cost effectiveness and complexity assessment in ship design...
TRANSCRIPT
Université de Liège
Faculté des Sciences Appliquées (FSA)
Département d’Architecture, Géologie, Environnement et Constructions (ARGENCO)
Secteur – Transport, Logistique, Urbanisme, Conception (TLU+C)
Service – Architecture Navale et Analyse des Systèmes de Transport (ANAST)
Cost Effectiveness and Complexity Assessment
in Ship Design
within a Concurrent Engineering and "Design for X" Framework
Thèse de Doctorat présentée en vue de l’obtention du grade de
Docteur en Sciences de l’Ingénieur
par
Jean-David Caprace
Liège, January 2010
Printed January 3, 2010 at 18:54
Il faut beaucoup de choses pour transformer le monde : lacolère et la tenacité. La science et l’indignation, l’initiativerapide, la longue réflexion, la froide patience, et la per-sévérence infinie, la compréhension du cas particulier et lacompréhension de l’ensemble. Seules les leçons de la réalitépeuvent nous apprendre à transformer la réalité.
Bertold Brecht
vii
Figure 1: Metaphor on cost effectiveness and complexity of ships [Cru05]
viii
Acknowledgements
I would like to express my deepest appreciation to my promotor Professor Philippe Rigo
for his guidance, kind support and encouragement during this study. His continued help,advice, and trust throughout my tenure at University of Liege from my Masters in navalarchitecture to graduation will stay with me forever. I am also thankful to the members ofthe jury for their suggestions and constructive criticism.
I would like to give many thanks to Professor Jean Marchal, Professor André Hage
and all my colleagues at ANAST, University of Liège, as well as my friends and family fortheir encouragement and support.
Special appreciation is extended to STX Europe for their availability in providing the re-quired data. All STX Europe personnel were particularly gracious and supportive, this hasbeen very much apprecied.
Certain individuals deserve special recognition. Mr Renaud Warnotte was especiallypatient and tremendously responsive to my requests. Mr Frédéric Bair and Mr JérômeMatagne demonstrated saintly patience and provided keen guidance for this PhD.
I also express my grateful thanks to all the official readers, Michel Caprace, EriksUskalis, Sandrine Le-Viol, Nicolas Losseau.
Finally, I would like to thank the National Fund of Scientific Research (FNRS) for theirfinancial support.
This PhD thesis was entirely typeset in LATEX using the TEXMaker software then con-verting to PostScript format with dvips and finally converting to Pdf format with Ghostscript[Rol95, GMS94].
ix
x
Abstract
Decisions taken during the initial design stage determine 60% to 95% of the total cost of aproduct. So there is a significant need to concurrently consider performance, cost, productionand design complexity issues at the early stages of the design process. The main obstacle tothis approach is the lack of convenient and reliable cost and performance models that can beintegrated into a complex design process as is used in the shipbuilding industry. Traditionalmodels and analysis methods frequently do not provide the required sensitivity to considerall the important variables impacting performance, cost, production and ship’s life cycle.Our challenge is that achieving this sensitivity at the early design stage almost requires dataavailable during the detail design analysis.
The traditional design methods do not adequately include, early enough, production andlife cycle engineering to have a positive impact on the design. Taking an integrated approachthroughout the life cycle of the ship and using concurrent engineering analysis tools canimprove these traditional design process weaknesses.
Innovation is required in structural design and cost assessment. The use of design for X,and particular design for production and cost schemes, during the design is the solution: toreduce failure during a ship’s life caused by design misconception, to reduce the overall designtime and to shorten the build cycle of ships, to enlarge the number of design alternativesduring the design process.
The author has developed some assessment methods for cost effectiveness and complexitymeasurements intended to be used by ship designers for the real time control of cost process.The outcome is that corrective actions can be taken by management in a rather short timeto actually improve or overcome predicted unfavourable performance.
Fundamentally, these methods will provide design engineers with objective and quantifiablecost and complexity measures making it possible to take rational design decisions throughoutthe design stages. The measures proposed in this PhD are based on several techniques likedecision analysis, data mining, neural network, fuzzy logic. They are objective facts, whichare not dependent on the engineer’s interpretation of information, but rather on a modelgenerated to represent the ship design. The objectivity aspect is essential when using thecomplexity and cost measures in a design automation system.
Finally, with these tools, the designers should obtain well-defined and unambiguous metricsfor measurement of the different types of cost effectiveness and complexities in engineeredartefacts. Such metrics help the designers and design automation tools to be objective andperform quantitative comparisons of alternative design solutions, cost estimation, as well asdesign optimization. In this PhD, these metrics have been applied and validated with successin real industrial conditions on the design of passenger ships.
xi
Résumé
Les décisions prises au cours de la phase initiale de conception déterminent 60% à 95%du coût total d’un produit. Il y a donc un besoin important d’envisager simultanément desaspects de performance, de coût, de production, de complexité au stade précoce du processus deconception. Le principal obstacle à cette approche est le manque de commodité et de fiabilitédes modèles de coûts et des modèles d’analyse de performance qui peuvent être intégrés dansun processus de conception aussi complexe que celui utilisé dans l’industrie de la constructionnavale. Les modèles traditionnels et les méthodes d’analyse les plus fréquemment utilisés nefournissent pas la sensibilité nécessaire pour examiner toutes les variables de conceptionimportantes qui ont une influence sur la performance, le coût, la production et le cycle de viedes navires. Notre défi est qu’atteindre cette sensibilité au niveau de la conception initialedemande presque toujours des données uniquement disponibles lors de la conception détaillée.
Les méthodes de conception traditionnelles ne prennent pas suffisamment en compte, suf-fisamment tôt, la production et l’ingénierie du cycle de vie afin d’avoir un impact positif surla conception. Adopter une approche intégrée sur l’ensemble du cycle de vie des navires etutiliser l’ingénierie simultanée peuvent améliorer les faiblesses des processus de conceptiontraditionnels.
L’innovation est nécessaire dans la conception structurelle et l’évaluation des coûts. L’uti-lisation du concept de "design for X", et en particulier le design en pensant à la production età la réduction des coûts lors de la conception est la solution : pour réduire les dysfonctionne-ments qui peuvent apparaître pendant la vie du navire causés par des erreurs de conception,de réduire le temps total de conception et de raccourcir le cycle de la construction de navires,d’élargir le nombre d’alternatives de conception évaluées au cours du processus de conception.
L’auteur a développé quelques méthodes d’évaluation des coûts et des techniques de mesurede la complexité destinées à être utilisés par les concepteurs de navire pour le contrôle entemps réel de la conception. Le résultat est que les actions correctives peuvent être prises parla direction dans un temps assez court pour améliorer réellement ou surmonter les prévisionsde performance défavorables.
Fondamentalement, ces méthodes offrent aux ingénieurs de conception des mesures quan-tifiables des coûts et de la complexité qui rend possible la prise de décisions rationnelles toutau long des étapes de conception. Les mesures proposées dans cette thèse sont basées surplusieurs techniques telles que l’analyse à la décision, l’analyse de données, les réseaux deneurones ou encore la logique floue. Ce sont des faits objectifs, qui ne dépendent pas del’interprétation de l’information par l’ingénieur, mais plutôt d’un modèle généré pour re-présenter le design du navire. L’aspect de l’objectivité est essentiel pour l’utilisation de lacomplexité et la mesure des coûts dans un système d’automatisation de la conception.
Finalement, avec ces outils, les concepteurs obtiennent des mesures bien définies et nonambigües des paramètres de mesure de coûts, d’efficacité et de complexité des artefacts d’in-génierie. De telles mesures aident les concepteurs et les outils d’automatisation de la concep-tion, à être objectifs et à comparer de manière quantitative les différentes alternatives lorsde la conception, de l’estimation des coûts, ainsi que de l’optimisation. Dans cette thèse, cesparamètres ont été appliqués et validés avec succès et dans des conditions industrielles réellessur la conception de navires à passagers.
xii
Resumen
Las decisiones tomadas en la etapa inicial de un diseño determinan el 60 % al 95 % deltotal del costo de un producto. Por esta razón, es necesario considerar al mismo tiemporendimiento, costo, producción y la complejidad en el proceso de la fase inicial del diseño.El principal obstáculo de este enfoque es la falta de práctica y fiabilidad de costos y demodelos de análisis de rendimiento, que puedan ser integrados en un proceso complejo deconcepción utilizado en la industria de la construcción naval. Los modelos tradicionales y losmétodos de análisis frecuentemente no tienen la sensibilidad necesaria para examinar todaslas variables importantes que influyen en el rendimiento, costo, producción y la vida útil delbuque. Nuestro reto es el que logrando de que ésta sensibilidad en la fase inicial de diseñocasi siempre requiera de datos únicamente disponibles durante un análisis de diseño requiredetallado.
Los métodos tradicionales de diseño no incluyen temprana y adecuadamente, la produccióny la ingeniería del ciclo de vida para tener un impacto positivo en el diseño. Adoptando unenfoque integrado a lo largo del ciclo de vida de la nave usando la ingeniería concurrentey herramientas de análisis se pueden mejorar estas debilidades del proceso tradicional dediseño.
Se requiere innovación en el diseño estructural y la evaluación de los costos. La utilizacióndel concepto “diseño para X”, y en particular el diseño pensando en la producción y en lareducción de costos durante la concepción es la solución: para reducir errores durante la vidaútil de los buques causados en la concepción del mismo, para reducir el tiempo total de diseñoy acortando el ciclo constructivo del buque, para ampliar el número de alternativas duranteel proceso de diseño.
El autor ha desarrollado algunos métodos de evaluación de costos y de técnicas de medidade la complejidad destinadas a ser utilizadas por los diseñadores de barcos para el controlen tiempo real del proceso de costos. El resultado es que las acciones correctivas puedenser adoptadas por la dirección en un tiempo suficientemente corto para realmente mejorar osuperar el rendimiento desfavorable proyectado.
Fundamentalmente, estos métodos ofrecerán a los ingenieros diseñadores con medidas ob-jetivas y cuantificables de costos y complejidad haciendo posible tomar decisiones racionalesa lo largo de todas las etapas de diseño. Las medidas propuestas en esta tesis doctoral sebasan en varias técnicas tales como análisis de decisión, análisis de datos, redes neuronalesy lógica difusa. Son objetivos hechos, que no dependen de la interpretación que el ingenierorealice sobre la información, sino más bien en un modelo generado para representar el dis-eño de los buques. El aspecto de la objetividad es esencial cuando se usa las medidas decomplejidad y costos en un sistema de automatización del diseño.
Finalmente con estas herramientas, los diseñadores podrían obtener medidas bien definidasy no parámetros de medición ambiguos de costos, de eficacidad y de complejidad en losartefactos de ingeniería. Estas medidas ayudan a los diseñadores y a las herramientas au-tomatizadas de diseño, a ser objetivos y a comparar de manera cuantitativa las diferentesalternativas de solución del diseño, estimación de costos, así como también la optimizacióndel diseño. En este doctorado, estos parámetros han sido aplicados y validados con éxito sobreel diseño de buques de pasajeros en condiciones industriales reales.
xiii
Übersicht
In der Anfangsphase einer Produktentwicklung getroffene Entscheidungen definieren 60%bis 90% der Gesamtkosten eines Produktes. Daher ist es absolut notwendig Performance, Ko-sten, Produktion, Designkomplexität in der Entstehungsphase eines Produktes zu betrachten.Das Haupthindernis hierbei ist das Fehlen von praktikablen und zuverlässigen Kosten- undPerformancemodellen, welche sich in den komplexen Designprozess wie in der Schiffbauin-dustrie integrieren lassen. Konventionelle Modelle und Analysemethoden berücksichtigen oftnicht all die wichtigen Performance-, kosten-, produktion-, und Lebenszyklus-relevanten Va-riablen. Unsere Herausforderung, dass das Erreichen dieser Sensibilität in der Anfangsphaseder Produktentwicklung, benötigt fast immer Daten, die erst in der Phase der Detailkon-struktion zur Verfügung stehen.
Die konventionelle Entwurfsmethoden berücksichtigen nicht angemessen und früh genugdie Produktion und das Lebenszyklen-Engineering, welche eine positive Auswirkung auf dasDesign haben. Ein integrierter Ansatz entlang des Lebenszyklus eines Schiffes und das Prakti-zieren von „Concurrent Engineering“ können die Schwäche des konventionellen Entwurfspro-zesses beseitigen.
Innovation ist notwendig im Strukturentwurf und in der Kostenanalyse. Die Anwendungvon „Design for X“ und besonders „Design for Production and Cost Scheme“ während derEntwurfsphase ist die Lösung: Um die Fehlerquote verursacht durch Missverständnisse unddie Entwurfs- und Produktionszeit zu reduzieren, sowie um eine höhere Anzahl von Entwurfs-varianten zu ermöglichen.
Der Autor hat einige Analysemethoden für Kosteneffektivität und Komplexitätsmaßnah-men für die Anwendung durch Schiffsentwerfer für die Echtzeit-Steuerung von Kostenprozessentwickelt. Das Ergebnis ist, dass das Management korrigierende Maßnahmen in kurzer Zeittreffen kann, um eine ungünstige Performance zu verbessern bzw. zu vermeiden.
Im Grunde beschaffen diese Methoden den Entwurfsingenieuren mit objektiven und quan-tifizierten Kosten- und Komplexitätsmaßnahmen. Dies vereinfacht das Treffen von vernünfti-gen Entwurfsentscheidungen über die gesamte Entstehungsphase eines Schiffes. Die in dieserArbeit vorgestellten Maßnahmen basieren auf Techniken wie die Entscheidungsanalyse, dasData-Mining, neurale Netze und die Fuzzy-Logik. Es sind objektive Fakten, welche nichtvon der Interpretation des Ingenieurs abhängen, sondern von einem generierten Model, umden Schiffsentwurf darzustellen. Der objektive Aspekt ist essenziell bei der Anwendung vonKomplexitäts- und Kostenmaßnahmen in einem automatisierten Entwurfssystem.
Mit diesen Werkzeugen sollte der Entwerfer gut-definierte und eindeutige Werte für Maß-nahmen von unterschiedlicher Natur von Kosteneffektivität und -komplexität erhalten. Die-se Werte helfen dem Entwerfer und automatisierten Entwurfssystemen, objektive zu sein.Ferner, ermöglichen diese Werte quantitative Vergleiche von Entwurfsvarianten, Kostenab-schätzung und Entwurfsoptimierung. Diese Werte wurden im Rahmen eines Studienfallesmit realen Voraussetzungen (Entwurf von Passagierschiffen) erfolgreich angewendet und va-lidiert.
xiv
Key words
Concurrent engineering – Design for X – Cost estimation – Cost assessment – Designcomplexity – Shipbuilding – Shipyard – Design optimization – Life cycle engineering
Mots-clés
Ingénierie simultanée – Design for X – Estimation des coûts – Évaluation des coûts -Complexité de conception - Construction navale - Chantier naval - Optimisation –
Ingénierie du cycle de vie
Palabras clave
Ingeniería concurrente – Desiño para X – Estimación de los costos – Évaluación de loscostos – Complejidad del diseño – Construcción naval – Astillero – Optimización –
Ingeniería del ciclo de vida
Stichwörter
Concurrent engineering — Design for X — Kostenabschätzung -– Kostenanalyse —Designkomplexität — Schiffbau -– Werft — Design Optimierung – Lebenszykluskosten
xv
Author
Jean-David CapraceResearch fellow of the Belgian National Fund of Scientific Research (FNRS)
University of Liège (ULG)
Faculté des Sciences Appliquées (FSA)
Département d’Architecture, Géologie, Environnement et Constructions (ARGENCO)
Secteur – Transport, Logistique, Urbanisme, Conception (TLU+C)
Service – Architecture Navale et Analyse des Systèmes de Transport (ANAST)
Chemin des chevreuils, 1
4000 Liège
Belgium
Phone +32 43669621
Fax +32 43669133
Mobile +32 494599607
Email [email protected]
Email [email protected]
xvi
Jury members
Jean Marchal (chairman)Professor
University of Liège, Liège, Belgium
Philippe Rigo (promotor)Professor
University of Liège, Liège, Belgium
Yves CramaProfessor
University of Liège, Liège, Belgium
Pierre DuysinxProfessor
University of Liège, Liège, Belgium
Frank RolandGeneral manager
Center of Maritime Technology, Hamburg, Germany
Tetsuo OkadaGeneral Manager, Structural Engineering
IHI Marine United Inc., Tokyo, Japan
Robert BronsartProfessor
Rostock university, Rostock, Germany
Floriano Carlos Martins PiresProfessor
Federal university of Rio de Janeiro, Rio de Janeiro, Brasil
Tran Ngoc DanProfessor
Hochiminh City University of Technology (HCMUT), Hochiminh, Vietnam
xvii
xviii
Contents
Acknowledgements ix
Abstract xi
Abstract (French) xii
Abstract (Spanish) xiii
Abstract (German) xiv
Author xvi
Jury xvii
Contents xix
1 Introduction 11.1 Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.1.1.1 Shipbuilding – An industry of labour . . . . . . . . . . . . . 2
1.1.1.2 Importance of cost control . . . . . . . . . . . . . . . . . . . 21.1.1.3 Design for X . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Focus of the research . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Research problem and hypothesis . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Research problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2.2 Boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.2.1 Where . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.2.2.2 What . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.2.3 How . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.2.2.4 Why . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Justification of the research . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.4 Outline of the study and description of work . . . . . . . . . . . . . . . . . . 8
1.5 Original contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.6 Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6.1 InterSHIP - Competitiveness . . . . . . . . . . . . . . . . . . . . . . . 121.6.2 LBR5 - Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.6.3 MARSTRUCT - Networking . . . . . . . . . . . . . . . . . . . . . . . 131.6.4 VISION - Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . 141.6.5 IMPROVE - Integration . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
xix
2 State of the art 172.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.2 Shipbuilding industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.1 Shipbuilding history . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.2.1.1 Welding and riveting . . . . . . . . . . . . . . . . . . . . . . 172.2.1.2 Blocks construction . . . . . . . . . . . . . . . . . . . . . . . 192.2.1.3 Industry of labour – Key players over the world . . . . . . . 212.2.1.4 Shipyard strategies . . . . . . . . . . . . . . . . . . . . . . . 232.2.1.5 Ship owner strategies . . . . . . . . . . . . . . . . . . . . . . 242.2.1.6 Tools to support design and production (Computer Aided
Engineering – CAE) . . . . . . . . . . . . . . . . . . . . . . 272.2.2 A non-conventional industry . . . . . . . . . . . . . . . . . . . . . . . 282.2.3 European shipbuilding industry . . . . . . . . . . . . . . . . . . . . . 30
2.2.3.1 Recent evolution . . . . . . . . . . . . . . . . . . . . . . . . 302.2.3.2 Order book of ships in Europe . . . . . . . . . . . . . . . . . 332.2.3.3 Research fields in Europe . . . . . . . . . . . . . . . . . . . 33
2.2.4 Design and conception . . . . . . . . . . . . . . . . . . . . . . . . . . 352.2.4.1 Concept design . . . . . . . . . . . . . . . . . . . . . . . . . 362.2.4.2 Preliminary design . . . . . . . . . . . . . . . . . . . . . . . 362.2.4.3 Contract design . . . . . . . . . . . . . . . . . . . . . . . . . 372.2.4.4 Detailed design . . . . . . . . . . . . . . . . . . . . . . . . . 372.2.4.5 Production design . . . . . . . . . . . . . . . . . . . . . . . 37
2.2.5 Production and fabrication . . . . . . . . . . . . . . . . . . . . . . . . 372.2.5.1 Group Technology (GTech) . . . . . . . . . . . . . . . . . . 382.2.5.2 Product Work Breakdown Structure (PWBS) . . . . . . . . 392.2.5.3 Ship Work Breakdown Structure . . . . . . . . . . . . . . . 402.2.5.4 Hierarchical work stages . . . . . . . . . . . . . . . . . . . . 402.2.5.5 Hierarchical work types . . . . . . . . . . . . . . . . . . . . 42
2.2.6 Productivity and competitiveness factors . . . . . . . . . . . . . . . . 422.2.6.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422.2.6.2 Measurement of the productivity . . . . . . . . . . . . . . . 43
2.2.7 Inefficiencies of traditional design and production processes . . . . . . 452.2.7.1 Poor production consideration during design . . . . . . . . . 452.2.7.2 Low productivity and high labour cost . . . . . . . . . . . . 462.2.7.3 Increasing number of production constraints . . . . . . . . . 46
2.3 Cost assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482.3.1 Cost "estimate" and cost "assessment" . . . . . . . . . . . . . . . . . . 482.3.2 Production cost assessment . . . . . . . . . . . . . . . . . . . . . . . 492.3.3 Types of cost assessment . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.3.1 Initial design stage . . . . . . . . . . . . . . . . . . . . . . . 502.3.3.2 Detailed design stage . . . . . . . . . . . . . . . . . . . . . . 512.3.3.3 Production design stage . . . . . . . . . . . . . . . . . . . . 51
2.3.4 State of art in shipbuilding industry cost assessment . . . . . . . . . . 512.3.4.1 Top-Down approaches . . . . . . . . . . . . . . . . . . . . . 512.3.4.2 Bottom-Up approaches . . . . . . . . . . . . . . . . . . . . . 552.3.4.3 Life cycle approaches . . . . . . . . . . . . . . . . . . . . . . 58
2.3.5 Challenges of cost assessment . . . . . . . . . . . . . . . . . . . . . . 592.3.5.1 Disconnection between decision and cost . . . . . . . . . . . 592.3.5.2 Inaccuracy of the cost assessment . . . . . . . . . . . . . . . 59
xx
2.3.5.3 Cost evaluated only once . . . . . . . . . . . . . . . . . . . . 592.3.5.4 Multiple versions of the cost . . . . . . . . . . . . . . . . . . 592.3.5.5 Measure of real costs . . . . . . . . . . . . . . . . . . . . . . 592.3.5.6 Uncoupling between design and cost engineering . . . . . . . 602.3.5.7 Cost assessment in early design stage - a real challenge . . . 602.3.5.8 Specificities of the shipbuilding industry . . . . . . . . . . . 602.3.5.9 Intricate control of schedules and costs . . . . . . . . . . . . 612.3.5.10 Cost variation factors . . . . . . . . . . . . . . . . . . . . . 612.3.5.11 Increasing Costs . . . . . . . . . . . . . . . . . . . . . . . . 632.3.5.12 Data and DB management problems . . . . . . . . . . . . . 63
2.4 Concurrent engineering (CE) . . . . . . . . . . . . . . . . . . . . . . . . . . . 662.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662.4.2 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662.4.3 Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.4.3.1 First principle . . . . . . . . . . . . . . . . . . . . . . . . . . 662.4.3.2 Second principle . . . . . . . . . . . . . . . . . . . . . . . . 682.4.3.3 Third principle . . . . . . . . . . . . . . . . . . . . . . . . . 682.4.3.4 Fourth principle . . . . . . . . . . . . . . . . . . . . . . . . 682.4.3.5 Fifth principle . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.4.4 Drawbacks and limitations . . . . . . . . . . . . . . . . . . . . . . . . 702.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3 Methodology 713.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713.2 Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713.3 Design for X (DFX) – Concurrent Engineering (CE) tools . . . . . . . . . . . 75
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753.3.2 Design for production (DFP) – Design for manufacturing (DFM) . . 763.3.3 Design for assembly (DFA) . . . . . . . . . . . . . . . . . . . . . . . . 793.3.4 Design to cost (DTC) . . . . . . . . . . . . . . . . . . . . . . . . . . . 803.3.5 Design for Simplicity . . . . . . . . . . . . . . . . . . . . . . . . . . . 803.3.6 Design for maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . 813.3.7 Design for environment (DFE) . . . . . . . . . . . . . . . . . . . . . . 813.3.8 Design for safety (DFS) – Risk Based Design (RBD) . . . . . . . . . 833.3.9 Design for life cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 833.3.10 Design for Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . 853.3.11 Design for process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.3.11.1 Design For Six Sigma (DFSS) . . . . . . . . . . . . . . . . . 863.3.11.2 Design for Lean Manufacturing . . . . . . . . . . . . . . . . 88
3.3.12 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 883.4 Selection of a cost estimation method . . . . . . . . . . . . . . . . . . . . . . 90
3.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 903.4.2 The different cost assessment methods . . . . . . . . . . . . . . . . . 90
3.4.2.1 Intuitive method (IM) – Expert opinion . . . . . . . . . . . 913.4.2.2 Case based reasoning (CBR) – Analogy analysis . . . . . . . 923.4.2.3 Parametric method (PM) – Statistical analysis . . . . . . . 943.4.2.4 Feature-Based Costing (FBC) . . . . . . . . . . . . . . . . . 963.4.2.5 Fuzzy logic method (FLM) . . . . . . . . . . . . . . . . . . 983.4.2.6 Neural networks method (NNM) . . . . . . . . . . . . . . . 1013.4.2.7 Simulation method (SM) . . . . . . . . . . . . . . . . . . . . 103
xxi
3.4.3 Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1053.4.4 Selection of cost assessment method . . . . . . . . . . . . . . . . . . . 1073.4.5 Multi-criteria decision analysis . . . . . . . . . . . . . . . . . . . . . . 108
3.4.5.1 Definition of alternatives . . . . . . . . . . . . . . . . . . . . 1083.4.5.2 Definition of criterion . . . . . . . . . . . . . . . . . . . . . 1083.4.5.3 Definition of weights and scenarios . . . . . . . . . . . . . . 1093.4.5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1113.4.5.5 GAIA visualisation . . . . . . . . . . . . . . . . . . . . . . . 1123.4.5.6 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . 112
3.4.6 Conclusions about the selection of a cost estimation method . . . . . 1163.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4 Analysis, developments and results 1174.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1174.2 Presentation of the developments . . . . . . . . . . . . . . . . . . . . . . . . 1184.3 A Feature Based Costing module for shipbuilding industry . . . . . . . . . . 121
4.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1214.3.2 Developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.3.2.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 1234.3.2.2 Cost structure . . . . . . . . . . . . . . . . . . . . . . . . . 1244.3.2.3 Cost Evaluation Relationships (CERs) . . . . . . . . . . . . 1254.3.2.4 Data flow – Outside components . . . . . . . . . . . . . . . 1314.3.2.5 Data flow – Inside components . . . . . . . . . . . . . . . . 133
4.3.3 Analysis and results . . . . . . . . . . . . . . . . . . . . . . . . . . . 1384.3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.4 Complexity of ships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1414.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1414.4.2 Objectives of the complexity . . . . . . . . . . . . . . . . . . . . . . . 1424.4.3 Definition of complexity . . . . . . . . . . . . . . . . . . . . . . . . . 1434.4.4 Macro complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
4.4.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 1444.4.4.2 Complexity of different types of ship (Ctyp) . . . . . . . . . 1444.4.4.3 Complexity of same type of ships (Carr) . . . . . . . . . . . 1464.4.4.4 Conclusion about macro complexity . . . . . . . . . . . . . 157
4.4.5 Micro complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1584.4.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 1584.4.5.2 Definition of micro complexity . . . . . . . . . . . . . . . . . 1584.4.5.3 Shape complexity – Csh . . . . . . . . . . . . . . . . . . . . 1604.4.5.4 Assembly complexity – Cas . . . . . . . . . . . . . . . . . . 1624.4.5.5 Material complexity – Cmt . . . . . . . . . . . . . . . . . . . 1644.4.5.6 Application . . . . . . . . . . . . . . . . . . . . . . . . . . . 1654.4.5.7 Conclusions about micro complexity . . . . . . . . . . . . . 166
4.4.6 Conclusion about the complexity . . . . . . . . . . . . . . . . . . . . 1664.5 Application of two cost evaluation methods of the straightening operation . . 170
4.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1704.5.2 A neural network cost assessment method (NNM) applied to straight-
ening operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1704.5.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 1704.5.2.2 Data Mining (DM) . . . . . . . . . . . . . . . . . . . . . . . 1714.5.2.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 172
xxii
4.5.2.4 Database architecture . . . . . . . . . . . . . . . . . . . . . 173
4.5.2.5 Data description . . . . . . . . . . . . . . . . . . . . . . . . 173
4.5.2.6 Data quality . . . . . . . . . . . . . . . . . . . . . . . . . . 173
4.5.2.7 Data exploration . . . . . . . . . . . . . . . . . . . . . . . . 173
4.5.2.8 Data modelling . . . . . . . . . . . . . . . . . . . . . . . . . 180
4.5.2.9 Application case . . . . . . . . . . . . . . . . . . . . . . . . 181
4.5.2.10 Conclusion about neural network method . . . . . . . . . . 184
4.5.3 A straightening fuzzy metric . . . . . . . . . . . . . . . . . . . . . . . 185
4.5.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 185
4.5.3.2 Fuzzy sets and membership function . . . . . . . . . . . . . 185
4.5.3.3 Fuzzy rule matrix . . . . . . . . . . . . . . . . . . . . . . . . 187
4.5.3.4 Overall performance of fuzzy estimator . . . . . . . . . . . . 187
4.5.3.5 Comparison with real data . . . . . . . . . . . . . . . . . . . 187
4.5.3.6 Optimisation of the fuzzy output . . . . . . . . . . . . . . . 189
4.5.3.7 Application case . . . . . . . . . . . . . . . . . . . . . . . . 191
4.5.3.8 Conclusion about fuzzy logic method . . . . . . . . . . . . . 191
4.5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
4.6 A statistical cost metric for a ship designer . . . . . . . . . . . . . . . . . . . 193
4.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
4.6.2 Database presentation . . . . . . . . . . . . . . . . . . . . . . . . . . 193
4.6.3 Principal Components Analysis (PCA) . . . . . . . . . . . . . . . . . 194
4.6.4 Database normalisation . . . . . . . . . . . . . . . . . . . . . . . . . . 195
4.6.5 Data modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
4.6.5.1 Dendrogram analysis . . . . . . . . . . . . . . . . . . . . . . 196
4.6.5.2 Multiple Linear Regression Analysis . . . . . . . . . . . . . 196
4.6.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
4.7 Cost assessment trough production simulation . . . . . . . . . . . . . . . . . 201
4.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
4.7.2 Test case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
4.7.2.1 The workshop . . . . . . . . . . . . . . . . . . . . . . . . . . 201
4.7.2.2 Two designs . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
4.7.2.3 Two blocks and sections splitting . . . . . . . . . . . . . . . 202
4.7.2.4 Two state of scantling . . . . . . . . . . . . . . . . . . . . . 202
4.7.2.5 Sister ships . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
4.7.3 Process flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
4.7.4 Budget assessment module . . . . . . . . . . . . . . . . . . . . . . . . 206
4.7.5 Production simulation module . . . . . . . . . . . . . . . . . . . . . . 207
4.7.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
4.7.6.1 Budget assessment . . . . . . . . . . . . . . . . . . . . . . . 208
4.7.6.2 Space allocation . . . . . . . . . . . . . . . . . . . . . . . . 209
4.7.6.3 Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
4.7.6.4 Lead time . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
4.7.6.5 Final interpretation . . . . . . . . . . . . . . . . . . . . . . 212
4.7.7 Conclusion about the production simulation . . . . . . . . . . . . . . 213
4.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
xxiii
5 Conclusion and recommendations 2175.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2175.2 Key findings and achievements . . . . . . . . . . . . . . . . . . . . . . . . . . 217
5.2.1 Key findings and achievements . . . . . . . . . . . . . . . . . . . . . . 2175.2.2 Main contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
5.3 SWOT analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
List of Figures 221
List of Tables 225
Bibliography 227
Glossary 253
Index 254
A Multicriteria analysis 257A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257A.2 Multicriteria Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257A.3 PROMETHEE method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258A.4 GAIA representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260A.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
B Survey about life cycle cost management 263B.1 Form of the survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263B.2 Results of the survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
C Feature based costing prototype 275C.1 Welding considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
C.1.1 Definition of the welding type . . . . . . . . . . . . . . . . . . . . . . 275C.1.2 Definition of the welding position . . . . . . . . . . . . . . . . . . . . 276
C.2 Database structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279C.2.1 CAD/CAM relational database . . . . . . . . . . . . . . . . . . . . . 279C.2.2 Cost relational database . . . . . . . . . . . . . . . . . . . . . . . . . 279
C.2.2.1 Data relating to the cost structure . . . . . . . . . . . . . . 279C.2.2.2 Data relating to the cost scales and design variables . . . . . 281C.2.2.3 Data relating to the cost results . . . . . . . . . . . . . . . . 285
C.3 Graphical User Interface (GUI) . . . . . . . . . . . . . . . . . . . . . . . . . 285
D Production simulation 291D.1 STS tool set for shipbuilding . . . . . . . . . . . . . . . . . . . . . . . . . . . 291D.2 Database structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
E Productivity measurement and improvement 295E.1 Productivity measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295E.2 Productivity improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
xxiv
Chapter 1
Introduction
The shipbuilding industry provides a good case study of how the traditional Europeanmaritime sector is facing up to the increasing pressures of global competition, most notablyfrom Asia. In the last decade, European shipbuilding has lost 36% of its jobs but gained 43%in productivity [EuG08]. This has resulted in a sector that is specialised in the productionof sophisticated vessels. Ships produced in Europe are outstanding in terms of complexity,safety and environmental impact, often well beyond regulatory requirements.
The ship design procedure has been improved to produce better and more effective prod-ucts. The demand has been driven by increasing customer specifications for more economicand viable designs and has resulted in a great deal of competitiveness between operatorsand designers [Bol02]. Furthermore, baseline design standards have been raised by societyin many areas, such as safety and pollution control, in attempts to reduce the number ofaccidents and the burden on the global environment. In order to deal with those demands,there has been a significant amount of development in tools, techniques and processes.
This kind of industry has a growing need for increased productivity to remain competi-tive regarding the worldwide competition. The models of cost effectiveness and complexityassessment presented in this PhD study could significantly improve the shipyard’ss compet-itiveness.
This chapter presents the foundations for the report. It introduces the research problemand research questions and hypotheses (see section 1.2). Then the research is justified (seesection 1.3), the report is outlined (see section 1.4), the original contributions are given (seesection 1.5) and the context is described (see section 1.6). On these foundations, the reportcan proceed with a detailed description of the research.
1.1 Framework
This section first outlines the broad field of the study and then leads into the focus of theresearch problems involved.
1
1.1.1 Background
1.1.1.1 Shipbuilding – An industry of labour
Shipbuilding activity requires a lot of labour. For example, for a passenger ship, the shiphull (steel part) represents approximately 20% of the cost of the ship and the cost of labourrepresents about 50% of the cost of the ship hull. For our western countries, this fact leadsto a relocation of ship manufacturers towards regions which have lower labour costs. Indeed,we can observe in history that the leadership in shipbuilding moved from the UK to Japanand later to Korea, whereas today China and Vietnam are getting an increasing part of themarket. Currently, Asian shipyards produce about 80% of regular ships, such as tankers andcontainer ships.
To avoid this relocation and to remain profitable, European shipyards (Italian, German,French and Scandinavian) decided to devote themselves only to ships with high added valueand with high technology such as cruise ships, fast ferries, advanced cargo carrying ships,dredgers, off-shore support vessels, fish production vessels, mega-yachts, research vessels,etc. To keep and increase their shipbuilding world market share the European shipyardsare always obliged to increase their competitiveness significantly. Moreover, the nature ofthe European shipbuilding market prevents large production series; each ship is unique (seesection 2.2.2), and the installation of full automatic processes remains complex.
Nowadays European shipbuilding is facing a new difficult conjuncture, so working methodsand management must evolve to allow the survival of the sector. Even if significant effortshave already been made by the shipbuilding industry to reduce the costs of each individualstage of ship construction, the European objectives (see section 2.2.3.3) in this domain muststill be achieved: reduction of the design and manufacturing costs (25 to 30%) as well asproduction time (20 to 30%). Another important research field concerns the developmentof the best product by using multi-purpose optimization integrated design, quality, safety,environment and efficiency.
Since the main part of the construction cost relates to the production and since theproducibility of a ship is basically defined at the design stage, the main promising track ofcost savings is to assess the cost effectiveness and complexity as soon as the main options ofconstruction are determined (Design for Production, Design to Cost, Design for Simplicity– see section 3.3).
1.1.1.2 Importance of cost control
The ability to assess ship construction costs is necessary for the commercial success of ashipyard:
• overestimating the cost will place the shipyard out of the competitive range, and
• underestimating the cost will result in financial loss and possible bankruptcy.
Cost has become a major business driver in many industries. It can be observed that thereis a lack of understanding about the process of estimating, managing and controlling costsacross the life cycle of a product [RK03]. In developed countries, where labour costs arehigher and remaining competitive with emerging countries is essential to stay in business,the necessity to modernize ship production is evident.
In order to achieve this goal, there are a number of approaches which can help to reduceglobal production costs and lead time without reducing the performance and quality of the
2
product. These approaches concern the whole ship, not only the hull structure. The mostpromising gains can be made by integrating Design for X concepts from the concept designstage (as early as possible in the design process). This is because the majority of the cost isestablished in the early design stages [LAA06].
1.1.1.3 Design for X
In an increasingly competitive world, product design may be the ultimate way to distin-guish a company’s capabilities. Because of the growing importance of product design, designfor production and design for maintenance concepts will be critical. It will be the key toachieving and sustaining competitive advantage through the development of high quality,highly functional ships effectively manufactured and highly maintainable ships through thesynergy of integrated product and process design.
The traditional role of the ship designer is the preparation of an overall design of a vesselwhich will have to satisfy the owners operational and functional requirements while com-plying with the statutory rules and regulations. The concept of design for efficiency andquality in production, however, requires that in satisfying these requirements, the ship de-signer must also give attention to ease of production [BHH+06]. There are thus several majoraspects to design, namely, design for performance, including design for safety and design forproduction. There are in addition some other aspects, considered later in the developmentcycle, including design for overhaul, repair and maintenance. To ease the workload on designmany decision support methods will be developed to find a compromise between the variousdemands required to satisfy each area. Clearly, there will be areas of interaction. The role ofthe ship designer can be seen in this context as an arbiter, having the ultimate responsibilityof deciding whether performance or production considerations shall take precedence in anyparticular case and deciding the nature of the compromise to be reached.
However, increasing global competition is challenging the shipbuilding industry to bringcompetitively priced, well-designed, and well-manufactured ships to market in a timely fash-ion. To achieve this goal, increasing research attention is being directed toward the integra-tion of ship design and production tools for ship designers [OAT04]. These attempts willlead to the evolution of Design For X.
1.1.2 Focus of the research
This study aims to provide a fresh look at the cost and complexity assessment problem andits associated tasks, to identify some approaches that will make the process easier and moreproductive for the shipbuilding industry. Of course, there have been many previous attemptsto produce new or better cost and complexity assessments with some success (see section2.3.4). However, it is impossible to ignore the fact that many of these tools use successfulprocesses developed over many years (see section 2.3.4). While this study recognises thistrend, it is clear that the present tools are not as effective for producing the product asthose used in other areas of engineering design such as auto-mobile or aerospace industry.Consequently, this study is focused on the search for a solution that involves the innovativeintegration of existing cost and complexity evaluation techniques by maximising the use ofknowledge and information inherent within the product itself, the ship.
This aim will be achieved by considering the following objectives:
3
1. To make a practical review of integrated shipbuilding processes in order to highlightwhy this industry is specific and why it is so difficult to improve the competitivenessand efficiency of ships through the entirety of their lives.
2. To review in detail and from a practical viewpoint the wide range of cost assessmenttechniques that have been used over the years. This will also allow the processesinvolved to be identified, the advantages and fundamental limitations of each and anunderstanding of why there are a lot of different techniques being used in modern shipdesign tools.
3. To make a practical review of the cost evaluation methodologies. To list the existingtools to identify where the designer’s requirements are not being met and to highlightany tools with innovative approaches which assist the designer to achieve particulartasks.
4. To establish potential solutions based on the thorough review of the processes andmethods presently being used in the ship design process. A clear and objective ap-proach will be used to resolve present problems by identifying any factors that havebeen overlooked by present tool developers.
5. To develop some pilot systems to demonstrate the effectiveness of the approach. Thishas been done whilst confronting the resolution of many individual problems facedwithin industrial tools.
1.2 Research problem and hypothesis
This section outlines the core of the research, starting from the research problem printedin bold in the section 1.2.1 of this PhD thesis.
1.2.1 Research problem
Ship designers quite often face the problem of selecting a design alternative (DA) from agiven set of a finite number of alternatives, particularly in the early stages of the ship designprocess [OAT04]. The chosen DA is the best one or a compromise option that meets certainpredefined objectives. Evaluation of DAs often needs to take into account many attributesso that the economic, technical, safety, and other aspects can be comprehensively assessed.
The hierarchy of parameters which affect the actual construction processes involved duringfabrication and assembly of ships is very extensive and complex. The list presented in Tab.1.1 is believed to be sufficiently comprehensive to highlight that a lot of design factors areinvolved in the producibility, complexity and cost assessment of a ship structure.
Speaking with a ship designer:
– When you are working in your CAD program, every change you make to a ship designhas an impact – positive or negative – on how much producing/maintaining the ship willcost. Wouldn’t it be nice if you could understand the impact of your design and trade-offdecisions every time you made a change?
4
To approve or disapprove the implementation of any design change the designer requiresanswers to the following questions:
• How much will it cost (or save) to implement this change?
• How will the schedule be impacted?
• What risk is involved?
• How will the ship’s performances be affected?
• How will the complexity of the whole structure be affected?
• How will the productivity of the ship be impacted?
• How will the maintenance of the ship be affected?
1.2.2 Boundaries
1.2.2.1 Where
The PhD research has been carried out in the Naval Architecture and Transport Sys-tem Analysis sector (ANAST) of the University of Liege (ULg) and promoted by ProfessorPhilippe Rigo. The financial support has been provided by the Belgian National Fund ofScientific Research (NFSR).
1.2.2.2 What
The developments of this PhD study have been applied to shipbuilding industry, but isis not confined to this specific sector. The tools developed highlight the specific problemsencountered in this technical area.
Applications and developments presented in chapter 4 are focused on ship structure (steelpart) and on labour costs.
1.2.2.3 How
The implemented methods during this research reflect the industrial purpose of the PhDthesis. In this industrial context, the problems faced are often specific and the analysisof the problem is immediately followed by the development of a specific solution. Thissolution is often non optimal because the time for reflection and development is very small
Welding considerations Pipefitting considerations Electrical considerations HVAC considerations
Welding length Pipe size Wireways Ducting
Welding type Pipe length Connections/hookups Size
Welding position Piping support needs Cable Length
Welding process Process Length Material
Number of passes Use of bends vs. fittings Size Configuration changes
Welding accessibility Connection type Equipment installation
Plate bevel angles Material type Insulation
Heat treatment
Table 1.1: Producibility and cost criterion
5
in order to continue the production as soon as possible. Although a shipyard is often facescomplex problems that require the completion of fundamental research, industrial dynamicsremain characterized by a specific and direct link between the occurrence of the problemand its resolution. In the results of our study, this aspect has been enhanced while showinginteresting generic methods and tools. The contact with the industry that we maintainedthroughout this research project has allowed us to develop solutions directly assessed withreal tested cases.
We would like to mention two problems here, which have always been important through-out the ages, to which nowadays the shipbuilding industry is finding even more urgently asolution. They are:
• reduction of production delay
• reduction of life cycle costs and complexities
How to find solutions in order to achieve these two objectives is the heart of the matter andhas been the main objective of the author’s work.
1.2.2.4 Why
In order to achieve the purpose of this study, many methods could be considered but theymust all have the same objective: "increasing the competitiveness of shipyards through thereduction of the life cycle cost".
Several paths are possible and could be classed in three major methods:
1. The optimization of the industrial process – quality management, 6σ (Motorola), leanmanufacturing (Toyota), CAD/CAM, scheduling, sequencing, etc.
2. The optimization of the product design – design for production, standardisation, mod-ularization, etc.
3. The optimization of the industrial layout – automation , mechanization, etc.
Research conducted in the last 15 years by Ulg-ANAST has addressed the first two pointsof the previous list. It is commonly recognized that a lot of research has succeeded to improvethe third point, while much remains to be done to improve the first two points. The PhDstudy [Bai09] conducted by ANAST focuses on the first point while the current one aimsspecifically at overcoming the second one.
1.3 Justification of the research
This section is devoted to the justification of the research. We highlight here that the re-search problems should be considered important in terms of several theoretical and practicalaspects.
Cost effectiveness and complexity assessmentin ship design
within a concurrent engineering and "design for X" framework
6
To introduce the developments and results of the PhD, we will highlight the principlesand the ideas that have guided our study, and then we will justify the specific nature of themethods used. The implementation of the "design for X" in the industry is one of the mostobvious successes of the research in terms of cost reduction, production time and higherquality. This concept is based on a design approach that takes into account all aspects ofthe product life cycle from design to retirement. This is seen as the result of an improveddesign, production and maintenance process under the responsibility of all the sectors ofbusiness.
The closer inter-dependencies between design, life cycle performance and fabrication tech-niques have been highlighted in a lot of papers [BHL+03, BHH+06, CEH+09]. These inter-actions are bidirectional:
• Construction cost and manufacturing conditions are to a large extent defined in earlydesign phases. It is therefore important that the designer is provided with methodsand tools which enable him to sufficiently consider design alternatives, cost aspects,new fabrication technologies and materials in his work.
• Manufacturing quality, imperfections and accuracy have a significant impact on struc-tural performance, repair and maintenance and life cycle cost.
Nowadays, market drivers induce permanent innovation through better designs and moreefficient fabrication techniques. Fuel costs force better hydrodynamics efficiency, harbourenvironmental concerns force lower slow-speed propeller wash, navigational constraints forcebetter manoeuvring and control systems, steel cost force better optimisation of the hullstructures, etc.
It is well known that the traditional approach to product development leads to long prod-uct development cycles and cost control difficulties in particular for shipbuilding. The desiredapproach to product development is a concurrent one with an emphasis on cost control. Oneof the targets of Concurrent Engineering (see section 2.4) is to reduce both the cost andmanufacturing time of a product through simultaneous consideration of product develop-ment activities. Research results have showed that over 70% of the production cost of aproduct is determined during the conceptual design stage [Hun93]. However, the designphase itself accounts for only 6% of the total development cost.
Recognition of the pre-eminent influence of design in the overall competitiveness of aproduct naturally leads to the inference that this stage also offers the greatest opportunityof cost control. Therefore it may be concluded that cost itself is a design element andshould be controlled as part of the design process [IH06, RG98]. Consequently, devoting agreater effort to design cost is a necessary step towards optimising ships. In other words, themajority of the life cycle cost is determined by a designer who might have little knowledgeabout the cost [OYL97]. Since a design modification requires only a little time but mightgreatly reduce life cycle costs and complexity, developing a tool to help the designer toevaluate the required life cycle cost is an important task.
So there is a significant need to concurrently consider performance, cost, production,design complexity issues at the early stage of the design process. The main obstacle to thisapproach is the lack of convenient and reliable cost and performance models that can be
7
integrated into a complex design process as used in the shipbuilding industry. Traditionalmodels and analysis methods frequently do not provide the required sensitivity to considerall the important variables impacting performance, cost, production, and ship life cycle. Ourchallenge is that to achieve this sensitivity at the early design stage requires the data whichare usually available during the detailed design analysis.
This PhD thesis gives grounds for a new philosophy to be used for ship structure designs.The new philosophy relies on the implementation of effective complexity and life cycle costevaluation during all stages of the design in order to enhance the Design for X concepts ina Concurrent Engineering framework. This ensures pertinent factors traditionally neglecteduntil late in the design process, which ultimately delays the delivery and increase the lifecycle cost. Early incorporation of these items allows assessment of their impact on the shipdesign and performance. An additional benefit of this approach is that the management isprovided with an early indication of the scale of the ship cost. This enables the managementto make more informed bid estimates at the conceptual design phase.
This PhD thesis provides methodology and tools to improve the existing solutions throughcost and complexity criteria during all ship design stages. So the designer will be oriented tofind the best design which reduces life cycle cost and complexity to a minimum, compatiblewith the requirements of the vessel, and to fulfilling its operational functions with acceptablesafety, reliability and efficiency.
The traditional cost assessment methods largely focus on two ways of reducing cost. Theseare through a reduction in material costs and labour costs. These reductions are vigorouslypursued through many ways including substituting cheaper material or automating labourpractices. Therefore, focusing only on these two areas ignores the equally important areas ofdesign changes. The impact of the cost evaluation on the design is precisely the scope of thisPhD thesis. The overall thrust of the thesis is to integrate ship synthesis with cost assessmentincluding all conception and design parameters to explore the most design alternatives inthe early stage of the design process. The innovation is to provide the designer a powerfultool which allows real-time monitoring of the future performance of the vessel through thelife cycle cost, so that it can evaluate different design alternatives and choose the best one.
1.4 Outline of the study and description of work
Each chapter of the PhD thesis is briefly described in this section.
The PhD thesis begins by making a brief review of the development of the integratedshipbuilding process. The importance of the techniques and approaches used in ship designbefore it became dominated by the scientific approach should not be underestimated andmany improvements were introduced by the innovative engineers working in the period.At the turn of the last century, the growing interest in science and the momentum of theindustrial revolution gave rise to the development of numerical approaches to improve shipperformance (competitivity, producibility, efficiency, etc.). An extensive review of many ofthe techniques that were developed, from the early beginnings to those of the present day iscarried out in section 2.2.
8
The cost assessment process has adapted to the capabilities of the design technologyto ensure that products remain competitive. As other technologies are being developedand being used in the design process, it is important to understand the changes that aretaking place to ensure that cost assessment tools will continue to remain effective in thefuture. A state of the art is made of the particular characteristic traits of available costassessment tools devoted to ships, identifying where these tools fail the design process andsome of the approaches that can enhance the development of cost and complexity evaluationdedicated to a ship structure (see section 2.3). Subsequently, a review is made of currentcost evaluation methodologies to understand how the development process functions andhow well the functionality provided by present tools compares (see section 3.4).
Before embarking on the development of potential solutions, a deep review of concurrentengineering and an exploration of the Design for X concept are made respectively in section2.4 and section 3.3).
The study uses the various findings identified by the review of the way of combining presenttechniques to produce a concept for an integrated cost evaluation dedicated to the overalllife cycle of ships. The paradigm of the study is developed in section 3.2.
The study analysis, developments and results are presented in chapter 4. A feature basedcosting, a complexity metric, two straightening cost assessment modules, a statistical costevaluation and a cost assessment trough production simulation are successively described,and the results are extensively discussed.
The thesis concludes with an evaluation of the concept. We compare the new approacheswith present techniques and present a discussion with regards to their performance, theirwider implications and their future development (see chapter 5).
The appendices provide additional details of many of the areas covered in the work. Theappendices also give details of some of the techniques and procedures used in the pilot systemthat were not directly relevant to the development of the approach.
1.5 Original contributions
The major original investigation of this PhD study is to provide some innovative solu-tions for cost and complexity assessment during ship design to enhance the "design for X"concept in a Concurrent Engineering framework. To the author’s knowledge, this kind of ob-jective measurement during the ship design has not yet been integrated in the most commonCAD/CAM commercial tools used in shipbuilding industry.
Moreover, the study places these developments in a holistic ship design optimisationstrategy where all design objectives are considered simultaneously in order to explore themost design alternatives in the early stage of the project (see Fig. 3.5). Having consideredall these techniques concurrently by the minimization of the life-cycle cost is an innovationin itself.
Other original contribution items presented in this PhD study are:
• An outstanding state of the art has been provided.
9
• The complex nature of decision making involved in the selection of the cost perfor-mance and complexity assessment methods at each stage of the ship design has beenhighlighted. Multi-criteria analysis has been used to identify the best cost assessmentmethods and particularly for uncertain decision environments.
• Two different methods of evaluating and measuring the complexity and efficiency inship design have been presented in this study; the first one for the contract designstage (macro complexity) and the second one for the detailed design stage (microcomplexity). The real-time evaluation of these metrics which requires less computingtime than the cost assessment is a new concept. This provides a quantifiable and aobjective assessment of the efficiency and the complexity of ships during various designstages.
• A fuzzy straightening cost metric has been introduced. This measure has been de-veloped by using the expert opinion based on accumulated experience because thisphenomenon is very difficult to tackle. Proof that is possible to minimize the errorsbetween the fuzzy output surface generated by the expert knowledge and the real mea-sured data has been provided in this study. In parallel, the limitations of the neuralnetwork analysis and production simulation to handle innovative design or to managedesign optimization have been highlighted.
• The author succeeded in integrating the new concept trough software development andapplications in shipyard environment.
• The development part of this PhD thesis is a huge work which is not visible in thisdocument.
• A strong link between theory and industry has been created. Indeed, this PhD thesishas two parts. The first one focusing on the theoretical approach (new concept) andthe second one focusing on the development of industrial applications.
• The author succeeded in putting into practice complicated and challenging conceptswith industrial concerns.
• The experience of previous European projects has been collected and integrated intothe new models presented in this PhD thesis. This provides a good dissemination anda good exploitation of those EU projects.
1.6 Context
Nowadays, the European shipbuilding industry faces competition from Asian competi-tors whose low labour costs are hostile (see section 2.2.3). The solution was therefore thedevelopment of tools and techniques to improve the performance of shipyards. The engi-neering departments and universities have therefore started to tackle this problem throughthe following three points:
• improve the quotation and design steps in order to quickly provide an optimum solutionin term of structure, production and cost,
• minimise costs and production time through automation and improved schedules,
• maximize product quality.
10
In this context, the European Community has launched projects such as InterSHIP (2004-2006), MARSTRUCT (2004-2009), VISIONS (2005-2009), IMPROVE (2006-2009) in whichthe ANAST research group is actively participates or has participated actively.
In the framework of cost reduction and reduced production times, various tools haveemerged:
• A first category of software aims to accelerate the initial design stages and to optimisethe structural design. The LBR5 [Rig99, Rig03a] software developed within ANASTdetermines the optimal structure of the ship under one of the following three aspects:minimum cost, minimum weight or maximum flexional inertia.
• Other tools can then generate 3D plans of the ship. These CAD data are then usedby software such as TRIBON or NAPA in order to generate production data (CAM)which detail parts and assemblies.
• Beyond that, software for cost estimating give the possibility of detailing the workloadlinked to each part of the ship in order to facilitate the dispatching of productionbetween the various workshops.
• More downstream, decision support and simulation tools [CBL+08, CFHR09] are usedto establish the production planning of the overall shipyard.
Consequently, this research project is the logical continuation of the research tasks carriedout by ANAST since more than 20 years. Indeed, the scientific skills developed by ANASTconcern ship design, analysis and optimisation of naval structures (LBR5, FEM, etc.), pro-duction simulation, design of advanced tools, strategic analysis and decision-making aids.All this research requires continued improvement by developing tools for costs productionevaluation integrated at all the design stages.
As mentioned before, during the research of this PhD thesis, we had the occasion tobe involved in several research projects within research and industrial frameworks. Theseprojects are presented in the next sections as well as how we took part in these researches.For each project we highlight the skill and the major aspects that have been useful for theknow-how of the author (see Fig. 1.1).
PhD
Competitiveness
Integration
Innovation
Optimization
Networking
LBR5
Inte
rSH
IP
MARSTRU
CT
IMPROVE
VISI
ON
Figure 1.1: Framework of the PhD Thesis
11
1.6.1 InterSHIP - Competitiveness
The European InterSHIP1 project aims to increase the competitiveness of European ship-builders by better integrating tools and methods for design and manufacturing of complexone-of-a-kind vessels. This IP project enabled shipyard engineers to consider leading edgeknowledge in environmental, safety, comfort and cost efficiency aspects in simultaneous en-gineering, thus making sure that optimum solutions can be obtained for the total life-cycleof complex ships. InterSHIP focused on improving vertical integration between shipyards,owners, suppliers, classification societies and horizontal cooperation between European ship-yards. Aspects like education and training, innovation management in shipyards and co-operation with other projects and industries were covered in horizontal activities, makingsure that the project results are quickly implemented for the benefit of the entire Europeanmaritime community.
The main objectives were:
• To increase significantly the competitiveness of European shipyards to reach a world-wide market share of 90% for cruises vessels and 50% for RoPax vessels and ferries.
• To develop the best product by seeking an optimum solution between the design,quality, safety, environment and efficiency.
• To drastically reduce building and development costs (25-30%) as well as lead time ofinnovative solutions (20-30%).
We can see that the objectives of this project were complementary to the purposes ofthis PhD thesis. Moreover, the author has been particularly involved in the sub-project II.1devoted to "Design for Production and Costs" where two different directions were explored.
1. The first one is a cost assessment during every design stage in order to direct and toinfluence the work of the designer by a real time follow-up of the life cycle cost. Mostof the costs in a shipbuilding process occur in the production phase, while they aredefined in design. Design for production and effective methods of cost estimation in alldesign phases are therefore the key for cost reduction. However considering cost andproduction requires that the designer has extensive knowledge and sophisticated tools.The major issue is to provide efficient cost and complexity evaluation tools so thatthe designer can assess the influence of various alternative designs on the productionperformances.
2. The second one is a production simulation. The increase of the productivity in eachproduction workshop is an essential problem to work on the efficiency and the compet-itiveness of a shipyard. The success of an optimal production consists in realising, inan appropriate manner, the division of the ship into sub-structures, the repartition ofthe sub-structures manufactured in the different workshops (space allocation) and thechoice of production parameters in each workshop (flow management). The major issueis to offer efficient production simulation tools so that the production managers canassess the production performance of various alternative forms of production schedul-ing on the overall fabrication performances. The PhD of Mr Bair has covered thisresearch field [Bai09].
1Integrated Collaborative Design and Production of Cruise Vessels, Passenger Ships and RoPax
12
In this PhD thesis, we explore the first way although the two approaches are complemen-tary and intimately connected.
1.6.2 LBR5 - Optimization
This project is related to the development of an optimisation software called LBR5. Thistool allows the optimization of ship structures following different objectives such as thehighest inertia, least weight and/or least cost. The scantling design of ships is always definedduring the earliest phases of the project. At this time, few parameters (dimensions) have beendefinitively fixed, and standard FEM is often unusable, particularly for design offices andmodest-sized shipyards. An optimization tool at this stage can, thus, provide precious helpto designers. This is precisely the way the LBR5 optimization software was conceptualized[Rig01].
The main features of the software are:
• The scantling optimisation of hydraulic and naval stiffened structures
• The 3D analysis of the mechanical behaviour of the structure
• The use of all the relevant limit states of the structure (service limit states and ultimatelimit states)
The interest European shipyards have in optimising the ship structure is basically relatedto the production cost and mainly to the labour cost. The implementation of some ofthe developments and results of this PhD study have improved the optimization solutionprovided by LBR5 software.
1.6.3 MARSTRUCT - Networking
The European MARSTRUCT2 project aims to improve the safety, the effectiveness, thereliability, the environmental behaviour and the comfort of ship structures through theapplication of advanced structural and reliability assessment within design, fabrication andoperation, leading to increased public and commercial confidence in the competitiveness anduse of water-borne transportation.
The objective is achieved through a programme for jointly executed research in the areaof the structural analysis of ships, the creation of research facilities and platforms and acontinuous programme of dissemination and communication of research results. The way inwhich the programme is designed contributes to mutual specialisation and complementaritythrough building up of the strengths and the reduction of the weaknesses of the participants.This programme will strengthen the scientific and technological excellence of the EuropeanResearch Area by integrating at a European Level a critical mass of resources and expertisethat will be able to provide European leadership in the design of efficient and safe shipstructures.
2Network of Excellence in Maritime Structures – FP6 project
13
The activities of the Network cover different areas related to advanced structural analysissuch as:
• Specification of the loading appropriate for the various modes of structural responseand strength.
• Methods and tools for the both numerical and experimental analysis of structuralstrength and performance, including aspects such as ultimate strength, fatigue, crash-worthiness, fire and explosion, resistance, and noise and vibration.
• Influence of fabrication methods and new and improved materials on the structuralstrength and performance of ships.
• Tools for design and optimisation of ship structures.
• Tools and methods of structural reliability, safety and environmental protection ofships.
This research network will be a powerful support structure to disseminate the results ofthis PhD study.
1.6.4 VISION - Innovation
The European VISIONS3 project follows three main overall objectives, which are closelyinterrelated:
• to strengthen the European scientific and technological capability to develop innovativeconcepts for the products of the future, by developing a mechanism which involves allthe stake holders and combines scientific excellence with market needs
• to contribute to the exploration of new market sectors with a competitive advantagefor European shipbuilders
• to contribute to the solution of medium and long term transport related problems, likethe congestion of roads and cities, environmental and safety hazards
Though most significant technological innovations in the maritime field have originated inEurope, the European shipbuilding industry has lost a significant part of its market shareover the last decades, mainly through aggressive competition and cheap labour. One ofthe concepts to counter this situation is the development of an integrated and systematicapproach within this project towards the development of innovative concept outlines forfuture products.
This PhD has taken advantage of this innovative framework.
3Network of Excellence of Visionary Concepts for Vessels and Floating Structures – FP6 project
14
1.6.5 IMPROVE - Integration
The main objective of the European IMPROVE4 project is to develop three new shipgenerations in an integrated multiple criteria decision making environment by using theadvanced design synthesis and analysis techniques at the earliest stage of the design process,which innovatively considers structure, production, operational aspects, performance, andsafety criteria on a concurrent basis. The product types focused on this project are newgenerations of LNG gas carriers and chemical tankers, and an innovative concept of a largeRo-Pax vessel.
The specific objectives of the project are to:
• develop improved generic ship designs based upon multiple criteria mathematical mod-els,
• improve and apply rational models for the estimation of the design characteristics(capacity, production costs, maintenance costs, availability, safety, reliability and ro-bustness of ship structure) in the early design phase,
• use and reformulate basic models of multiple criteria ship design, and include theminto an integrated decision support system for ship production and operation.
This research study has been involved in the development of an integrated design optimi-sation platform specifically regarding life cycle cost assessment and production simulation.
1.7 Conclusion
This chapter has laid the foundations for the thesis. It has introduced the research problemand research questions and hypotheses (see section 1.2). Then the research was justified(see section 1.3), the report was outlined (see section 1.4), the original contributions weregiven (see section 1.5) and the framework has been described (see section 1.6). On thesefoundations, the report can proceed with a detailed description of the research.
4Design of Improved and Competitive Products Using an Integrated Decision Support System – FP6project
15
16
Chapter 2
State of the art
2.1 Introduction
This chapter aims to present the three major elements cited in the title of the PhD thesis:the shipbuilding industry (see section 2.2), cost assessment (see section 2.3) and ConcurrentEngineering (see section 2.4). The chapter aims to present the theoretical foundation onwhich the research is based by reviewing the relevant literature to identify research issues,difficulties and challenges. The inefficiencies of contemporary practices are also highlightedhere.
2.2 Shipbuilding industry
As mentioned in section 1, the present PhD thesis concerns a specific sector: the ship-building industry. In this section we present how and why this industrial sector is a non-conventional industry and what its main properties are.
2.2.1 Shipbuilding history
2.2.1.1 Welding and riveting
Initially welding was used in ships as a means of repairing various metal parts. During theFirst World War various authorities connected with shipbuilding, including Lloyd’s Register,undertook research into welding and in some cases prototype welded structures were built.However, riveting remained the predominant method employed for joining ship plates andsections until the Second World War (see Fig. 2.1 and 2.2). After the war the use anddevelopment of welding in shipbuilding was widespread, and welding has now totally replacedriveting [Eyr01]. There are many advantages to be gained from employing welding in shipsas opposed to riveted construction (see Fig. 2.3). Welding has many advantages in bothbuilding and in operating the ship.
The advantages are:
• It is easier to obtain watertightness with welded joints (structural continuity).
• Joints are produced more quickly (less labour is required).
17
Figure 2.1: Ship production progressed considerably
Figure 2.2: Shipbuilding in the United States during WWII (1943)
18
• Reduced hull steel weight (see Fig. 2.3); therefore more deadweight1 (DWT) .
• Smoother hull surface with the elimination of overlaps leads to reduced hull frictionresistance, which reduces fuel consumption.
However, the welding process has the drawback of facilitating the propagation of cracks.It is for this reason that the riveting process is still used to assemble the fuselages of planes.
(a) Riveted assembly (b) Welding assembly
Figure 2.3: Welding and riveted assemblies
2.2.1.2 Blocks construction
During the Second World War (WWII) a large number of merchant and war ships werebuilt in a short period of time. This requirement speeded up the adoption of welding inshipyards, and often led to the application of mass production techniques in shipbuilding.Subsequently, the traditional assembling strategy has been replaced with a modular construc-tion strategy as dry dock and lifting capabilities at shipyards have improved. Prefabricationof ship units, which means the construction of individual sections of the ship structure priorto erection (see Fig. 2.4), became a highly developed skill (see Fig. 2.1). This is also knownas block construction [Eyr01]. This change enabled shipbuilders to effectively outfit2 largersections of the ship earlier in the construction process rather than relying on the less efficientold practices3 that characterize dry dock construction.
Using a modular construction strategy, relatively small steel parts are joined to form sub-assemblies; sub-assemblies are joined to form assemblies; units or blocks, and units are joinedtogether to form larger blocks of the ship. Typically, the ship is then erected in the dry dockblock-by-block until the ship is complete. The various elements (subassemblies, assemblies,units, and blocks) are almost always unique in size, shape, and weight (see Fig. 2.5) andlargely depend on:
• the size of the final ship,
• the lifting and handling capabilities of the shipyard,
• and the extent of outfitting completed within the ship.
1DWT – Deadweight Tonnage is a measure of how much mass or weight of cargo or burden a ship cansafely carry. Deadweight tonnage was historically expressed in long tons but has now largely been replacedinternationally by tonnes. Deadweight tonnage at any given time is defined as the sum of the weights ormasses of cargo, fuel, fresh water, ballast water, provisions, passengers and crew.
2Pre-installation of equipments, pipes, electrical cables, etc.3i.e. assembly plate by plate
19
Figure 2.4: Exploded view of ship by blocks
(a) Section with a simple shape (b) Section with a complex shape
Figure 2.5: Shape of simple and complex sections
Figure 2.6: Turning over of a section
20
Units or sections may be constructed under cover, which is an attractive advantage not onlybecause of working conditions, but because of the better welding conditions. It is possible toturn units over to allow flat welding which is easier to perform and likely to provide betterquality (see Fig. 2.6). There is great advantage in keeping vertical and overhead4 weldingto a minimum [Eyr01]. Also central services are more readily available at the workshop,with gases for cutting, air for chipping, and electric currents for welding, being placed whereneeded. Production planning techniques may be adopted with prefabrication sequences, thematerial and labour being planned in unit groups and the whole shipbuilding sequence beingcontrolled to fit the time allowed on the berth, or dry dock.
2.2.1.3 Industry of labour – Key players over the world
The shipbuilding industry is on a never-ending tour of the world. Over the past halfcentury it has put on great performances in Europe, but since the end of the twentiethcentury it has been an almost entirely Asian show. European predominance was challengedin the late 1950s by Japanese builders, and by the mid-1960s Japan had become the dominantplayer on the shipbuilding scene [Reg04] (see Fig. 2.7(a)).
The GT6 shown in Fig. 2.7(a) fails to present the differences in the amount of work requiredto build tonnage of various vessel types. A 50 000GT cruise vessel demands a lot more yardcapacity to build than does a 50 000GT bulk carrier. The Compensated Gross Tonnage(CGT) is the GT value multiplied with a compensating factor as illustrated in Fig. 2.7(b).Vessel deliveries expressed in CGT thereby provides a more balanced view of the productionthan DWT, GT or the number or vessels (see Fig. 2.7(c)).
Historically, the industry has suffered from the absence of world wide rules and a tendencyto (state-supported) over-investment due to the fact that shipyards offer a wide range oftechnologies and employ a significant number of workers.
For the governments of industrialising countries, shipbuilding is sometimes seen to be avery practical political instrument to develop technologies and acquire key new knowledge[CES07a]: examples from the past are Japan in the 60’s, Korea in the 80’s and todayChina and Vietnam. Japan used shipbuilding to rebuild its industrial structure, Koreamade shipbuilding a strategic industry in the 1970s and China is now in the process ofrepeats these models with large state-supported investment in this industry. The evolutionof Chinese shipbuilding over the last decade has been impressive with a six fold productionincrease from 2000 to 2007 and quasi exponential growth of new orders and order books. In2006, we have seen that over 80% of the ship orders have been placed in Asian countries[VN06]. As a result the world shipbuilding market suffers from over-capacity, depressedprices7, low profit margins, trade distortions and wide-spread subsidization.
Where state subsidies have been removed and domestic policies do not provide support,in high cost nations8 shipbuilding has usually gone into steady, if not rapid, decline. TheBritish shipbuilding industry is one of many examples of this. From a position in the early
4Ceiling5Included scheduled 2004-20056Gross Tonnage (GT) is a unit less index related to a ship’s overall internal volume. GT is calculated
based on the moulded volume of all enclosed spaces of the ship and is used to determine things such as aships manning regulations, safety rules, registration fees and port dues.
7Although the industry experienced a price increase in the period 2003–2005 due to strong demand fornew ships which was in excess of actual cost increases
8Especially European countries.
21
(a) Based on DWT
(b) Approximate CGT value 50 000GT ship
(c) Based on CGT
Figure 2.7: Vessel deliveries5 by year and building country [Reg04]
22
1970s where British yards could still build the largest types of sophisticated merchant ships,British shipbuilders today have been reduced to a few defence contracts and repair work.
Thanks to the superior quality and productivity of its shipyards, South Korea is theworld’s largest shipbuilding nation in terms of tonnage and numbers of vessels built, in spiteof high labour costs. China is currently the third largest shipbuilding country and willovertake Japan in the near future.
However, current trends indicate that China9 will soon be the largest shipbuilding nationin the world. Recently, as of May 2007, 425 new building contracts have been taken byChinese shipbuilding companies; this comes to 45% of the global vessel order intake. Interms of orders and projected output, China is now statistically the largest shipbuildingnation in the world, owing largely to recently increased investment in the industrial base, aswell as the fact that South Korean and Japanese shipyards are currently operating at 100%capacity and are predicted to be backlogged for the next two years at least.
Strong new development has been recently seen in countries like Vietnam and Brazil.They are waiting to get a bigger slice of the cake.
2.2.1.4 Shipyard strategies
There are three strategies which cover the range of production scenarios that are encoun-tered in the shipbuilding industry. The scenarios are:
1. Multiple ships based on standard shipyard designs (sister ships)
2. New ship design which is familiar to the shipyard
3. New ship design which is unfamiliar to the shipyard
Many shipyards have tried to emphasize the first type in their business line in order tomaintain productivity, especially in western countries [Koe02] (see learning curve in section4.3.2.3). When standard ships are not demanded, shipyards will book orders for the secondproject type. The third type of the project represents a strategic entry into a new market.For instance, we can cite the STX France shipyard (the former Akeryards France) who hadto decide to manufacture LNG carriers in the nineties in order to complete their order bookwhen they did not have enough cruise ships to produce.
Two global trends can be observed in the world shipbuilding market. The constructionof high value ships in Korea, Japan and Europe, and the construction of low value shipsin China and Vietnam. Both types of vessel are targeting completely different goals, forshipyards and for ship owners (see Tab. 2.1 and Fig. 2.8).
Why do we observe these two types of production strategy? The answer to this questionis dependent primarily on the production volume and variability. At one extreme of theshipbuilding industry, a few shipbuilders are building highly customized products (nichemarket), one or very few at a time. At the other extreme, a few shipyards are building onlystandard ships of a single type in large numbers (mass market).
9China is expected to reach the top position by 2015 - at least in DWT terms [Reg04]
23
In the case of the European shipbuilding industry focusing niche market, the attentionis more focused on the product than on the process. The production process should betailored to the specific parameters of the ship under construction (often a prototype). Thus,the attention to the life cycle of the product and the production process becomes naturalwhen the products are numerous, as in the case of vehicle production (see Fig. 2.9). In theshipbuilding field, designers are often more interested in the performance prediction of theproduct and verifying compliance with the required functionality.
Requiring the implementation of new tools for a quick and reliable assessment of theproposed solutions in the design phase is crucial for all industry sectors, but even more forshipbuilding.
High value ships Low value ships
Market type Niche markets Mass markets
Ship type Cruise Liner, Ferries, RoPax, Ro/Ro, Yachts Container, Bulker, Tanker
Series Small series (1-5) Long series (20-60)
Modifications in series Significant Limited
Customer requirements Highest Moderate to low
Design type Custom made design Not custom made
Quality Moderate to high Low
Delivery On time Uncertainties
Table 2.1: High and low value ships
Technolo
gy
Price level
Low HighMedium
Low
Hig
hM
ediu
m
Marketability
NicheMarket
MassMarket
Vietnam
China
Japan
EuropeKorea
Figure 2.8: The key players over the world – Marketability
2.2.1.5 Ship owner strategies
The relative importance of the different features of a ship’s performance has been evaluatedby [Bux76]. Tab. 2.2 shows the economic benefit based on 100 ships. The increase inNet Present Value (NPV) resulting from 10% improvement in various design, building oroperating features has been calculated, while holding other features constant, except thosedirectly related.
24
Features Type NPV (£M)
Fraight rate Design 180.6
Load factor Design 154.9
Steaming distance Operating 112
First cost Building 56.6
Power requirement Operating 38.1
Fuel cost per tonne Operating 35.1
Specific fuel consumption Operating 34.4
Building time Building 33.3
Port turnround time Operating 30.3
General cargo handling cost per tonne Operating 26.5
Hull steelmass Design 22.9
Crew costs Operating 22.5
Lightship displacement Design 20.4
Life of ship Design 19.9
Maintenance cost Operating 18.8
Registered tonnage Operating 12.6
Port charges Operating 12.1
Hull construction labour cost Building 11.5
Other running costs Operating 10.6
Speed loss with age Operating 8.8
Steel cost per tonne Building 8
Time out of service Operating 5.4
Table 2.2: Relative importance of a 10% improvement of various features of ship performance[Bux76]
25
Increasing freight rate by 10% shows the greatest improvement because it applies to theentire income side of the economic calculations, but of course it is outside the control of theshipowner and shipbuilders. Increasing load factor is also valuable, obtainable by greaterversatility of design (e.g. multi-purpose ships) or better management of operations (e.g.computerised operation ship scheduling). The effect is less than that of freight rate, as cargohandling costs will also be increased, and will offset some of the gains. While reducingsteaming distance10 appears very worthwhile, 10% reductions in practice are not possibleoverall. Fractional gains are, however, possible from better course-keeping or more accuratenavigation or the opening of short-cut channels. Reducing initial costs is always worthwhile,but the expected gain is so small. Reduced power requirements, resulting in both reducedmachinery cost and reduced fuel consumption, are very important. The reduction of fuelcost per tonne is also a very efficient method to increase the NPV. The reduced building timefrom contract to delivery is only assessed by its effect in bringing forward in time all cashflows.
Tab. 2.2 assumes a standard improvement of 10% but clearly gains of this amount are notequally attainable for each feature in the table. It would be far more difficult and costlyto reduce first cost by 10% than speed loss with age. It is possible to make a subjectiveassessment of the level of improvement which might be attained over a given time period.The results show that the greatest pay-offs may be expected from the following features[Bux76]:
1. Reduced building time
2. Reduced crew costs
3. Reduced specific fuel consumption
4. Increased life of ship
5. Reduced hull labour costs
6. Reduced maintenance costs
7. Reduced steel mass
8. Reduced first cost
9. Reduced power requirements
This suggests that improvements in the following areas will lead to the largest benefits:
• Ship production technology
• Design procedure
• Maintenance and reliability
• Improving material and coating
This PhD focuses on the three first points cited earlier.
10Steaming distance is the shortest distance between two ports, which a ship traverses while sailing fromone port to another
26
2.2.1.6 Tools to support design and production (Computer Aided Engineering– CAE)
Whereas up to 1950, the drawing of the hull lines and the main ship elements was per-formed at full-scale on a floor, and thereafter projected via microfilm onto sheets, today CADand more recently CAM have completely replaced these old fashioned methods. Common toall world-class shipyards is the move towards a greater integration of design and production.This process encompasses engineering, planning, management and procurement. Such a fullyintegrated approach is supported by computing technology and is referred to a ComputerIntegrated Manufacturing (CIM) to support the design and production of ships.
Indeed, progress in these fields has led to the improvement of quality and accuracy,which are two essential preliminaries for effective production. In order to generate theenormous quantities of information for each ship in a relatively short period, it is necessaryto concurrently design hull structures, machinery, equipment and electrical systems (seesection 2.4).
CIM systems for shipbuilding support the increase of productivity during the productionstage by linking the design system with the production support system (see Design forProduction in section 3.3.2). Many advanced CIM systems used in shipbuilding incorporateadvanced production support systems. Such systems lead to improvements in the quality ofproduction planning and scheduling, consequently enabling improved production flow. Thesystems also enable the introduction of automated facilities/robots by electronic data of thedesign information [SSH+00]. The CIM software technology takes into account:
• Computer aided design (CAD) – is the use of computer technology to aid inthe design and especially the drafting11 (technical drawing and engineering drawing)of products. Today, shipyards use various software dedicated for this purpose suchas TRIBON, NAPA, CATIA and FORAN. They integrate not only tools for thedrawing of smooth hull forms, structural details and piping, but also tools for structuralanalysis, stability calculation, resistance, propulsion and sea behaviour. Today, CATIAis even able, along with DELMIA, to virtually simulate the production to increaseefficiency and workshop productivity [Sas03, SS00].
• Computer aided manufacturing (CAM) – is the use of computer-based softwaretools that assist engineers and workers to bridge the gap between ship design and con-struction. CAM is a programming tool that allows you to manufacture physical modelsusing computer-aided design programs. Nowadays in the shipbuilding industry, thiskind of software is able to transfer the component nomenclatures to various productionrobot interfaces.
• Product data management (PDM) – is a category of computer software usedto control data related to products (tracking the creation, change and archive of allinformation). PDM creates and manages relations between sets of data that define aproduct, and store those relationships in a database. It is a vital tool for shipbuildingindustry in order to manage the life cycle of the product.
• Enterprise resource planning (ERP) – is the planning of how business resources(materials, employees, customers etc.) are acquired and modified from one state toanother. This system is maintained in a single database where the storage and access-ing of the data are performed in real time. This system contains the data needed for
11Drafting can be done in two dimensions (2D) and three dimensions (3D)
27
a variety of business functions such as planning, manufacturing, supply chain manage-ment, financial management, material management, inventory, human resources andcustomer relationship management [ZL06]. The information should be reliable, acces-sible, and easily shared. The shipbuilding industry greatly requires this kind of systemdue to the huge number of employees and the complexity of the products. The majorbenefits of instituting these systems in shipbuilding would include improvement of userefficiency, optimization of workflows, improvement of the quality of data, and the in-creased flow of information through the various departments or sections of a shipyard[BBT05].
• Computer aided process planning (CAPP) – CAPP is an important element ofCIM because it is a vital link between CAD and manufacturing. Significant portionsof the final cost of a product are highly dependent on the specifics of the process usedto manufacture the product and not simply the material from which the product isto be made. CAPP allows for the identification and cost assessment of the machines,tooling and devices required in manufacturing the product. This PhD thesis focuseson the improvement of this type of tool, keeping in mind the link with the other designtools.
2.2.2 A non-conventional industry
Is the shipbuilding industry the same as other manufacturing industries? The answeris definitely "no". It is not the same, as we will see. However, many basic managementprinciples hold for the different industries. In a broad sense, the organization is the same.Moreover, the mechanical process in ship construction is not so very different [Fer44]. Wecan find welding, electrical work, piping, wood work and painting in other industries likethe automobile industry, aeronautical industry as well as in the construction industry. Thenhow does shipbuilding management differ from these industries or any other "repetitive"manufacturing? The marine industry might be characterized, in comparison with otherindustries, as a synthesis of many technologies from various disciplines. So, the difficulty ofbuilding a ship is significant as it combines the following elements:
1. Small series – In straight manufacturing, and especially in mass production of iden-tical items with the same design, we find a "repetition" of operations and processes,day after day. In shipbuilding, in normal times, no company is contracted to pro-duce a hundred, or even ten, ships of identical type and design, except for maybesmall "pleasure crafts". Nowadays, in our EU countries, we are lucky to get two orthree of a kind (see Fig. 2.9(a)). This difference is available for all types of ship, butparticularly for passenger and naval vessels and others of special design which havenot yet been completely "standardized". What is notable for marine structures is theabsence of prototypes which inhibits the collection of the statistical data of past per-formance [HBC+03]. This is in contrast to the aeroplane and the automobile industry.The advantage of the use of prototypes is that it reduces system complexity after afirst prototype has been established. The prototypes integrate and highlight a numberof different aspects that enhances system reliability (e.g. identification of locationssusceptible to cracks, production factors, operability).
2. Short time to market – The duration taken in product development process fromproduct concept to the finished product is quite short with regards to the other heavyindustries (see Fig. 2.9(b)).
28
3. High complexity – Shipbuilding is a complex production system that requires designand engineering skills, creativity, technical experience and material knowledge, timelymaterial procurement, and efficient planning and scheduling [HBB+06]. Products havebecome almost unbelievably complex. Since 3 decades, design complexity (see section4.4) and the expanding responsibilities on the design processes grown in an exponentialway. The complexity of a production system is observed in two major areas: operationsscheduling and production structure [PEKW94].
• The complexity which characterizes the structure of ships in terms of number ofcomponents is extreme. The number of ship components is about ten million, i.e.ten or a hundred times higher than a plane and a thousand times higher than acar [Whi94] (see Fig. 2.9(b)). The huge number of parts and systems requires aconsiderable variety of different professions in order to perform the integration ina confined environment like a ship. Moreover, the huge number of items has theeffect of increasing the workforce needed for assembly and fabrication.
• As an example of a project-oriented production, sequencing and scheduling of op-erations in ship production is highly complex, as interim products and productionstages are interrelated with each other throughout the production process. Thecomplexity of the production structure is affected by the type, quantity, and or-ganization of the operational equipment. The more the complexity of a systemgrows, the less it is possible to make specific statements about its behaviour.
4. Tripartite collaboration – There is also the fact that the shipbuilding industrystill requires tripartite collaboration between ship owner, shipyard and class and flagcompanies. Each side has a different and opposite goal. The shipyard wants to makemoney, while the owner wants the best quality for the lowest purchase and operatingcost while class and flag societies are involved in safety (see Fig. 2.10). The constructionof a ship therefore asks for constant compromise between these three stakeholders. Thisexchange of information ensures that the final product meets the design requirements,which represent the needs, desires, and specifications of the three partners.
5. Bad working conditions – The working conditions become increasingly more dif-ficult, time consuming, and expensive as the work progresses from the workshop toon board installation. Ideally the most preferable place to do work would be [Mir06]under cover, in an easily accessible area, where less support services are required andwhere tools and equipment are readily available. Thus the most cost efficient place toconstruct and assemble the vessel is the workshop. The least desirable work locationto do the work is on board. The point about work done on the vessel is that theworkers move with their tools while the material stays in one place. Many workers do"handwork" only, and have to move from space to space, especially for outfitting work.The block level, when the blocks are not too large, is a more convenient workplace thanon-board. The productivity is improved if the blocks can be assembled under coverto eliminate the effects of weather conditions. Lastly, the unit level, which involvesthe assembly of outfit modules, can produce cost savings since it is easier to access,assemble and outfit a unit on block rather than on board.
6. Low standardisation – Outfitting work like piping, electrical work and HVAC sys-tems are not the same in each space or compartment. In general one doesn’t get thechance to "repeat" operations, space after space. Ships are radically unlike a house,with its rectangular construction and straight lines. A ship has curved lines for allsurfaces close to the water.
29
7. Confined space and bad accessibility – In shipbuilding, unlike house construction,the spaces in the ship are confined and limited. Limited spaces or compartments inmost ships require the planning of fewer men at the same time in one space. A lotof workers cannot be expected to work in one small working area at same time (forinstance inside a double hull). Moreover the accessibility inside the confined space ofthe ship is an insurmountable problem and reduces work effectiveness.
8. High variety of ships – Since the construction of ships became more sophisticated,we can observe that ships for a dedicated purpose have become much more prevalentand the new design of multi-purpose ships is now uncommon. So today we can say thatfor every job, there is a different type of vessel. We can classify the ships accordingto their cargo type (see Fig. 2.11(a)), or according to their size or their hull type (seeFig. 2.11(b)). Nowadays, there are so many different types of vessel that each shipyardhas to be specialized in one or two types of vessel to stay competitive.
9. Increase of ship size – During recent years, the average size of ship has increasedrapidly for some types of ship such as container carriers, passenger ships (see Fig. 2.12)or LNG carriers. So the design and production constraints are becoming increasinglylarge. For the ship owner, big ships mean economy of scale benefits, offered by theincrease of payload.
10. Design spiral – Ship design is an iterative process, especially in the early stages. Theultimate design is postulated and then analysed and modified (see Fig. 2.16). Themodified design is re-analysed until all the parameters of the ship converge on a moreaccurate and improved solution. The reason for iteration is that ship design has so farproven to be too complex to be described by a set of equations which can be solveddirectly. Typically, a number of cycles (design iterations) are required to arrive at asatisfactory solution. In fact, major problems are identified in the course of design andthe act of resolving these problems typically perturbs the design effort in a number ofother design disciplines, requiring restarting or reworking tasks previously completed[Lam03b]. Time and cost are additional constraints that should always be taken intoaccount in the design process.
11. Contracting Business – Shipbuilding and repairing is a contracting business [Sum73].Most work involves bidding and negotiations for a price that once established cannotbe readjusted without major change in the scope of work. The price being fixed in thecontract, even if you find a design bug, you cannot modify your sale price. Moreover,the ship price must be high enough to cover all shipyard costs for production and beless than the other competitors.
2.2.3 European shipbuilding industry
As explained in section 1.6, the present PhD research is positioned at the interface ofdifferent European projects. We explain here the specificities of the European shipbuildingindustry and what are the main research trends decided by the European Community.
2.2.3.1 Recent evolution
Since Europe’s shipyards are facing strong competition from countries with lower labourcost, they have not yet developed appropriate answers. Moreover, almost every successfulmaritime invention has been developed in Europe.
30
Contr
act or
deliv
ery
tim
e
Car
Number of build
101
102
103
104
105
106
107
108
3
6
9
12
15
18
21
100
Airplane
Cargo ship
Navy ship
Power plant
(a)
Com
ple
xity
Car
Time to market
Airplane
Cargo shipNavy ship
Power plant
Long
Hig
hLow
Short
(b)
Figure 2.9: Shipbuilding industry versus other industries [TC02]
Floating vessel
Catering
Navigation & Shipping
HotelFactory
Entertainment & Leisure
Ship Owner
Ship
Shipyard
Stability
Manoeuvrability
Propulsion
Energy production
Structural integrity
Production scheduling
Components integration
Class and Flag Society
Health & Safety
Quality
Environment
Social Responsability
Figure 2.10: The ship complexity - three actors, three points of view
31
(a) Cargo Type [Eyr01]
(b) Hull Type
Figure 2.11: Classification of ship type – (1) Displacement hull, (2-3) monohull, (4) multihull,(5) pre-planning and planning hull, (6-7) hydrofoils, (8) SES and SWATH, (9) hovercraft
0
50000
100000
150000
200000
250000
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Years
GT
W W
I
W W
II
Figure 2.12: Largest passenger ship size (GT) from 1810 to 2010
32
Within two decades, European yards have gone through a fundamental transformationof production methods through consequent rationalisation and innovation coupled with ac-tive outsourcing strategies. The CESA 12 report [CES05, CES06, CES07a] shows that animpressive rationalisation coupled with active outsourcing strategies have reduced the totalworkforce directly employed by shipyards by 36% (see Fig. 2.13(a)) while the production hasincreased by 43% (see Fig. 2.13(b)).
Innovative solutions, Concurrent Engineering (see section 2.4) and the use of advanceIT applications are facilitating a high degree of specialisation allowing the European ship-building industry to assume global leadership in the production of highly sophisticated andcomplex ships with the highest added value such as cruise ships, fast ferries, advanced cargocarrying ships, dredgers, off-shore support vessels, fish production vessels, mega-yachts, re-search vessels, etc.
(a) Workforce (b) Production
Figure 2.13: The shipbuilding progression between 1985 and 2004 in CESA [CES06]
2.2.3.2 Order book of ships in Europe
After three consecutive years of increasing new orders [CEH+09], the order book of Euro-pean shipbuilding has reached a record level with nearly 17 Mio CGT at the end of 2006.Production of ships in Europe has increased by 20% as compared to 2005 while turnoverincreased by 44% in the same period. At the same time, the direct workforce employed byEuropean shipyards slightly decreased to 137 500 at the end of 2006. The major marketsectors of European shipbuilding remain passenger vessels (with a slightly increasing share)and container vessels. However, the segment of other non-cargo vessels, such as offshoresupport vessels, rapidly grew and reached close to 14% at the end of 2006 (see Fig. 2.14).
2.2.3.3 Research fields in Europe
Research developments in world shipbuilding over recent years have influenced the areasrelated to shipyard manufacturing technologies mainly from the point of view of productivityand product performance as the most important contributors to shipyard competitiveness.Europe conducts a large amount of research in joining techniques, hydrodynamic research,and design tools and methods. Based on the requirements of shipyard productivity and
12The Community of European Shipyard Associations (CESA) representing the shipyards in 15 Europeancountries covers 99% of EU shipbuilding production and more than 85% of the wider geographical Europeincluding the shipbuilding nations Bulgaria, Turkey and Ukraine. European shipyards offer products in threeareas: construction of merchant vessels, repair and conversion, and naval production.
33
product performance, various technologies were developed including processes for the effi-cient production of relative thin plate structures (e.g. laser welded sandwich structures,laser-arc hybrid welding, automatic plasma welding), to improve the economy of scale inshipyards, manufacturing by modularization and increased pre-outfitting as well as simula-tion techniques to enhance the management of the shipyard production chain [BHH+06].
Shipyard production technologies are heavily influenced by the types of ship being builtin individual shipyards, the economy of scale (size of series of similar products), interna-tional rules and regulations as well as the strategies of individual shipyards and shipbuildingnations. This explains why the equipment and techniques used in different yards as well asthe main directions in research and development differ significantly.
Flexibility in production is high in European yards through multi-skill practices, control-ling the supply chain, and better ERP methods. Advanced CAD/CAM, laser and automationtechnologies have also been developed in Europe [Gol01].
The actual priority research fields in Europe are:
• Life cycle cost assessment is becoming increasingly important to make the customeraware of the life cycle benefits of innovative solutions. An integrated view requiresdedicated methods to compare production and operational costs, safety and environ-mental aspects as well as tools for life cycle optimisation in the different design andproduction phases of a ship.
• Increased work sharing in the shipbuilding production chain, including shipyards andsubcontractors calls for integrated IT solutions for design and manufacturing, modularproduct concepts and new process chains for a more efficient work sharing.
• Similar to other countries, European shipbuilding is increasingly suffering from a short-age of skilled workforce. In order to maintain know-how within the shipyards and to
Figure 2.14: Oder book of European shipbuilders (CESA) by ship types (in % of CGT)
34
ensure that the specific skills of elderly personnel are kept, Knowledge Management isan increasingly important topic both for design and production.
• Processing of thin plate material remains an ongoing research issue for European ship-yards, with innovative laser systems being increasingly applied. A strategic line ofresearch is the flexible use of laser systems to increase economic feasibility.
• After extensive applications and researches in the naval field, composite and otherlightweight materials are increasingly finding their way into commercial shipbuilding.Further research is however required to make joining and outfitting of lightweightstructures more efficient, to find the optimum material mix and to solve problemsrelated to safety and operational requirements.
The present PhD study targets the first two European research areas related to life cyclecost assessment and integrated IT solutions for the improvement of manufacturing [CEH+09].
2.2.4 Design and conception
The designers must make enormous efforts to fulfil the ship owner’s satisfaction. Thismeans that it must be safe, reliable, and as economical to operate and maintain as possible,within the constraints imposed by technology and the ship owner’s budget. At the sametime, the designer must also integrate into the design the manufacturing or shipbuildingprocesses and procedures so as to create production-friendly features in order to ensure thatthe ship construction costs will be minimized. This requires a deep knowledge, on the partof the designer, of the most current and advanced procedures used by shipyards.
Initial Design
Concept Preliminary Contract Detailed Production
Post Contract Design
Contract
Figure 2.15: Design stage during the timeline of the project
As shown in Fig. 2.15, the initial design of a ship, also called basic design, generallyproceeds through three stages: concept, preliminary and contract design. The output of thesestages leads to a set of specifications and constraints which dictate how the product should bemade [EA98]. In the early stage of design, little information is available and few constraintshave been specified so all possible manufacturing processes should be considered. As thedesign progresses to the post contract design, more information on the product becomesavailable and a set of constraints is specified.
The process of overall ship design is an iterative one, proceeding cyclically to resolve theconflicts among the many systems which comprise the ship, each of which has its own separateobjectives and constraints, but with each contributing to the overall goals of ship performance[Wil75]. These conflicts are often serious and difficult to resolve and the compromises whichmust be made sometimes produce marked changes in the objectives of individual systems.The process of initial design is often illustrated by the design spiral (Fig. 2.16) which indicatesthat, given the objectives of the design, the designer works towards the best solution byadjusting and balancing the interrelated parameters as he goes [Eyr01]. Finally, as thedesign reaches its final stages it becomes detailed enough to allow cost evaluation.
35
A concept design should, from the objectives, provide sufficient information for a basictechno-economic assessment of the alternatives. Economic criteria that may be derived forcommercial ship designs and used to measure the shipowner profitability are the net presentvalue of ship life, discounted cash flow or required freight rate. Preliminary design refinesand analyses the agreed concept design, fills out the arrangements and the structure andaims at optimizing service performance. At this stage the builder should have sufficientinformation to tender. Contract design details the final arrangements and systems agreedwith the ship owner and satisfies the building contract conditions.
Total design is not complete at this stage, it has only just started; post contract designentails in particular detailed design and design for production where the structure, outfit andsystems are planned in detail to achieve a cost and time effective building cycle. Productionof the ship must also be given consideration in the earlier design stages, particularly whereit places constraints on the design or can affect costs.
Figure 2.16: The design spiral [Eyr01]
2.2.4.1 Concept design
The first step in concept design translates the mission requirement into Naval Architectureand engineering characteristics. Feasibility studies determine such fundamental elements ofthe proposed ship such as dimensions, shapes, configuration, General Arrangement (GA),depth, draft, power or alternate characteristics, all of which meet the required speed, range,cargo cubic and deadweight. A designer’s creative ability is much required in this designstage. The concept design is used for obtaining approximate construction costs, which oftendetermines whether or not next level of development will be initiated [Kis80].
2.2.4.2 Preliminary design
Preliminary design further refines the major ship characteristics affecting cost and perfor-mance in order to offer detailed specifications, delivery date, price, etc. to the ship owner.It is the second iteration in the design process. Certain controlling factors, such as length,breadth, horsepower, payload or deadweight would not be expected to change significantlyin this phase.
It is in preliminary design, however, that basic decisions are made, such as structuralcomponents, scantlings and the principal structural materials such as high strength steel,
36
high yield steel, ordinary steel or combination of these. Its completion provides a precisedefinition of a ship that will meet the requirements of the ship owner and sufficient detail topermit the verification of both the technical and economical feasibility of the ship [Kis80].This provides the basis for the development of contract specifications.
2.2.4.3 Contract design
The third design stage involves the preparation of both the contract specification andthe contract drawings. This stage yields a set of plans and specifications which form anintegral part of the shipbuilding contract document, further refining the preliminary design[Kis80]. The contract design becomes the legal and binding document used, especially duringarbitration, to resolve and settle the cost of the ship. A second reason for carrying out thecontract design is to increase the amount of detail and improve the accuracy of the design,so that the designer can continue to evaluate the economic and technical feasibility of thedesign, but with increasing accuracy, greater reliability and lower risk.
2.2.4.4 Detailed design
Immediately following the contract, the initiative in the design process shifts from the shipowner to the shipyard.
Detailed design is the final stage, and is the development of detailed building plans. Asonly the key plans of mid-ship part hull structure are usually made in the initial design stage,shipyards drawings of forward and aft part structures are also made in this stage. Theseplans are the fabrication, installation, and construction instructions to the professionals,machines, equipment and computers which will ultimately build the components of the ship.They include production details tailored to meet shipyard unique requirements, restrictionsand limitations.
2.2.4.5 Production design
Most necessary designs for the production are determined in this stage. It is ideal that in-formation of the detailed design is made as electronic data so that they are directly applicableto the production without additional process of work [PBB+03]. Subsequently, productionof the ship is carried out (see section 2.2.5).
2.2.5 Production and fabrication
Modern ship construction is based on modular construction (see section 2.2.1.2), with eachmodule (unit or block) passing through a series of stages (see process flow Fig. 2.17), eachof which is normally associated with specific work sites, where the process can be performedmost efficiently [Pro93]. Many work processes can also be accomplished at later stages ofthe construction process, but the work then will be done less efficiently, and thus will requiremore man-hours.
As a result, the vast majority of shipbuilding involves the simultaneous production ofmultiple products of different types [Lam03b]. The concept for building a ship as a seriesof sub products is commonly called Product-Oriented Design and Construction (PODAC)[KF93, Mir06]. This method classifies the different sub products that are assembled andsubdivides the required work accordingly.
37
2.2.5.1 Group Technology (GTech)
A key factor of Group Technology (GTech) production is to recognize that, despite productvariety and variation, there is a very high degree of similarity between most ship sub products(blocks, sections, sub section, panels, etc.). Another key factor is to recognize that even ifthe number of ships built simultaneously might be relatively low, the volume of similar subproducts (blocks, sections, sub section, panels, etc.) is substantial. These two key factorstaken together, similarity and volume of sub products, provide the foundation of the GTechproduction system [Lam03b]. The objective of this technology is to exploit the similaritiesbetween sub products to gain production economies of scale for non standard products evenwhen produced in moderate volume.
Once sub products are defined, GTech principles can be applied to systematically classifythem into groups or families having design and manufacturing attributes sufficiently similarto make batch manufacturing practical. Process workshops can then be established forthe efficient manufacturing of similar interim products providing for efficiencies of batchmanufacturing for a small numbers of ships. This specialization of work by sub productscreates the opportunity for the increased use of semi-automatic and automatic machines andtools, which can improve productivity.
The key to minimizing the cost and lead time13 in a Group Technology based shipyardis to most efficiently and cost effectively utilize production resources across all the ships tobe built at the same time. All planning and scheduling must be based on this aggregateperspective so that the focus is not on the individual ships but on the sub products.
Critical to production process control is control of sub product and associated processcost, productivity and lead time. Shipyards must develop cost and time monitoring systemsfor each workshop, work station and work team. This kind of system can be used to con-
13Lead time in the shipbuilding industry is often the time between the machining of the first steel plateand the ship delivery
Production design
CAM data
Production drawings
Delivery of plates and profiles
Surface treatment
Cutting, marking and numbering
Bending
Blocks erection
Launching
Outfit work
Sea trials
Delivery
Assembly of sub assemblies Assembly of sections/panels Assembly of blocks
Outfit work Outfit work
Production process
Figure 2.17: Production and fabrication process flow
38
tinuously track cost and time variance by individual sub product. Quality control is alsocritical for the production. If elements or sub products don’t fit with the initial requirements,rework or replacement must be carried out causing increased material and labour costs andcascading schedule problems.
2.2.5.2 Product Work Breakdown Structure (PWBS)
The design for production concept (see section 3.3.2) implies that schedule and cost areinextricably linked. In order to align cost drivers with an organized structure, particularlylabour cost, which is the most variable entity, it is essential to define a specific hierarchicalstructure sensitive to product and process requirements. This structure leads to the use ofimproved design for production trade-offs during whole design process [WKK+97, EDL+98,KF93].
The cost assessment needs to reflect the impact that the proposed shipyard’s build strat-egy has on the price information. The concept of modular construction points the way for aneed for modular cost assessment. The Product Work Breakdown Structure (PWBS) allowscosts to be packaged in terms of the modular building environment.
The goals of this structure are to provide a tool to the shipbuilding industry that allowsit to:
• enable realistic design for production trade-offs and investigation of alternative designand production scenarios at the early design stage,
• perform top-down/bottom-up cost modelling based on how ships are actually built andsupply a framework for improved cost collection (see section 2.3.4),
• create an integrated structure including design, scheduling and cost concepts,
• improve data transfer amongst design, cost assessment, procurement, production usinga common framework and description of both the material and labour content of a shipproject,
• provide a structure for 3D product model data.
The PWBS is defined as a hierarchical representation of work associated with the designand building of a ship based on product structure, product stages, and product work type[Ges93, KC99]. It is represented by the logical progression of all sub-product elements.This structure is used to summarise the scope of work and should provide the format foridentifying and cataloguing the details of the cost assessment.
Under this concept area three substructures have been created to identify the logic andinformation process necessary to develop schedule and cost alternatives during the wholeship design process.
These three substructures are:
1. Ship Work Breakdown Structure
2. Hierarchical work stages
3. Hierarchical work types
39
2.2.5.3 Ship Work Breakdown Structure
Once any complex product such as a ship has been designed, planning and engineeringwork need to be applied toward maximizing production efficiencies. The modular construc-tion technique (see section 2.2.1.2 and Fig. 2.4), including block construction and advancedoutfitting methods is the preferred shipbuilding technology for this purpose.
This work entails organizing work and resources that promote productivity and minimizenon-value added costs. The modular concept technology leads to breaking down the ship intodefinable sub-products with a path like Ship – Block – Sections – Assemblies – Sub Assemblies– Parts (see Fig. 2.18 and 2.19). The product structure sorts shipbuilding project data byproducts. All work on a ship design and construction project can be assigned a location14 inthe product structure. These products can take advantage of significant cost and schedulesavings because they enable the work to be performed under more convenient and moreeasily performed work conditions.
These technologies, however, requires more advanced product engineering to gain the fullpotential of efficiencies and cost savings. This therefore places added pressure on the alreadydifficult position of the designer who now has to incorporate and anticipate the method andsequence of production. This ship systems-oriented technique of organizing work requiresa cost collected by sub products (sub-assemblies, assemblies, hull blocks, ship zones) andmanufacturing process (machining and cutting, welding, assembling, etc.).
Ship
Block Block Block
Section Section Section
Assembly Assembly Assembly
Sub. Ass.
Part
Part
Part
Part Part Part
Sub. Ass. Sub. Ass.
Figure 2.18: Hierarchical ship structure
2.2.5.4 Hierarchical work stages
The work stage is defined as the division of the shipbuilding process, classified into con-struction and non-construction activities (see Table 2.3). Non-construction stages includedesign, planning, procurement, and other information generation activities. Fabrication,block lying, and erection on the other hand are examples of construction stages.
The work stage can be divided into different levels relating to the management processfrom design to delivery (see Fig. 2.20). As a consequence of the progress of work, thesuccessive phases can overlap for different elements of the SWBS. Costs can be assessed atvery high levels during the concept stages of design or they can be estimated at very low levelsfrom detail material bills. In between these levels, there are cost assessment relationships
14Production path
40
(a) Ship (b) Block erection
(c) Double bottom section (d) Accomodation section
(e) Stiffened panel (f) Various subassemblies
Figure 2.19: Ship work breakdown structure
41
that provide more accuracy from available design information but without the precision thatmight be obtained after detail design and engineering has been completed.
Early design
Basic design
Detailed design
Production design
Fabrication
Erection
Delivery
Su
ba
ssem
bly
Ass
embly
Pan
el
Block
Grand block
On
Board
Ou
tfi
ttin
g
Launch
Test
Figure 2.20: Ship cycle
Design Fabrication (shape, plate) Block Launching
Planning and scheduling Sub Assembly Huge block Testing and evaluation
Procurement Assembly Ship erection Delivery
Material Management Panel On board outfitting Post delivery
Table 2.3: List of typical hierarchical work stage for a ship
2.2.5.5 Hierarchical work types
The shipyard typically has its own hierarchical work types that are the basis for thecompanies’ operating systems that collect and manage return costs. These return costsprovide the historical database used by the cost estimator. The hierarchical work type (seeTable 2.4) is a means for summarizing the scope of work and should provide the templatefor identifying and cataloguing the details of the cost assessment. Work types are definedto differentiate the work by skill, facility and tooling requirements, special conditions andorganizational entities work types. The work type is used to link a range of work to anintermediate product at a specified stage.
2.2.6 Productivity and competitiveness factors
2.2.6.1 Definition
Competitiveness [Mar92] defines competitiveness as "the ability to win and execute ship-building orders in open competition and stay in the business". Many factors influence thecompetitiveness of individual shipyards.
42
The three main components in competing for a customer’s decision to buy are the quality,the time and the cost. We have an equilibrium between these factors. For instance, passengerships require high quality and have rather generous delivery times and production costs whilecargo ships are subject to less time to delivery and production cost for acceptably less quality.
When is a shipyard competitive regarding another? A shipyard is competitive if it is ableto provide a ship faster and cheaper than the others for a given quality - there is no incentiveto deliver more quality than required. The quality aspect comes from quality of design aswell as quality of production process. Nevertheless, quality management is a critical point forproduction due to the cost and time aspects. The challenge of quality management consistsof delivering the minimum required quality within the minimum time and minimum costbudget.
A cost-effective production system, production-friendly ship design, shorter lead timesand financial engineering maintain the competitive position of a shipyard [HK96].
Productivity is an essential factor in competitiveness, but it is not the only parameter.
Productivity The words "productivity" and "efficiency" are frequently employed in thecontext of the performance assessment of processes, companies, etc. Productivity refers tomeasures of output from production processes, per unit of input [Lam02]. Labour produc-tivity, for example, is typically measured as a ratio of output per labour-hour, an input.Productivity may be conceived of as a measure of the technical or engineering efficiency ofproduction. Having made this simple definition it suddenly becomes very complex. How doyou measure output and input?
The efficiency of a company consists of two components: technical efficiency and allocativeefficiency [CROB05]. Technical efficiency concerns the company’s capability to produce themaximum output for a given input vector, or alternatively to produce a determined outputvector with a minimum input. Allocative efficiency concerns the capability to employ inputsin optimal proportion for determined prices and production technology.
For this study, the relevant indicators are related to technical efficiency. Technical ef-ficiency measures can be input or output oriented. Input-oriented efficiency correspondsto the minimum input for a given output. Output-oriented efficiency corresponds to themaximum output for a given input.
In the shipbuilding industry it is common to consider labour productivity. This can bedone at any level of the ship construction process, so a wide variety of measures has beendeveloped. The most famous one is the man-hours per tonne of steel weight.
2.2.6.2 Measurement of the productivity
Cost assessment should directly reflect the shipyard’s relative level of productivity [DT04].
Electrical Welder Paint
Hull Outfitting Production services Material Handling
HVAC Block assembly Machinery
Operations control Engineering Management
Structure Piping Quality assurance
Materials Testing Administration
Table 2.4: List of typical hierarchical work types for a ship
43
Nevertheless, it is important to consider why productivity is measured, and this can bereduced to two reasons:
• to determine what level of productivity is appropriate;
• to measure changes in the level being achieved over time.
The challenge for a shipyard that wants to maintain its competitive advantage by reducingcosts and contract schedules is to find areas where savings can be achieved. Savings can besignificant and can come from a variety of sources. In order to identify what changes willprovide the most significant savings, a shipyard must be able to evaluate its operationsin quantitative terms. This means that the shipyard must have implemented means formeasuring productivity and efficiency as accurate’s as possible. The performance indicatormeasurement system should provide an overview of performance that will indicate if a no-change or a change in the design or the production process is proving to be effective.
Shipbuilders have used some form of productivity metric as a basis for estimating andplanning over their long history. Often, the measurement of productivity regarding thefabrication of the hull structure of ships is the man-hours per ton of steel weight. Forinstance, the productivity might be 12 man-hour/ton to assemble flat steel panel sections,such as deck assemblies. This performance indicator is based on a facility using largelymanual welding of stiffeners to the plates. After the investment of an automated panel linethe performance indicator could be 9 man-hour/ton. It means that the productivity hasbeen improved by 25%. We further on demonstrate that this is an inefficient metric (seesection 4.4.4.2).
Imbalances in the production system are inevitable, as a result of product changes, andthese must be removed by management action. A good level of overall productivity and atight management control system are essential. However some other aspects of productivitymeasurement can cause problems [BC92]:
• The accuracy and the consistency of the data which is fed back from production andwhich forms the basis of the productivity measurement (see section 2.3.5.12).
• The performance of any individual department, or work station, is irretrievably con-nected to those which precede and follow it.
It follows that the absolute level of productivity is relatively difficult to assess. Howevertwo authors have recently improved the measurement of productivity in the shipbuildingindustry.
[Ber03] presents a model for the assessment of competitiveness of shipyards where theproductivity is evaluated with the equation 2.1.
P D = P T × LP (2.1)
where P D Productivity,P T Technical Productivity [CGT/man-hour],LP Labour Performance [man-hour/man-year] indicates the actual hours
worked by an average employee per year. If no statistical data is avail-able, it is possible to estimate this figure on the basis of : number ofworking days per week, number of working hours per day, number ofholidays per year.
44
[Lam02] presents two equations for the calculation of productivity of shipyards basedon ship database coming from Lloyd’s register between 1997 and 1999. The first one (seeequation 2.2) takes into account few parameters with a correlation coefficient of r2 = 0.981and the second one (see equation 2.3) takes into account some additional parameters suchas production ratio or vertical integration with a correlation coefficient of r2 = 0.987.
P D = 3302 × BP −4.73× TE0.24 (2.2)
P D = 111 × BP −3.00× TE0.27
× P R0.6× DP 0.41
× V I−0.66× ST −0.88 (2.3)
where P D Productivity,BP Best Practice rating – The best practice rating of the shipyards was de-
veloped based on the technology level approach used in the [SCL95]. Theapproach assigns five technology levels to eight elements, namely: steel-work production; outfit production; other pre-erection; ship construc-tion and outfit installation; layout and environment; amenities; design,drafting, production engineering and lofting; organisation and operatingsystems,
TE Total number of Employees,P R Production Ratio – It is the ratio of the total number of employees TE
by the number of production employees,DP Dual Purpose – The DP = 1 if the shipyard is building commercial or
naval ships only, and the value of DP = 2 for a shipyard producingcommercial as well as naval ships,
V I Vertical Integration – Vertical integration is the ratio of value added bythe shipyard versus the total ship value and is defined by the percentageof labour cost to total cost,
ST Ship delivered/ship Types – ST is a parameter that takes into accountthe number of total ships built compared to the number of "series" shipsbuilt over a given time, such as three years.
2.2.7 Inefficiencies of traditional design and production processes
Shipbuilding is one of the oldest industries, and that may partly account for the historicfact that the old ways of building ships continued to exist. With the advent of new industries,with new ideas, new methods such as automobile and aircraft manufacturing, shipyards allover the world take notice more and more of modern methods. Nevertheless, we can observea certain inertia to changing due to past history.
2.2.7.1 Poor production consideration during design
As mentioned in section 2.2, there are many characteristics of the ship which are estab-lished very early in the design process.
The traditional design method does not adequately include production engineering ormaterial/supplier/logistical concerns early enough to have a substantial, positive impact onthe design [Bar96]. Taking a integrating approach, and using Computer Aided Cost (CAC),analysis and synthesis tools can mitigate these traditional design failures.
45
2.2.7.2 Low productivity and high labour cost
The ratio between productivity (see section 2.2.6) and labour cost is quite low for Europeanshipbuilding in comparison with the other shipbuilding actors (see Fig. 2.21). Increasing theproductivity of European shipyards is a critical task to regain competitiveness against theother players.
Several solutions can be investigated:
• The optimisation of the industrial process
– Improved accuracy and quality control in order to reduce rework, reduce skillcontent and increase use of robots
– Schedule compliance and schedule-driven process improvement
– Sequencing improvement
– Computer-Integrated Manufacturing (CIM) – integrated design, planning, pro-curement, production rules, etc.
– Etc.
• The optimisation of the industrial layout
– More use of automation – welding robots, painting robots, automatic line-heating,etc.
– Reduction in the number of workers
– Minimization of staging
– Laser steel processing
– Etc.
• The optimisation of the product design
– Faster design-build time – benefits of series production used and throughput ismaximised
– Reduction of number of parts
– Design for production
– Standardisation
– Modularization
– Etc.
2.2.7.3 Increasing number of production constraints
The increasing automation of the fabrication process leads to more and more constraintsconcerning the design [Neu97]. In order to have an adaptation between the product andthe shipbuilding workshops it is mandatory to define all these production constraints duringthe design. There is a general tendency to apply CAD/CAM/CIM systems to develop theprocess chain from design to production. Nevertheless, these design systems are not yetsufficiently effective to enable the introduction of all the production constraints.
46
The levels of most of the life cycle costs are set out during the design, and therefore to alarge extent it is the designer who is responsible for the cost effectiveness of fabrication andease of maintenance. The problem is that there exist no general clearly established rules forsuch design, because it depends on the:
• ship type
• facilities of the shipyard
• etc.
To increase the effectiveness of the designer and the quality of the product, CAD/CAM/CIMsystems must be continuously improved.
Figure 2.21: Some general competitiveness factors [KNB03]
47
2.3 Cost assessment
2.3.1 Cost "estimate" and cost "assessment"
The word "estimate" is used with several different meanings. Dealing with the cost of workit is necessary to confine the discussion principally to one specific definition and meaning.The word "estimate" may mean:
To make a "rough" approximation of probable cost; "an estimate of what it wouldcost"; "a rough idea how long it would take". This is the only kind of estimate possible whenthe specifications and parameters are not yet defined very well. This kind of preliminaryestimate is based on the best information available; comparison with similar projects andknown construction cost trends. They are therefore not very precise. We like to comparethe precision of a cost assessment to the shooting at a target. Precision is the tightness ofthe cluster of shots (see Fig. 2.22(b)).
Rough estimates are useful to prospective customers when the total funds available arelimited and it is necessary to decide without delay what work shall be undertaken and whatpostponed or not undertaken. However, a "rough estimate" should never be submitted asa bid but only as a matter of general information. A rough estimate in bidding may causeeither the loss of a desirable contract, or may lead to an undesirable contract which losesmoney.
To make an "assessment" of probable cost starting from data obtained during theadvance planning stage; i.e., comprehensive design, technical drawings, etc. After the designdevelopment phase is completed, a more accurate cost estimate can be prepared. We like tocompare cost accuracy to the rings on a target. Accuracy is the relative distance from thecenter of the target (see Fig. 2.22(a)).
(a) Good accuracy, low precision (b) Good precision, low accuracy
Figure 2.22: Accuracy and precision of cost assessment
When the project is in an early conceptualization, the cost estimate is less precise, similarto the outer ring of the target. As the project design progresses, the estimate approachesthe smaller rings and is therefore more precise. Focusing on accuracy before precision resultsin an inherently unstable costing system where two different people cannot obtain the samecost and the same person cannot duplicate the same cost twice. Similar to target shooting,
48
once shots are clustered, it is a much easier task to move the cluster towards the center ofthe target.
In this PhD thesis, the second definition is considered; with the use of the word "assess-ment". However, we try to meet both the precision and the accuracy of the cost assessment.
Total cost is composed of material, labour, overheads, etc. Usually the largest portion ofa cost assessment consists of the cost of shipyard labour. In this PhD thesis, attention willbe especially devoted to estimating the labour. It is also the most difficult. Estimating thecost of material presents no major difficulty for an experienced estimator after a completelist of the material required is prepared.
2.3.2 Production cost assessment
For most companies, cost is the most influential factor in the outcome of a product orservice. Most often, reducing cost is essential for survival. To compete and qualify, companiesare increasingly required to improve their quality, flexibility, product variety, and noveltywhile consistently maintaining or reducing their costs [RK03]. In short, customers expecthigher quality at an ever-decreasing cost. Not surprisingly, cost reduction initiatives areessential in today’s highly competitive market place. Concurrent engineering is one of manyinitiatives (see section 2.4). Since cost has become such an important factor of success,project development needs to be carefully considered and planned. It is essential that thecost of a new project development be understood before it actually begins. It could meanthe difference between success and failure.
The cost of a ship is the sum of all the labour and material costs involved in the construc-tion including any overhead costs. The final price of a ship will include allowances for capitalcost financing, inflation, and shipyard profit [Mir06]. Due to the particular nature of theshipbuilding industry (see section 2.2.2) and the high degree of variation between particularprojects, the necessity of reasonably accurate cost assessments is imperative (vital).
Cost assessments during the early stages of ship development are crucial. They influencethe go, or no go, decision concerning a new development [RK03]. If an estimate is too highit could mean the loss of business to a competitor. If the estimate is too low it could meanthe company is unable to produce the ship and make a reasonable profit.
An ability to carry out effective, detailed, and reliable ship cost assessment could createa change in the way the shipyards are able to negotiate their contracts [Mir06]. A greaterunderstanding of the factors that drive costs can hopefully lead to a reduction in cost overrunsfor two reasons:
• firstly, designers will be in a better position to quickly carry out trade off studies andtherefore develop a better understanding of how their designs affect cost,
• second, with an ability to carry out reliable cost assessments at the preliminary level,the shipyards will be able to negotiate more favourable contract terms that could reducecosts.
49
2.3.3 Types of cost assessment
Cost assessment occurs at various stages of the ship design development (see section 2.2.4).The importance of a good cost assessment, particularly at the early levels of design, can becrucial when comparing different design proposals [Bro88]. The approach used to developthe cost assessment largely depends of the availability of the data for the cost assessmentprocess.
Today, cost assessment is a key task of an integrated ship design. This means (seeFig. 2.15):
• At the initial or basic design stage: to validate the budgets and give a reliable biddingprice,
• At the detailed design stage: to plan the deadlines and establish the production sched-ules15,
• For an effective design for production: to distribute the workload between the variousproduction workshops and assess the productivity.
2.3.3.1 Initial design stage
This stage is used to validate the budget and give a reliable bidding price in order to helpthe manager to make, as soon as possible, the best strategical decision.
Concept design stage The cost assessment during the concept design stage is at a veryhigh level and requires rather broad assumptions about the general methods and organizationof the design, engineering and construction processes. This level is used to validate theeconomic feasibility of the project.
Preliminary design stage The cost assessment during preliminary design stage remainsat relatively high level, but there is more detailed information about the ship design withregard to hull structure, equipment and outfit systems. Preliminary design cost assessmentis able to reflect the effects of alternative building strategies. This level is often used toevaluate and sanction the project. The accuracy and the validity of the cost estimate areimperative since at this stage the design will be evaluated against various options. It istherefore very important for the designer to understand how his decisions will affect costsince major trade-off studies are performed and major decisions about the future of theproject will be made at this stage.
Contract design stage The cost assessment at the contract design stage describes costson the basis of sub-products (hull blocks, outfit modules, etc.) and manufacturing processes(machining, prefabrication, manufacturing, assembly, installation, testing, etc.). Contractdesign cost assessment is able to reflect the effect of design alternatives in terms of productionengineering and manufacturing processes. The cost assessment at this stage is a tool to obtaina successful strategy for managing the detailed design process and will help to ensure thatthe final design stays within prescribed cost objectives.
15Here the target is more workload than cost.
50
2.3.3.2 Detailed design stage
At this stage, cost assessment is used to plan the deadlines and establish the productionschedules in order to help the manager to make the best tactical decision as soon as possible(see Fig. 2.29).
2.3.3.3 Production design stage
At this stage, cost assessment is used to distribute the workload between the variousproduction workshops to help the manager to take as soon as possible the best operationaldecision (see Fig. 2.29).
2.3.4 State of art in shipbuilding industry cost assessment
The usual methods of cost assessment are various and diversified. Large volumes have beenwritten on this subject; many books for each class of the construction industry [HSGC04,BBBB09, Kar07], but very few about the shipbuilding industry.
To succeed commercially, shipyards must be able to accurately assess costs. Cost assess-ment is necessary for the bid process, to change orders, and for trade-off studies. Numerouscost assessment approaches exist. They can be based on extrapolations from previously-builtships, detailed parameters, or integrated physics-based analyses. The solution for the pro-duction cost assessment differs in the required information (input data). The less informationis needed, the earlier a method can be employed in the design process. The more informationis used, the finer differences between design alternatives can be analysed. [BMCR05]
(a) Top-down (b) Bottom-up
Figure 2.23: Top-down and bottom-up methodology
The methods for estimating production cost are classified into:
• Top-Down (macro, cost-down or historical, weight-based) approaches (empirical, sta-tistical and close-form equations, etc.), see Fig. 2.23(a)
• Bottom-Up (micro, cost-up or engineering analysis, process-based) approaches (directrational assessment), see Fig. 2.23(b)
2.3.4.1 Top-Down approaches
The top-down approach is a parametric cost assessment methodology which uses empiricalrelationships between product parameters and costs as a means to estimate the cost of newships [GD96]. In this case, the top-down method means that the ship cost is predicted from
51
its higher level specifications, instead of its detailed design which may not be available atthe time of estimation. Parametric relationships are estimated by using statistical regressiontechniques from a historical cost database. A parametric estimating system can then becontinuously refined and re-calibrated.
The top-down approach, also called weight based approach, determines the productioncost from global parameters such as the ship type and size, weight of the hull, the blockcoefficient, ship area, complexity, etc. The relationships between cost and global parametersare found by the evaluation of previous ships [Bar96]. Thus, the top-down approach is onlyapplicable if the considered design is similar to these previous ships. Also, the cost estimationfactors in the approach reflect only past practices and experience.
Cost reductions resulting from newly adopted and developing shipbuilding technologiesand production methods can not be reflected in the existing historical based cost estimatingtechniques [CW92]. Advanced shipbuilding technologies typically involve a modular, productoriented approach which cuts across elements of the existing Ship Work Breakdown Struc-tures (SWBS - see section 2.2.5.3). Moreover, these weight based cost assessment approachesdo not reflect improvements that may occur in the production process [CF86, EDL+98]. Forinstance, if a new welding technique is used which takes 25% less man-hours per meter ofweld, no change would be reflected in cost, because there is no change in the weight of theship. Therefore, if a change in design or production process has no impact on weight, thenthe cost assessment will not change.
However this approach is often used for its ease of use at the very early design stages dueto the fact that it is easier to apply and obtain quicker "results", and does not require manydesign details. Weight is often used as the primary driving factor for cost assessment as itencapsulates the amount of material and to some extent the work associated with an item.Weight is an important characteristic to establish early in the design of any vessel and thereare several parametric rules such as [SB98a], which can be used to estimate weight based onsuch minimal information as the main dimensions and hull form coefficients.
One approach in a top-down cost calculation is to determine material and labour coststogether. In addition to those referred by [Car77], [SB98b] proposed an extensive study onthe development of such empirical-statistical relationships.
For instance, [Car77] gives the following cost equation which brings together the termsrelated to the main headings of steel, outfit and machinery:
Cs = A′·
W 2/3s L1/3
Cb
+ B′· Ws + C ′
· W 2/3o + D′
· W 0.95o + E ′
· P 0.82 (2.4)
E ′ = F ′ + G′ (2.5)
52
where Cs (euros) Total cost of the ship,A′ Factor embracing steelwork labour,B′ Factor embracing steelwork materials,C ′ Factor embracing outfit labour,D′ Factor embracing outfit materials,F ′ Factor embracing machinery labour,G′ Factor embracing machinery material,Ws (tonnes) Net steel weight,L (m) LBP Length,Cb (-) Block coefficient,Wo (tonnes) Outfit weight,P (bhp) Service power.
The constant factors A′, B′, C ′, D′ and G′ are calibrated by statistical analysis.
Another interesting cost assessment equation is given by [SB98a], derived from a diagramof [Ker85]. This equation gives the material and labour cost together (the unit cost per tonof steel installed), and multiplies these unit costs by the steel weight (see equation 2.6).
The man-hours number differs widely, depending on the production methods and shipcomplexity within the yard. [SB98a] gives the specific hull steel production costs as:
KSt = kSt · WSt (2.6)
kSt = k0 ·
(
4
(L/m)1/3+
3
L/m+ 0.2082
)
·
(
3
2.58 + C2b
− 0.07 ·0.65 − Cb
0.65
)
(2.7)
where KSt Hull steel cost (euro = monetary unit),kSt Specific cost for installed steel (euro/t),WSt Hull steel weight (t),Cb (-) Block coefficient,L Length of the ship (m),m Breadth of the ship (m),k0 Production costs of a ship 140m in length with Cb = 0.65,0.5 ≤ Cb ≤ 0.8 Limit for Cb,80m ≤ L ≤ 200m Limit for Length.
The [DT04] discussion on "Cost Estimating" is disappointing, as it does not really providean up to date survey and only presents top-down approaches. In their last section, theyrefer to many series of systems (tools) used for navy ships: ASSET, ACEIT, UPA, PRICE,and finally the PODAC (Product-oriented Design and Construction) cost model. PODAC[KF93, EDL+98, WKK+97], is also a rather sophisticated top-down approach. However,PODAC can be linked to other ship design tools with cost estimating capabilities thatoperate at a more detailed level of analysis:
• Parametric Flagship, a system developed under a Marintech Advanced ShipbuildingEnterprise project, links various ship design and naval architecture analysis systemsdirectly to the PODAC cost model.
• Intergraph’s multi-discipline GSCAD system was also linked with the PODAC costmodel
53
• The Navy’s ASSET design tool was linked to the PODAC cost model
Unfortunately all these tools are oriented towards combat ships and the real methodologyand details are not published.
This parametric model based on a detailed Product Work Breakdown Structure (PWBS– see section 2.2.5.2) provides the mechanism for entering the parameters available at thevarious design levels for specified ship types and their associated costs thanks to empiricalparameters (structural weight, power, etc.) or direct parameters (welding length, pipe length,etc.) if they are available. The parametric module is structured to use statistical analysisthat carefully consider factors like ship type, complexity, and basic ship characteristics suchas displacement, speed, individual weights, hull form, etc. So a new ship’s cost predictioncan be correlated empirically to those parameters. For instance at the concept level, theprice of the total ship is given by the equation 2.8.
P rice = CF · a · Displb· Speedc (2.8)
withCF = SF · TF (2.9)
where P rice (e) price of the total ship,CF complexity factor,a, b, c coefficients determined by historical database analysis,Displ displacement (m3),Speed design speed of the ship (knot),SF size factor of the ship (SF = 32.47 · Displ−0.3792),TF ship type factor16.
[RA05] propose a ship cost assessment method based on weight estimating at the earlydesign stage of the project. According to them, weight is the most important attribute onwhich initial design cost can be based. Weight can be estimated parametrically at the earlystage of design process, and is thus more immediately available than attributes such as cablelength and surface area. They implemented a computer aided approach to assess weight(equation 2.10) and cost (equation 2.11) to support the initial design process where thekey factors (k1 and k2) are calculated from historical data. Additional coefficients may beincluded in the cost equation if the associated information is known, like material purchasecost changes, currency inflation rate changes, production processes (robotic or manual),consideration for fatigue, etc.
W = k1 · Lpp · B · D · C0.5B (2.10)
CH = k2 · W (2.11)
where W is the lightship weight (tons),k1 is a coefficient dependent on ship size and ship type,Lpp is the length between perpendiculars (m),B is the beam (m),D is the depth to uppermost continuous deck (m),CB is the block coefficient,CH is the cost of the hull structure (euro),k2 is a coefficient dependent on ship type.
16for combatant frigate T F = 7, for passenger ship T F = 3, for a LNG T F = 1.12, for container shipT F = 0.96, for a tanker T F = 0.8, etc.
54
Despite their popularity and frequent references in the literature, top-down approacheshave serious disadvantages, which are often overlooked:
• A top-down approach uses only global information and is thus impossible to reflect localform changes or details of the design improving producibility. Top-down estimates donot allow the cost comparison of the features or details of a design which is necessaryfor selecting the lowest cost design approach at each step [KI90]. They don’t takeinto account changes in costs resulting from different labour productivities, differentproduction facilities, or different production procedures [KMS83].
• A top-down approach is usually based on weight. This aspect of macro parametersleads to an overemphasis of weight as a mean of cost control. Any change whichincreases weight will automatically increase the cost estimate regardless of the realeffect on cost. Extreme lightweight designs may drastically increase the number ofrequired hours, while large frame spacing may increase weight, but reduce the necessaryman-hours. This is often not reflected in the formulae!
• A top-down approach is based on historical data, i.e. historical designs and historicalproduction methods. In view of the sometimes revolutionary changes in productiontechnology over the last decade, the data and formulae may sometimes be called "pre-historic". Macro parameters tend to promote past practises and inefficient decision-making choices. They do not reflect new approaches in structural design or productiontechnology.
• A top-down approach is probably based on data which are inaccurate even at the timethey were derived. Shipyards are traditionally poor sources of cost information [KI90].The data are frequently skewed, reflecting the pressures of the first-line managers andother factors.
• A top-down approach is not suitable for structure optimisation as there is no linkbetween the cost and the design variables (e.g. scantlings).
A statistical cost assessment is only as good as the information supporting the assessment.For shipyards, historical cost information is not reliable for developing cost assessments fornew projects. However, historical data needs to be both accurate and collected in waysmeaningful to the assessing process. Thus, it is very important that the shipyards have acost planning and cost data collection system that is able to organise costs in order to havesufficient cost details and accuracy.
Clearly, top-down cost assessment is not helpful in improving productibility in ship de-sign.
2.3.4.2 Bottom-Up approaches
The traditional cost assessment top-down method using system-based, weight-driven costmodels are not always sensitive to changes in production processes and advanced manufac-turing techniques [EDL+98]. Thus the need exists for a cost model which can better relate todesign, construction product and process issues, to enable cost conscientious decision makingand more affordable ships.
55
The alternative method to calculating the product cost is called bottom-up approach. Thisengineering analysis cost assessment approach breaks the project down into smaller andsmaller intermediate products until the most basic product (e.g. plate) is described. Allcosts for machining, tracking, coating, assembling this product, along with its associatedintermediate products, into the next, more mature intermediate product are estimated. Theestimated cost of each layer intermediate product is summed up with all preceding layers,thus obtaining an aggregated cost which reflects an engineering analysis of the building pro-cess [GD96, Bar96]. In fact, the bottom-up approach breaks down the project into elements ofwork and builds up a cost estimate in a detailed engineering analysis. [Sou80, ML68, WB86],and [Rig01, Rig03b] developed simplified cost models based on direct calculation using quan-tities and unit cost to assess the global production cost. Tab. 2.5 shows some simple costfactors for different production stages. Taking the technology of plate welding as an example,there are many factors affecting the project time and working hours such as welding length,plate thickness, welding position, bevel type and welding accessibility.
The major advantage of this technique is that it specifically considers the actual workcontent of the product and provides a realistic cost estimate for the construction effort.
Name Main factors Secondary factors
Plate alignement area weight
Locating locating length component height, component weight
Tacking tacking length
Plate welding welding length, plate thickness, welding position bevel type, accessibility
Stiffener welding welding length, welding throat, welding position bevel type, accessibility
Block turning-over weight
Table 2.5: Main cost factors on different production stages [WjLlW+09]
The bottom-up approach requires more effort and detailed information than the top-downapproach, but unlike the top-down approach, the bottom-up approach captures differences indesign details and is thus suitable for scantling and shape optimisations [CRWV06, Bol06].Changing the local hull geometry influences the number of frames which require bending, theeffort in plate bending, and the degree of weld automation which depends on the curvature ofthe weld joints. All these effects are reflected by an appropriate break down of the total workprocess into its individual components. At present, such an approach is not yet available inmost shipyards; and neither are historical databases from which it could be developed. It isthus necessary to develop an appropriate approach, and collect the data required to use theapproach. An advanced optimisation application in this field is the work of ANAST for shipstructures using the LBR5 system [RMC05, TPCR07]. This is maybe the only system thathas been, to our knowledge, applied in shipyard work.
[Wol79] gives an example for welding man-hours in a shipyard panel line workshop (seeequation 2.12). The same approach could be used for all other shipbuilding processes. Untilsuch an approach will be fully developed for all processes, a less precise but similar approachcould be used by applying available data to the various design and production factors foreach design alternatives.
Wcost = 2.79 · NP S + 0.0215 · JLF B · tF B + 0.097 · JLCB · tCB + 0.017 · JLF · FCSA (2.12)
56
where Wcost is the welding man-hours,NP S is the number of panel starts,JLF B is the joint weld length of flat panel butts (m),tF B is the thickness of flat panels (m),JLCB is joint weld length of curved panel butts (m),tCB is the thickness of curved panels (m),JLF is the joint weld length for fillet welds (m),FCSA is the cross-sectional area for fillet welds (m2).
The Smart Product Model (SPM) of Proteus Engineering [RMH02, RH02, Ros04] wasdeveloped in such a way that during the design of a vessel it will provide the best costestimates for the available information. The system estimates ship production costs in aSWBS hierarchy for three independent levels of detail (see section 2.2.5.2):
• for concept design, the cost estimate will be based on whole-vessel technical and para-metric relationships derived from approximately 20 data items (e.g., length, beam,displacement, installed power).
• for preliminary design, system information becomes available, and the cost estimate isbased on about 125 data items.
• for the contract design, the cost estimate is based on hundreds or even thousands ofdata items. As the design progresses, more and more technical data and cost informa-tion moves from statistical to physics-based parameters, and the accuracy of the costestimate improves.
The cost estimating software is divided into two linked elements, one focused on engi-neering and the other focused on cost. Each element has modules for specific operations.The cost element has the four following modules:
• Parametric cost – Cost is estimated for the design ship based on proportionality withregard to the base ship.
• Assigned costs – Assigned costs are directly entered into the module. These costs arebased on data such as initial estimates from vendors and from purchase orders.
• Cost source selection – The user selects parametric or assigned values.
• Cost reports – This module produces three reports: 1-digit, 2-digit, and 3-digit SWBScost estimates, with overall confidence levels provided for each cost entry.
Mitsubishi Heavy Industries (MHI) developed the CAPP system interfaced to its structuraldesign CAD system, [SMT+01, SSI02, Sas03]. Production planning information estimationfunctions and levelling functions were added to a commercially available line simulation sys-tem. Application of the system to an actual vessel results in a reduction of 2 or 3 weekscompared to the time required usually. The process planning system with 3D visualizationfunction enables the designer to semi-automatically define the hull block assembly sequenceand to calculate the key parameters (cost, time, weight, etc.) for each process design can-didate (Fig. 2.24). The cost estimation function (equation 2.13) in the Mitsubishi CAPPsystem multiplies weld length (for each weld class) by a factor K reflecting the work difficulty(e.g. 1 for downward, 2 for upward, 1.5 for horizontal)
57
CP roduction =∑
(WConversion · CUnitconst) (2.13)
With
WConversion = Wreal · K (2.14)
where CP roduction is the production cost of one block (euro),WConversion converted welding length 17,CUnitconst the cost to weld one meter for each posture (euro/m),Wreal the actual welding length (m),K a coefficient to express the difficulty of the welding work.
Figure 2.24: Application of bottom-up approach comparing a shell-base production (left)and a deck-base production approach (right). Analysis gave a 3% cost advantage for thedeck-base approach.
2.3.4.3 Life cycle approaches
In order to improve the design of products and reduce design changes, cost, and time tomarket, life cycle engineering has emerged as an effective approach to address these issues intoday’s competitive global market. As over 70% of the total life cycle costs of a product iscommitted at the early design stage [Eyr01], designers can substantially reduce the Life CycleCost (LCC) of products by considering the life cycle implications of their design decisions.
People are always concerned about product cost, which encompasses the entire productlife from conception to disposal. Manufacturers usually only consider how to reduce the costof materials acquisition, production, and logistics. In order to survive in the competitivemarket environment, manufacturers now have to consider reducing the cost of the entire lifecycle of a product, called the LCC, [SPJW02]. In case of a ship, it consists of the fabricationphase in the shipyard, such as design and assembly, and a maintenance phase in service,such as inspection, repair and painting, as well as disposal costs.
[LBD07, GZ07, TlL+09] have recently implemented methods for the investigation ofeconomic and environmental costs within a marine system. The LCC assessment approachis a promising future holistic methodology in order to maintain the effectiveness of shipsduring their overall life. But all the authors are unanimous concerning the difficulties dueto the variety of levels of production and maintenance.
17Considering the different difficulties of welding depending on posture.
58
2.3.5 Challenges of cost assessment
The organisation of a production control and cost control system in a shipyard is not aneasy task for different reasons [Fer44].
2.3.5.1 Disconnection between decision and cost
Cost assessments are typically not available at the point when a decision needs to be made.Any cost estimate (be it high or low quality) is typically not available until after the part issourced or in production. However, this not very helpful when it is a well-accepted fact that70-90% of product costs are decided in the first 20% of the product development cycle (seesection 2.4).
2.3.5.2 Inaccuracy of the cost assessment
Cost assessments (especially early ones) often lack sufficient rigour and thoroughness tobe used reliably in decision-making. Because historical costs lag behind the point in timefor decisions to be made, the shipyard is forced to make rough estimates from as much ofthe fragmented information as it can gather.
2.3.5.3 Cost evaluated only once
Once a cost assessment has been created, it typically doesn’t change or get updated as newinformation becomes available. If an employee is lucky enough to be provided with a costestimate early in the development process, this estimate is likely not included in ERP of theshipyard. Therefore, if new or better information becomes available to help the user refinethe cost assessment and increase the quality of the cost number, it is an arduous process totake advantage of the new information and recalculate the cost assessment.
2.3.5.4 Multiple versions of the cost
Multiple cost assessments exist which have been created from different sources (i.e. dif-ferent quantity, at different times, etc.). Because most cost assessments are generated byitinerant spreadsheets or from systems used only by cost experts and are off-line from thepeople that actually make the design, different costs start appearing. Sometimes this isbecause not everyone has received the latest quantity from the designer or they have anold version of the spreadsheet. At other times this occurs because different people in theshipyard have separate methods for calculating the same cost.
2.3.5.5 Measure of real costs
To ensure the effectiveness of any cost model, a basic requirement is that the measurementof the real cost has the same breakdown structure as the model. Within cost assessmentthe estimator is bound to use some "standard" of comparison, not a fixed or immutablestandard, but nevertheless a standard, whether such a standard is the written record ofactual previous costs of identical or similar products. Good estimates are made by accuratecomparison with the previous costs, allowing for any special changes in equipment or method,and hence a good estimate becomes a standard with which to compare the actual cost whenthe production is completed, and thus shows any increase or reduction in the efficiency ofthe shipyards in respect to the particular product.
59
2.3.5.6 Uncoupling between design and cost engineering
Cost is often a secondary consideration for the engineers concentrating on delivering thetechnical aspects of a new design [Bol07]. Indeed, cost evaluation can only be performed oncethe technical details have been resolved. Moreover, after that, it is possible to review thecomposition of the design. This two stage process results in a degree of separation betweentechnical and cost engineering departments working on the project and creates the situationwhere there may be a need for further design iterations. While these two engineering groups(design and cost engineers) operate separately there may be little opportunity to go throughan optimisation process to improve cost.
It can be noted that there is a big gap between the existing and the past possibilities of thecost optimising and controlling the achievement of the allowable cost. These circumstancesconcern not only reducing cost but also the aim of gaining time to create and comparedifferent design alternatives.
2.3.5.7 Cost assessment in early design stage - a real challenge
New production technologies and organizational improvements in shipyards have resultedin a continuous reduction in man-hours/ton over time especially inside the dry dock (see Fig. 2.1).In general, the nature of the shipping business leads to situations that place an excessivedifficulty on effective cost estimation even for the most experienced ship cost estimator.
In most cases ship construction contracts are signed before the completion of a detaileddesign. The reason for this is that detailed designs with a detailed cost assessment arevery expensive and excessively time consuming [Bar96]. Shipyard work in terms of workspecifications is difficult to formalize and predict directly from intricate detailed ship designs.This induces a very large risk to both the buyer who might end up overpaying for a shipand the seller who might have to incur exorbitant costs due to the lack of a clear definitionof the work.
The overwhelming majority of overhead costs are time related. This leads to an under-standing that by reducing construction time there are significant benefits, even if labourman-hours don’t decrease.
2.3.5.8 Specificities of the shipbuilding industry
Cost assessment and production simulation allows management to predict the effectivenessof processes in the shipyard. These methods are most frequently used in industries involvedin mass production. This is not the case in shipbuilding (see section 2.2.2), which can becharacterized by:
• small series production,
• short time to market,
• many different work disciplines,
• large number of different work tasks,
• high complexity,
• high degree of manual work,
60
• difficult working conditions,
• very difficult to identify and quantify work activities.
Thus, with production processes more complicated, and production parameters moredifficult to quantify, production simulation and cost assessment are not as used in the ship-building industry as in certain other industries such as automobile industry.
2.3.5.9 Intricate control of schedules and costs
The importance of planning and control to ship construction is generally, if not universally,accepted. Management of work is synonymous with management of costs [Fer44, BR91].The preparing of detailed schedules for every piece of material that enters into the ship isrequired, so as finally to ensure that the various groups or sub-assemblies of materials willarrive at the building sites or at the outfitting quay in the proper way and without anymissing pieces. The first thing to remember is that the objective is to gain and keep controlof the project. That is, the plan must be produced enough to be used as an actual drivingforce and the control system must give enough information to permit corrective action whennecessary. Hierarchical planning is essential. This typically gives three levels of planning,which correspond to different time horizons and levels of detail. Typically these will be:
• Strategic – covering all project or development with a time horizon of years
• Tactical – covering a project or development with a time horizon of months
• Operational – covering work stations with a time horizon of weeks
2.3.5.10 Cost variation factors
Worldwide competition and an increase of the costs of materials place massive pressureon shipyards to release a cost linked to design. However, capturing construction cost duringthe design of a new ship is one of the most difficult parts of the design process [Bol07, Mir06,FH08]. Construction costs must be tracked during the design process to ensure that theproject remains viable to both yard and customer particularly as late changes introducedinto the design can have considerable cost impacts. The factors that cost depends on arealways changing and only once the production design is finalised is it possible to make adirect evaluation.
The main factors that could result in cost changes are:
1. Technology change:
• New production processes
• New materials
• New designs
2. Social, economic and political situation:
• Changing workforce (productivity)
• Economic downturn and unrest
3. Shipyard backwardness:
61
• Intense backwardness causes confusion
• Few orders results in loss of skill
4. Labour rates:
• Different for each shipyard
• Effect of learning (see section 4.3.2.3)
• Unpredictable changes
5. Material cost:
• Vendor base changes
• High fluctuation of steel rate
6. Regulation:
• New rules
7. Inflation:
• Fluctuates unpredictably
• Different rates for each item
Cost assessment in different production assemblies is complicated by the fact that:
• there is insufficient cost data and the quality of this information is often quite low inthese early phases
• the data is usually distributed on different ERP and CAM systems which are compli-cated to handle
• sometimes the required cost information exists only in printed tables, or even in theknowledge of a single expert
• the production process are changing continuously in a shipyard so that the historicalcost database cannot be used for a long time
• different types of ships induce different types of cost (see Fig. 2.25) and it is oftenimpossible to compare their relative cost data
These circumstances produce different difficulties:
• Estimating and planning the ship costs takes a lot of time for manual system queriesand following cost aggregations. The pressure on time makes the assessment of exactand robust cost information difficult.
• The possibility of cost estimating belongs to a few experts. Besides this fact, thequality of estimations, based on the knowledge of experts, differs widely (variation onaverage ± 30%)
• The existing Data Base (DB), generated from past projects, is far from being com-plete, due to the lack of an integrated system for managing and providing the costinformation.
62
2.3.5.11 Increasing Costs
For the last two years, all shipbuilding costs have continuously increased all over theworld. These costs include the cost of building materials like steel (see Fig. 2.27) and non-ferrous metals (copper, nickel and aluminium), the cost of marine equipment and suppliesand, last but not least, labour costs (see Fig. 2.26), [CES07a]. In principle, most marineequipment markets can be considered as global with similar conditions for shipyards aroundthe world. Nevertheless, yards are also in some areas confronted with specific local marketconditions. While generally benefiting from a tight network of highly specialised qualityproducers in Europe, the example of steel plates for shipbuilding is quite illustrative of amajor competitive disadvantage.
The price differential between Europe and Asia for shipbuilding steel plates is quite sig-nificant and has recently passed and been maintained at over 450$ per ton (see Fig. 2.27),indicating that European shipyards have to pay a significant additional cost compared totheir Asian counterparts. For steel intensive ships, steel can represent up to 25% of the totalcost of the vessel. In 2009, with the current price gaps, it has become nearly impossible forEuropean yards to offer competitive prices in such market sectors. In addition, the acceler-ating cost inflation impacts greatly on the profitability of ships which are usually contractedat a fixed price with delivery several years after contract signature. This close dependenceon the changing market complicates the creation of cost assessment models.
2.3.5.12 Data and DB management problems
The quality of cost assessment which requires accurate, reliable, repeatable and under-standable results depends mainly on the quality of the input data. Moreover, the imple-mentation of cost assessment models necessarily involves the manipulation of large amountsof data from both the ship and the manufacturing environment. It follows several types ofproblems, often very cumbersome, tedious and time consuming to solve. Here are presentedthe major problems that we may encounter:
(a) Bulk Carrier [Asi08]
Figure 2.25: Cost Breakdown structure of ships
63
Figure 2.26: Relative Cost of Shipbuilding Labour – Average hourly labour costs in currentUSD [oLSB04]
450
550
650
750
850
950
1050
1150
1250
1350
1450
Ste
el p
rice
$/t
on
ne
Jan
-06
Fev
-06
Ma
r-0
6
May
-06
Jun
-06
Jul-
06
Au
g-0
6
Se
p-0
6
Oct
-06
No
v-0
6
De
c-0
6
Jan
-07
Fev
-07
Ma
r-0
7
Ap
r-0
7
May
-07
Jun
-07
Jul-
07
Au
g-0
7
Se
p-0
7
Oct
-07
No
v-0
7
De
c-0
7
Jan
-08
Fev
-08
Ma
r-0
8
Ap
r-0
8
May
-08
Jun
-08
Ap
r-0
6
World
Europe
Asia
Figure 2.27: Price comparison for hot rolled plate [MEP08]
64
• Lack of available data – Initial weight and cost estimates usually make use ofpreviously designed and constructed ships of similar type and size. Unfortunately,access to relevant data may be hindered by proprietary restrictions, interdepartmentalcommunication gaps, etc. In certain cases, when the shipyard is entering a new market,no relevant data is available.
• Insufficient data definition – Fields of Data Bases (DB) are rarely defined preciselyenough to permit direct comparison of data from different studies or reformulations tosuit overlapping or otherwise differing definitional boundaries [Koe02].
• Inconvenient data format – The data may be in a hard copy and not in an electronicDB, necessitating time-consuming keypunching. The data hierarchy may be unfamiliar,and require resorting a hierarchy appropriate to the present project. The constantevolution of the production tool implies a constant evolution of the data structure sothat it is very difficult to compare past data with actual projects.
• Unknown validity of data – The data may itself be an estimate and not a record ofthe actual costs. Costs may have changed because a supplier has gone out of businessor has moved out of country. The year in which the data was generated may not beknown, so that the impact of inflation is in question. The data may been provided bya consultant, and its validity unknown.
• Inaccessibility of data – Object oriented databases are often used inside CAD/CAMtools such as TRIBON, CATIA, etc. in order to store the ship models. For designpurposes it is the most appropriate solution because of the hierarchical structure of theproduct (eg. Scheme→Plates→Holes). Nevertheless, for general-purpose queries on thesame information, pointer-based techniques will tend to be slower and more difficultto formulate than relational databases. In fact there is an intrinsic tension betweenthe object encapsulation, which hides data and makes it available only through theinterface methods of the CAD/CAM tool. So that, the extraction of the data from theCAD/CAM tools to a traditional relational database is required for cost assessmentpurposes. It is a time consuming task mainly because the extraction of the data inducesthe implementation of macro using the specific export modules of the design tools.
• Quality of the data – It is clear that the effectiveness of any cost estimation ulti-mately comes down to the quality of the data used in the calculation. While moreadvanced cost models are being developed, in our experience, the collection of data todrive these models often seems poor. Furthermore, the more advanced models requirericher data sets.
• High quantity of data – The design and production description of a ship containsan amazing quantity of information. For example, it is a question of several million ofrows with several hundred attributes for the steel structure description of a passengervessel of 300 meters in length (eg. about 200 000 steel items).
• Data integrity – During the management of data it is very important to have theassurance that data are consistent, correct and complete. It may happen that this isnot the case so it is necessary to report these data as outliers.
• Data temporal heterogeneity – The continual changes in technological processesand business practices invalidate the data of previous ships.
Consequently, without disciplined capture of cost data and full support for this activityfrom management it is very difficult for any cost assessment process to function adequately.
65
2.4 Concurrent engineering (CE)
2.4.1 Introduction
Today the design method used in shipbuilding plays a primary function at the first stagesof the project. According to a traditional approach, during these phases, the majority of thedecisions are taken based on experiment and the opinion of the designers. However, thesedecisions have a strong influence on the ship and also on its entire life cycle, production,maintenance, etc.
One particularly effective technique, which facilitates the integration of design and pro-duction, is Concurrent Engineering [SSH+00]. Concurrent Engineering is not new. It hasbeen used by many highly successful companies in many industries to vastly improve per-formance. But industrials typically spend too much, too late for the product or componentto be effective in the market place.
CE is a subject of considerable significance as illustrated by the annual CE conferences(ISPE) held since 1994.
2.4.2 Definition
Concurrent Engineering is a systematic approach to the integrated, concurrent design ofproducts and their related processes, including manufacture and support. This approachis intended to cause the developers, from the outset, to consider all the elements of theproduct life cycle from conception through disposal, including quality, cost, schedule anduser requirements.
It is been said [Bar96] that design involves seeking the right problem to solve, whileengineering involves the "right problem". Concurrent Engineering seeks to perform bothdesign and engineering simultaneously.
The major concept of CE is to put the majority of effort in the product design stageto analyse the factors which might affect subsequent production processes, and hence saveoverall product development time; that meet the given function and quality requirements inthe shortest time and lowest overall cost. The potential CE benefits are reported in Tab. 2.6.
2.4.3 Principles
2.4.3.1 First principle
Design is the primary driver of quality, time and cost. Design drives 70% or
more of the typical cost and product success in the commercial market place.
Product design heavily influences cost, quality, and time-to-market and thus a company’sprofits. It is widely recognized that the product design stage influences nearly 70% of the finalproduct costs (see Fig. 2.28), even though only a small amount of expenditure is incurred atthis stage [BDK02, Hun93, FH08].
66
Category Benefit Type of benefit
Development time 30-70% Reduction
Engineering Changes 65-90% Reduction
Time to Market 20-90% Reduction
Overall Quality 200-600% Improvement
Productivity 20-110% Improvement
Table 2.6: Potential Concurrent Engineering benefits [Bar96]
100%
0%
Costs
Time
70%
Co
ntr
act
AccumulatedCosts
CommittmentCosts
Co
nce
pt
Pro
du
ctio
nD
esig
n
De
taile
dD
esig
n
Pro
du
ctio
n
La
un
ch
ing
Se
a T
ria
ls
De
live
ry
Figure 2.28: Degree of commitment costs to ship design
67
2.4.3.2 Second principle
While the cost of implementing changes increases tenfold with the passing of
each stage of production, design flexibility decreases over the design cycle
inversely with cost
In order to compensate cost increases or quality decreases due to design flexibility loss(see Fig. 2.29), the shipbuilding industry tries to apply the concurrent engineering conceptrather than sequential engineering (see Fig. 2.30). The decisions of each stage are madeby considering the constraints imposed by the other stages of the ship life cycle. Now theproblems that were only checked at the end of the project are now included in the design stageto reach a better solution. Each department no longer waits any more until the previous onehas finished but has to consider that a decision can occur over the course of project [Boc98].
As illustrated on Fig. 2.29, product information grows when the design freedom flexibilitydecreases. Moreover, one of the effects of Concurrent Engineering is to move the informationcurve upstream because the effectiveness and the quality of the information on the ship areimproved from the first stage of the project. The consequence is that we have a reductionof production delay and an increase in design freedom flexibility. This aspect is particularlystrategic as the design process has a cost which varies from 5% to 15% of the total cost andmoreover decisions taken during this initial stage determine about 60% to 95% of the totalcost [SM94].
2.4.3.3 Third principle
The multi-functional team is the key to the effective total design equation
An optimised Concurrent Engineering environment provides an opportunity to substan-tially reduce the total cost of a project. This is because integrated product teams containingmembers of various skilled disciplines enable a simultaneous contribution to an early productdevelopment and definition. It is impossible to achieve low production cycle times, shortdelivery times and high productivity unless the engineering and production functions workclosely together in multi-disciplinary teams. Therefore, within a fully integrated productdevelopment cycle, multidisciplinary teams working together increase the likelihood of a re-duced life cycle cost by avoiding costly alterations later in the design process. With thisview in mind, Concurrent Engineering is a great step forward when compared to an "overthe wall mentality", where each department works in "isolation" [RK03].
2.4.3.4 Fourth principle
The more information is known earlier, the better the decisions made in the
design process
Everyone understands that as the design process advances, more information becomesknown, but at the same time it becomes more and more costly to make design changes. Withbetter technical and cost information in hand, better technical decisions could be made. Inaddition, management could be invited to play a more significant role than is traditionallythe case. These ideas are reflected Fig. 2.29, which depicts the decreasing ability to influencethe outcome of a design, and the traditional reluctance of management to participate until
68
6
-0% 100%Time of the project
Time reduction
Project
Increasing of
Model data
Information
BasicDesign
DetailedDesign
ProductionDesign
> 18 months 8-12 months 1-2 months
Strategical
Decision
Tactical
Decision
Operational
Decision
Design
Flexibility
Figure 2.29: Design stages within shipbuilding industry
Needs
definition
Product
Design
Checking
Prototype
Test
Revised
Modifications
Checking
Production
Needs
definition
Product
Design
Checking
Production
Productibility
Service
Cost
Performance
Quality
Project
Seq
uen
tialen
gin
eeri
ng
Co
ncu
rren
ten
gin
eeri
ng
Figure 2.30: Concurrent engineering
69
late in the design process, until considerable funds has already been spent [RH02]. The goalis to focus the design up-front.
2.4.3.5 Fifth principle
Great innovation is only as good as great implementation. The success of
Concurrent Engineering is predicated on implementation of the entire design
process. The earlier the innovations are made the most benefit can be
gathered.
This phenomenon is particularly applicable with the introduction of advanced technologiesin ship design. The greatest savings result from the introduction of innovations at the earlystages of development. Incorporating new technologies at later stages when the design isalready committed to a certain technology will result in increased labour costs and changeorders [Mir06].
2.4.4 Drawbacks and limitations
There are some drawbacks associated with the initial implementation of concurrent engi-neering, including the need for considerable organizational restructuring and extensive re-training of workers. Such potentially disruptive changes and added work requirements maybe met with resistance from managers and other employees. Also, there are usually consid-erable difficulties in transferring data among employees in different departments, which mayrequire the additional tracking software applications.
Moreover, the division of the expertise into parallel tasks can lead to issues which aredifficult to manage. For instance a task B, started on the basis of the preliminary data of atask A currently in progress, should ultimately be completely or partially restarted due toinitial misconceptions of the task A.
In addition to these significant up-front investments, organizations pursuing a concurrentengineering work model must typically wait several years before seeing the benefits of thistransition.
2.5 Conclusion
The identification and the evaluation of the pre-existing contributions to the researchproblem defined in chapter 1 has been carried out. A complete and comprehensive reviewof the published research results and current industrial practices has been presented in thischapter. The inefficiencies of traditional ship design (see section 2.2.7) and the challenges ofcost assessment in the shipbuilding industry (see section 2.3.5) have been highlighted, leadingto the conclusion that the research question has not been previously answered efficiently.
70
Chapter 3
Methodology
3.1 Introduction
This chapter describes the major methodologies used to answer to the main challenges ofthe research.
The key paradigm of the PhD thesis is presented in section 3.2. Subsequently various "de-sign for X" (DFX) approaches have been collected in section 3.3 and offered in a harmonizedway. Finally, a cost assessment method selection is developed in section 3.4.
3.2 Paradigm
Sustainability of technologies has been the central focus of many international debates,seminars and forums [Dun06]. Designing for sustainability (see Fig. 3.1) requires the consid-eration of social, economical and environmental factors throughout the product life. The LifeCycle Performance (LCP) as a measure of sustainability and competitivity covers a numberof key aspects, such as Life Cycle Cost (LCC), environmental friendliness, end-of-life impactsor safety. This PhD study focuses on the economical aspects.
Figure 3.1: Scheme of sustainable development [UCN06]
In the early stages of design and development all technical and ecological requirementshave to be considered in terms of their long-term impacts on the entire ship life cycle. An
71
engineering design should not only transform a need into a description of a product butshould ensure the design compatibility with related physical and functional requirements.Therefore it should take into account the life of the product as measured by its performance,effectiveness, producibility, reliability, maintainability, supportability, quality, recyclability,and cost.
"Design for X" (DFX) is commonly regarded as a systematic and proactive designing ofproducts to optimise total benefits over the whole product life span (fabrication, assem-bly, test, procurement, shipment, delivery, operation, service, disposal), and meet targetquality, cost, performance and time-to-market. DFX involves different methodologies forproduct design and optimisation (like Design for: Manufacturing, Assembly, Maintenance,Simplicity, Environment, Safety, Robustness, etc.) which provide useful results; however,they address only specific aspects of product life cycle. However, considering that the shipdesign problem is a multi-objective optimisation problem, it is actually always a "Design forX, Y, Z, ..." problem. The "Design for X" procedure will be holistic and lead to ship designssimultaneously optimizing a variety of objectives.
Various "design for X" approaches have been brought together in section 3.3 and presentedin a harmonized way via DFX methodologies.
Nowadays we are lacking an integrated approach and a unified measure of LCP in aholistic way. While the current market position of European shipyards in complex one-of-a-kind products proves their capability to cope with specific customer requirements at shortdelivery times and at competitive prices (see section 2.2.3), there is no doubt that a widerapplication of LCP assessment has an immense potential. Since different approaches usedifferent measures for concept design evaluation (e.g. Design for Quality minimizes reworkdue to poor quality, while Design for Assembly cuts assembly time) it is not clear how thosediverse results can be judged and compared. In this context, the need for holistic and unifiedviews on design concepts evaluation is evident.
Due to the lack of methods and data for a holistic life cycle assessment, the current practicein shipyard is the following:
• No tools are available which would suit the needs of the shipyards both in concept de-sign and system design and hence support a continuous application of LCC assessmentthroughout the different design stages (see section 2.3.5)
• Life cycle "thinking" is not yet fully implemented in the culture of the shipyards. Theapplication is mainly limited to sales and basic design personnel – having the closestcontact with the customer, but it focuses on limited aspects and is not carried onto a technical levels. Life cycle optimisation – in a sense selecting the right designoptions on ship and system levels – is poorly applied. Methods and tools are needed,which connect technical design parameters to life cycle performance, allowing technicalexperts to quickly assess the impact of design options and parameters on the overallship performance.
The main idea behind this PhD thesis is that the LCP can be measured by the LCC ofthe product. A good assessment of LCC during all the stages of the design will lead to theimprovement of the sustainability and competitivity of the product (see Fig. 3.3). Designersare in a position to substantially reduce the LCC of the product by giving due considerationto the LCC implications of the design decision they make. While the LCC is the aggregate
72
of all the costs incurred in the product’s life, it must be pointed out that the developedapproach focuses on the cost that can be influenced by designer [DE07].
Building cost means total cost spent in a shipyard for building a ship. This is a part ofLCC and occupies a large amount. It is considered that acquisition cost occupies about 2/3
of life cycle cost and maintenance cost occupies 1/3. Fig. 3.2 shows the break down of theacquisition and maintenance cost of a VLCC1 ship.
Material - Other
30.0%
Material - Steel
20.0%
Material - Engine
10.0%
Fabrication-Steel
Structure 15.0%
Fabrication-Other
15.0%
Design-Basic 5.0%
Other 5.0%
Acquisition cost
67%
Maintenance cost
33% Lubricants 14%
Stores 12%
Maintenance and
Repairs 50%
DryDock 24%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Acquisition cost Total cost Maintenance cost
Figure 3.2: Acquisition and maintenance cost of a VLCC ship [KS01, JR05, GZ07]
The key issue of this research is to provide the means for the designer to reliably estimateand verify the costs of different design concepts at different stages of product development.Section 3.4 provides a structured cost assessment method specifying how and when thedifferent methodologies can be applied.
It is widely understood that design decisions have a great impact on the LCP of the system[Mat83]. It is fundamental that ship designers know how to evaluate their design for X and,thus, be able to attach a cost, directly or indirectly, to any part of the design that they aredeveloping. The main expected result is that designers will not only think about how tosolve their specialized problems, but will rather think about the ship as a whole. Indeed,the complexity of the design of a ship requires cross-disciplinary knowledge and masteringdifferent processes and technologies. The skilled engineer has to decide the solution to bechosen and developed with a balance of required performance, alternative costs and eventualalternative production processes, and finally different build strategies. Therefore differenttime schedules and costs are involved.
With the aid of computers it is possible to make a study of a large number of varyingdesign parameters and to arrive at a ship design which is not only technically feasible but,more importantly, is the most economically efficient (see chapter 4).
1VLCC – Very Large Container Carrier
73
Figure 3.3: Design for X and Life Cycle Cost
74
3.3 Design for X (DFX) – Concurrent Engineering (CE)tools
3.3.1 Introduction
Ship design was in the past more of an art than science, highly dependent on experiencednaval architects, with good backgrounds in various fundamental and specialized scientificand engineering subjects, alongside with practical experience. The design space (multitudeof solutions for the design problem) was practically explored using heuristic methods, namelymethods deriving from a process of trial and error often over the course of decades. Gradually,trial and error methods were more and more replaced by gained knowledge [PAK+09].
Today ship design can be viewed as an ad hoc process. It must be considered in thecontext of integration with other design development activities, such as production, costing,quality control, etc. In that context, it is possible for the designer to work on a difficultproduct, requiring high material or labour cost, and containing some design flaws that theproduction engineers have to correct or send back a new design before production. Anyadjustment required after the design stage will result in a high penalty of extra time andcost [OAT04]. Deficiencies in the design of a ship will influence the succeeding stages ofproduction. In addition to designing a ship that fulfils producibility requirements, it is alsodesirable to design a ship that satisfies risk, performance, cost, and customer requirementscriteria. More recently, environmental concerns, safety, passenger comfort, and life-cycleissues are becoming essential parts of the current shipbuilding industry.
With this paradigm, the selected design will be a producible, cost-effective, safe, clean,and functionally efficient design. This will enable shipyards to obtain great rewards, suchas the reduction of construction time and costs, reduction of lead time, improving productquality, simplification of products, and gaining sustainable competitive advantages in theshipbuilding market.
Throughout the engineering disciplines, many "Design for X" process have been devel-oped in order to correct the inadequacies of the designs during the ship design stages. DFXmethodologies were developed to support the designer by generating feedback on the conse-quences of design decisions on the product. In other words, DFX is the process of pro-activelydesigning products to optimise all the functions throughout the life of the product: fabrica-tion, assembly, test, procurement, shipping, delivery, service, and repair, and assuring thebest cost, quality, reliability, regulatory compliance, safety, time-to-market, and customersatisfaction. This has been called "Design for X" where X is whatever the specific focushappens to be. So "Design for X paradigm" covers many areas such as:
• Design for Production (DFP) or Design for Manufacturing (DFM) – see section 3.3.2
• Design for Assembly (DFA) or Assembly-Oriented Design (AOD) – see section 3.3.3
• Design to Cost (DTC) – see section 3.3.4
• Design for Simplicity (DFS) – see section 3.3.5
• Design for Maintenance – see section 3.3.6
• Design for Environment (DFE) – see section 3.3.7
75
• Design for Safety (DFS) or Risk Based Design (RBD) – see section 3.3.8
• Design for Life Cycle Cost (LCC) – see section 3.3.9
• Design for Robustness (DFR) – see section 3.3.10
• Design for Six Sigma (DFSS) – see section 3.3.11.1
Other "design for X" concepts are also described in the literature, such as design forreliability, design for quality or design for lean manufacturing but are not described in thissection.
3.3.2 Design for production (DFP) – Design for manufacturing(DFM)
For most ships, productibility has become a major design attribute. If a ship cannot bemanufactured or assembled efficiently, it is not properly designed. Any adjustment requiredafter the design stage will result in a penalty of extra time or cost. Deficiencies in the designof a ship will influence the succeeding stages of production [OYL97, OAT04].
The overall objective of design for production DFP can be defined as "Design to reduceproduction costs to a minimum, compatible with the requirements of the vessel to fulfil itsoperational functions with acceptable safety, reliability and efficiency" [SSH+00, BHH+06].DFP optimises all the manufacturing functions (fabrication, assembly, test, procurement,delivery, service, repair, etc.) that reduce the production work content while still meetingthe specified design requirements and quality. The goal is to include the impact of designdecisions on the production process.
Time pressures on commercial ship contracts result in the overlapping of phases of designdevelopment, procurement and production (see Concurrent Engineering section 2.4). Thismakes the impact of engineering changes more difficult to manage. There is a need to sys-tematically study the detail design process and its impact on construction with the objectiveto improve the process and its integration with construction [MD05].
The extension of the design process to include the DFP activity has the following primaryobjectives:
• to produce a design which represents an acceptable compromise between the demandsof performance and production and, where appropriate, takes into account the needsof overhaul, repair and maintenance,
• to ensure that all design features are compatible with the known characteristics of theshipyard facilities,
• to apply the individual DFP principles and procedures insofar as they are relevant tothe particular vessel and to the particular shipyard where the vessel is to be built, and
• to coordinate the interrelationship between the machinery, electrical and outfittingwork with the structural work, in order to create a fully integrated design model.
76
DFP must be incorporated into the design from the start. Traditional engineering leavesit up to another department, such as Production or Manufacturing Engineering to developthe technical documentation required by the production workers. This is an unnecessaryduplication of effort and is a non-value added task that takes time. The basic goal of DFPis to reduce work content and design detail decisions should be based on this.
The producibility evaluation process requires a close cooperative relationship betweenship designers and shipyard production personnel as early as possible in the design cycle. Ingeneral, ship design is characterized by the fact that some of the most important decisionsregarding the ship are taken at the early stages of the process. Later design actions areinevitably restricted by the earlier decisions, allowing for little possibility to positively affectcost and performance. As the design process progresses, the knowledge of the design objectincreases, while the freedom to make changes decreases due to the large costs associated withthese changes. There is evidently a need for knowledge feedback in the early stages of thedesign process, where the cost impact is lower and the freedom to make changes is greater.The improvement of producibility leads to a delivery of cost reduction targets establishedunder the DFP concept.
DFP is the deliberate act of designing a product to meet its specified technical andoperational requirements and quality so that the production costs will be minimal throughlow work content and ease of construction. All designs should be prepared to suit theshipyard facilities and preferred methods [LAA06].
The obvious objective of DFP is to increase the productivity (see section 2.2.6) of theshipbuilder. Reducing the build duration is a secondary objective. There are two mainprinciples for DFP for ships, namely:
1. all designs should drive for simplicity, and
2. all designs should be the most suitable given the shipyard facilities.
DFP can significantly reduce the costs, since ships can be quickly assembled from fewerparts. Thus, ships are easier to build and assemble, in less time, with better quality. Partsare designed for ease of fabrication. DFP encourages standardization of parts and modulardesign. Designers will save time and money by the reduction of the production complexity(see section 4.4).
Ease of manufacturing Designing for an easy construction of parts, material processing,and product assembly is a primary design consideration. An example is available in Fig.3.4, where two designs of a man hole with and without DFP are presented. Particularlyif labour costs are a big percentage of the cost, problems in fabrication, processing, andassembly can generate enormous costs, cause production delays, and demand the time ofprecious resources.
Some basic rules can simplify manufacturing in shipbuilding industry [Mir06]:
• Avoid using thin plate to avoid distortions, reworking and straigthening
• Do not plan hull curvature into the structure (hull plating)
• Eliminate cruiser sterns and cambered transoms
• Maximize use of flat panels, straight frames, and reduce plate curvature
• Simplify bow and stern shape by removing unnecessary curvature
77
• Run strakes2 in the same direction as primary framing
• Design for maximum use of high productivity tools such as automatic welding
• Design bilge strakes with the same thickness as bottom plates
• Make port side and starboard unit similar (symmetry)
• Allow for large deck space to facilitate outfitting
• Etc.
(a) Without DFP (b) With DFP
Figure 3.4: A manhole with and without DFP
Standardisation Products do not normally have all-unique components. Periodicity,seen as a matter of information theory, can be regarded as a way of economising informationin the assembly process. Considering periodicity in assembly makes it possible to distinguishbetween the amount of instructions needed for one component and the amount of instructionsneeded for recurrent components. For a component used once in an assembly, instructions arecreated and used once, whereas for recurrent components, assembly instructions are producedand recycled, with an obvious benefit [RJS04]. A logical consequence of highlighting theperiodicity of components in assemblies is that it represents a lower product complexity (seesection 4.4). Standardisation, as a means of reducing complexity and component variants,actually boosts the manufacturability of the product itself. It also increases the chances ofautomated assembly, for it presents a repeated mode of assembly.
The following rules can simplify the manufacturing in shipbuilding industry:
• Maximum use of standard plate and stiffener sizes
• Design to facilitate assembly and erection with structural units, machinery units, andpiping units
Modular design Modular design or "modularity in design" is an approach that subdividesa ship into smaller parts (modules) that can be independently created. Besides reduction incost (due to lesser customization, and less learning time), and flexibility in design, modularityoffers other benefits such as the reduction of lead time during production (see section 2.2.5.1).
2A continuous band of hull planking or plates on a ship
78
3.3.3 Design for assembly (DFA)
DFA methodologies were developed to aid the designer by generating feedback on theconsequences of design decisions on product assembly. The aim of DFA is to simplify theproduct so that the cost of assembly is reduced. However, the consequences of applying DFAusually include improved quality and reliability, and a reduction in production equipmentand part inventory. The application of DFA guides the designer towards a product with anoptimum number of parts, that requires simple, cost-effective assembly operations and themost appropriate manufacturing processes and materials for its components.
The most important task for the designers is to balance the relationship between costinformation and design decisions. Therefore it is important to develop a method that givesthe designers quick and accurate estimates of the financial consequences of their designdecisions, and procedures to determine optimal design parameters [SPJW02]. The presentstudy focuses on the development of methods and tools to guide the decision of the designerregarding the potential design alternatives. The assembly operation of a product is one ofthe most important cost drivers during production (see 3.2). We will take into account ofthis specific key factor in the development of our tools (see chapter 4).
Traditionally, different engineering departments carry out the design, planning and manu-facture of the product with no integration or feedback and so assembly problems are identifiedonly at the later stages of production. In order to reduce lead times and product costs effec-tively, manufacturing and assembly issues must be detected and considered during design.This requires the introduction of Assembly-Oriented Design (AOD) so that product devel-opment and assembly planning can be performed simultaneously rather than consecutively.
Some basic rules can simplify the assembling of steel parts in the shipbuilding industry[CS08]:
• Minimise the number of different parts - use "standard" parts.
• Minimise the number of parts (incorporating multiple functions into single parts, mod-ularize multiple parts into single sub-assemblies, etc.)
• Standardise to reduce part variety (especially for small steel parts like brackets).
• Provide alignment features (i.e. like position line and identification number of stiffenerson steel plates).
• Make parts so that it is easy to identify how they should be positioned for insertion.
• Insert new parts into an assembly from above.
• Assemble in open space, not in confined spaces (avoid work inside double bottom ⇒
increase block dimension)
• Prioritise easily handled parts.
• Eliminate the need for workers to make decisions or adjustments.
• Ensure accessibility and visibility.
• Avoid or minimise part orientation during assembly (i.e. privilege symmetrical parts).
• Maximise part/assemblies symmetry.
79
3.3.4 Design to cost (DTC)
One reason product designers may be unaware of actual product cost is management ac-counting’s focus on external reporting requirements and the reluctance of management todisclose cost information to designers [GD96]. Traditional product design engineering hastaken an "behind the wall" approach where design is done in relative isolation from otherindustry functions. A second reason for lack of design cost optimisation is the culture ofengineering which trains designers almost entirely in the technical side of design, while onlygiving minimal attention to their other implications. This culture makes product perfor-mance a higher priority than cost. When that is combined with deadline pressure, costconcerns may be given only minimal attention.
Design to cost (DTC) is a management strategy and supporting methodologies to achievean affordable product by treating target cost as an independent design parameter that needsto be achieved during the development of a product. DTC is an area which has attractedmuch attention recently. The objective with DTC is to make the design converge to anacceptable cost, rather than to let the cost converge to design [RR00, OYL97]. DTC canproduce massive savings on product cost before production begins.
The basic concept of DTC is to estimate the manufacturing cost during the conceptualand early design stages in order to achieve the following objectives:
• To identify the model parts which might cause high manufacturing cost.
• To provide an environment to estimate alternative cost for comparative design models.
The general approach is to set a cost goal, then allocate the cost goal to all the elementsof the product. Designers must then confine their approaches to set alternatives that satisfythe cost constraint [MW89]. The control of costs to meet these objectives is achieved bypractical trade-off’s involving mission capability, performance and other schedule objectives.However, this is only possible once cost engineers have developed a tool set that designerscan use to determine the impact of their decisions as they make them. The present PhDstudy focuses on the development of this kind of tool to help the designer analyse the impactof their decisions on the ship cycle.
When imposed on a project with a global cost constraint, the process of cost assessmentis carried out throughout the detail design development. Cost, as a key design parameter,is addressed on a continuing basis and is an inherent part of the design development. In thefinal analysis, each system, subsystem and component must be considered with respect to itscost and its effect on the cost of the project. Often the principles of lean design3 are appliedto these systems and components as a means of reducing their cost by virtue of simplifyingthe design (see section 3.3.5), reducing the number of parts and making them easier and lessexpensive to build (see section 3.3.3).
3.3.5 Design for Simplicity
This section refers to the minimization of complexity (!= simplicity) during the design ofa product. More considerations about the complexity assessment are available in section 4.4.
3Derived from the lean manufacturing concept
80
3.3.6 Design for maintenance
The overall competitiveness of the shipbuilding industry depends on the ability to designand build high-level ships with a maximum of benefits for the customer and a minimumof investment of resources. This refers to the reduction of the ship lead time and buildingcosts during the shipbuilding process as well as to the minimising of efforts and costs formaintenance and repair in the operating phase of the ship, and reduce any damages thatmay lead to ship unoperability (for unplanned maintenance).
Consideration of product maintainability and reliability tends to be an afterthought inthe design of ships. The design of the support processes needs to be developed in parallelwith the design of the ship and not after. Parallel design can lead to lower overall life cyclecosts and a product design that is optimised to its maintenance processes. Maintenancecharacteristics of the design and particularly unplanned maintenance are very importantmainly because they leads to a reduction in operating benefits.
Engineering techniques can be applied to systems design to minimize the time and effortrequired to perform periodic preventive maintenance as well as unscheduled maintenance.Some recommendations can be given to achieve higher quality, better reliability, lower op-erating cost, and better maintainability. For instance [Cro07]:
• Reduce the number of parts to minimize the possibility of a defective part or an as-sembly error
• Reduce the complexity and time of the assembly/disassembly process
• Improve the accessibility for testing or inspections of the components of the product
• Apply DFA to minimize non-value-added manual effort during the assembly of theproduct
• Use modular design for components with greater probability of replacement to facilitateassembly/disassembly
• Utilize standard parts to minimize the amount of spare parts
• Provide self test and self-diagnosis as more as possible
• Ease piping connections (i.e. flange connection)
3.3.7 Design for environment (DFE)
DFE is a relatively new field, developed in parallel to pollution prevention. The aims ofDFE are to minimize raw material consumption, energy and natural resource consumption,waste/pollution generation, health and safety risks, and ecological degradation over the entirelife of the ship [Bar96].
81
DFE integrates environmental considerations into the design of ships with a better envi-ronmental performance over the ship’s entire life cycle. Decisions made about the types ofmaterials and other resources, a well as manufacturing processes to be used during produc-tion, affect the environmental performance of the ship. Following the finished design, theship’s environmental attributes are generally fixed and cannot be changed [OAT04]. There-fore, a systematic integration of environmental considerations into the earlier stages of designis essential for a greater environmental benefit. Incorporating DFE attributes into ship de-sign has some benefits, such as reduced energy and material use, reduction of emissionsand waste, focus on material selection issues: design for recycling, design for disassembly,management of toxic materials, and evaluation of environmental attributes.
Some of the benefits of DFE are:
• lowering life cycle cost,
• conserving energy and other natural resources,
• simplifying environmental management.
How can the environmental performance of ships be evaluated by means of a set of criteriaso that design, construction, equipment, operation, maintenance and scrapping contribute tosustainability? Applying some kind of environmental score to each process and quantifyingthe impact is a logical extension of the cost assessment methodology. The environmental costassessment could apply energy, waste/pollution and natural resource budgets to each processfactor and monitoring these budgets as fabrication progresses. Two different methodologiescould be used to assess the environmental impact of a product:
• Each structural process factor could be considered as responsible for a little pollution.Each polluter could be analysed to determine the categories of pollution caused directlyand indirectly by each activity. For the purpose of this PhD thesis, pollution is definedas any emission, effluent, spill, discharge, or disposal into the air, land or water, whetherroutine or accidental. The sum of each small pollution will lead to the assessmentof the overall pollution. But the drawback of this method is that a lot of detaileddata are required. SIMAPRO is a LCA software which can assess the environmentalimpact of product with this method. But the methodology is highly time consumingespecially regarding the filling of the input database and simplifications are requiredfor an efficient use in the shipbuilding industry [MS99].
• A more achievable method of including the environmental impact of structural fabri-cation steps would be to determine a relative environmental consequence or damagefactor (penalty) for each process. The sum of these impact factors would be a measureof the relative risk of a particular design. And, like the cost assessment, the envi-ronmental impact could be used as a measure of the overall performance of the shipdesign.
For lack of time, the present PhD thesis doesn’t take into account environmental factors.We are aware that this is a gap to be filled in later work.
82
3.3.8 Design for safety (DFS) – Risk Based Design (RBD)
The term Design for Safety (DFS) refers to the general engineering principle of includingnavigation safety as a design characteristic at the early stage of design. DFS is a philosophyrather than a prescriptive approach and as such it is highly flexible [Cai02].
Safety is becoming of paramount importance as rules are becoming ever stricter in everyarea of maritime activities. Improvement in safety rules traditionally has taken place as areaction after every accident. History is full of examples. Improvements in stability andsafety of Ro-Ro4 ferries took place after Herald of Free Enterprise and Estonia disasters[OAT04]. Improvements on bulk carriers took place following a number of bulk carrierlosses. In recent years, there has been a big change in the safety culture of the maritimeindustry. Rather than waiting for an accident to happen and then act in haste to set up newrules, all pertinent knowledge deriving from such accidents could be analysed and stored toimprove the safety as early as possible in the design process [Pap09]. Today, it is widelyaccepted that rules provide minimum standards on average and in some areas there are noteven rules to provide minimum standards of safety. Consequently, DFS should systematicallyintegrate risk analysis in the ship design process with prevention/reduction of risk (to life,property, and the environment) embedded as a design evaluation attribute.
DFS is a real opportunity for ship owners to have ships customized to their needs whilemaintaining the same safety levels. However, DFS is a very expensive and time consumingapproach [BSCC00]. Indeed, the resources required for additional safety during the designstage will inevitably have a cost. It is from this background that the marine DFS emerged.The key drivers of the philosophy are to keep safety as an important functional characteris-tic of the design and to speed up the process of risk and cost analysis, so that the processitself becomes more usable. IMO, MSC, SOLAS, ISO, IACS, MARPOL are continuouslyimproving and implementing the safety requirements in the shipbuilding industry. In par-ticular, the IMO Maritime Safety Committee MSC recently adopted a new philosophy anda working approach for developing safety standards for passenger ships [PAK+09]. In thisapproach, modern safety expectations are expressed as a set of specific safety goals andobjectives, addressing design, operation and decision making in emergency situations withspecial attention paid to flooding survival analysis and fire safety analysis.
3.3.9 Design for life cycle
In order to improve the design of products and reduce design changes, cost, and time tomarket, life cycle engineering has emerged as an effective approach to address these issuesin today’s competitive global market. As over 70% of the total life cycle cost of a product iscommitted at the early design stage, designers can substantially reduce the life cycle cost ofproducts by giving due consideration to the life cycle implications of their design decisions[SPJW02].
Life cycle assessment (LCA) is defined as the assessment of the environmental impact ofa product over its entire lifetime whereas life cycle cost (LCC) analysis is the systems engi-neering method of considering all costs over the product life in the design process [HBC+03].The present section speaks only of the LCC analysis.
4Roll-On/Roll-Off
83
People are always concerned about product cost, which encompasses the entire productlife from conception to disposal. Manufacturers usually consider only how to reduce the costof materials acquisition, production, and logistics. In order to survive in the competitivemarket environment, manufacturers now have to consider reducing the cost of the entire lifecycle of a product, called LCC.
Design improvement such that maintenance is easier and that ship problems are lessfrequent or less significant may certainly reduce the cost of exploitation and increase safety.Currently, the LCC is not yet a major issue for shipyards. This is an economic and strategicmistake. Integration of the LCC, including maintenance costs and operating costs in thedesign procedure, could be used by designers and shipyards as a huge selling argument. If theshipyard can show to the ship-owner that the proposed design satisfies the standard technicalrequirements and the usual ship-owner specifications but also considers maintenance andoperation issues, the shipyard may get the order even if its offer is not the cheapest. Ship-owners want to minimize short term investment but above all maximize their profits.
The primary objective of the design work, besides creating the information needed to buildthe ship, is to satisfy the ship owner’s requirements at minimum cost. An owner requires aship which will give him the best possible returns for his initial investment and running costs[Eyr01]. Life cycle costs have often been a major consideration for commercial ship ownerswho must look at the bottom line for profit and a return on their investment. For instance,if the cost of design and production cannot be coupled within a reasonable amount of timethe ship will not be built. In the same way, if the operating and maintenance costs exceedoperating revenues, again the ship will not be built. Design methods for minimising the lifecycle cost of the product thus become very important and valuable.
A ship LCC includes design, manufacturing and acquisition costs as well as operationsand retirement costs throughout the life of the ship (see equation 3.1).
CLifeCycle = CConception + CAcquisition + COperating + CDisposal (3.1)
where CLifeCycle Life cycle costs,CConception Conception costs (including design and production costs),CAcquisition Acquisition costs,COperating Operating costs,CDisposal Disposal costs.
Usually, the life of a ship is divided into essentially four stages [Ull97]:
• Conception stage: All activities necessary to develop and define a means for meetinga stated requirement. For ships and equipment, this normally includes research anddevelopment, design, contract specifications, identification of all support necessary forintroduction into service, and identification of the funding required and a managerialstructure for the acquisition. This stage includes the design and manufacturing costs.
• Acquisition stage: All activities necessary to acquire the ship and provide support forthe ship and equipment identified in the conception stage.
• Operation stage: All activities necessary for operation, maintenance, support, cleaningand modification of the ship or equipment throughout its operational life. The in-service stage is normally the longest stage.
84
• Disposal stage: All activities necessary to remove the ship or equipment and its sup-porting materials from service (disassembling, recycling, etc.).
3.3.10 Design for Robustness
Robustness is defined as insensitivity (or stability) with respect to uncontrollable param-eters and is becoming a standard concept, particularly for innovative designs. Many inputparameters (e.g. loads, material data, thickness, etc.) held constant during the optimisationprocess, are subject to uncertainties causing variations of the values in the criteria set and/orviolation of constraints (infeasible design). They can also be costly to control. One way isto introduce safety margins on the constraints, but this leads to a reduction of the designspace. Robust design has been developed with the expectation that an insensitive designcan be obtained (robust means that the product or process performs consistently on targetand is relatively insensitive to factors that are difficult to control).
The robustness measure η, developed by Dr. Genichi Taguchi [Ros88, Mon91, CAB+06],is the ratio of the mean of the attribute value µ to the standard deviation σ resulting fromuncertain parameter values. In fact it is the ratio of predictability versus unpredictability.For the "nominal is the best" type of attribute, the "signal-to-noise" ratio is given by Equation3.2. A larger value for η indicates less sensitivity to noise conditions and hence higherreliability.
η = 10 log10(µ2
σ2) (3.2)
Taguchi advocates a philosophy of quality engineering that is broadly applicable. Heconsiders three stages in a product’s or process’s development:
1. System design (an engineer uses scientific and engineering principles to determine thebasic configuration)
2. Parameter design (specific values for the system parameters are determined)
3. Tolerance design (it is used to determine the best tolerances for parameters)
Taguchi recommends that statistical experimental design methods be employed to assistin quality improvement, particularly during parameter design and tolerance design. Exper-imental design methods can be used to find a best product or process design, where by bestwe mean a product or process that is robust or insensitive to uncontrollable (noise) factorsthat will influence the product or process once it is in routine operation.
A key component of Taguchi’s philosophy is the reduction of variability. Generally,each product or process performance characteristic will have a target or nominal value. Theobjective is to reduce the variability around this target value. Taguchi models the departuresthat may occur from this target value with a loss function. The loss refers to the cost thatis incurred by society when the consumer uses a product whose quality characteristics differfrom the nominal.
The robust design method greatly improves engineering productivity [Pha09]. Variationreduction is universally recognized as a key to reliability and productivity improvement.There are many approaches to reducing the variability, each one having its place in theproduct development cycle. The robustness strategy provides the crucial methodology forsystematically arriving at solutions that make designs less sensitive to various causes of
85
variation. It can be used for optimising product design as well as for the manufacturingprocess design.
3.3.11 Design for process
Strong emphasis has been placed on shipbuilding process improvement over the pastdecade. Six Sigma and Lean Thinking have been the dominant models. Six Sigma focuses onreducing process variation through consistent repeatable processes, process design/redesign,defect prevention, statistical analysis, and voice of the customer. Lean focuses on eliminat-ing non-value added activities by identifying the value stream, eliminating over-production,and over-processing, and creating process flow [INW+06]. Comparisons of the two modelsare provided in Tab. 3.1. The integration of Lean and Six Sigma provides a rapid processimprovement strategy for attaining organizational goals. When separated, Lean Manufac-turing cannot bring a process under statistical control, and Six Sigma cannot dramaticallyimprove cycle time or reduce invested capital [Men04]. Together, synergistic qualities arecreated to maximize the potential for process improvement. Lean Thinking provide toolsto "Make it simple" by the elimination of waste while Six Sigma provides tools to "Make itperfect" by the elimination of variations.
Program Six Sigma Lean Thinking
Theory Reduce variation Remove waste
Application guidelines 1. Define 1. Identify value
2. Measure 2. Identify value stream
3. Analyse 3. Flow
4. Improve 4 Pull
5. Control 5 Perfection
Focus Problem focused Flow focused
Assumptions A problem exists Waste removal will improve
Figures and numbers are valued business performance
System output improves if Many small improvements are
variation in all processes is reduced better than system analysis
Primary effect Uniform process output Reduced flow time
Secondary effect Less waste Less variation
Fast throughput Uniform output
Less inventory Less inventory
Fluctuation - performance New accounting system
measures for managers Flow - performance measure for managers
Improved quality Improve quality
Criticisms System interaction not considered Statistical or system analysis not valued
Process improved independently
Table 3.1: Comparison of Lean and Six Sigma methodology [INW+06]
3.3.11.1 Design For Six Sigma (DFSS)
DFSS is a separate and emerging business-process management methodology related tothe traditional Six Sigma concept.
Six Sigma was originally developed as a set of practices designed to improve manufactur-ing processes and eliminate defects, but its application was subsequently extended to other
86
types of business processes as well [Mot09]. The particulars of the methodology were firstformulated by Bill Smith at Motorola in 1986.
Originally, σ is used to represent the standard deviation of a statistical population.The term "six sigma process" comes from the notion that if one has six standard deviationsbetween the mean of a process and the nearest specification limit, there will be practically noitems that fail to meet the specifications. This corresponds to 3.4 defective parts per millionopportunities (DPMO) or 99.9997% efficiency. However, Six Sigma has evolved into oneof the most powerful management strategies to fundamentally change the way corporationsdo business and to improve their bottom line. Six Sigma began as a quality improvementattempt to identify and reduce all sources of variation. It then evolved into an overallmanagement strategy for measuring and improving performance across different processes.The main emphasis of Six Sigma is the application of statistical tools in a disciplined manner,which requires data-driven decision making. Six Sigma is about controlling processes to getthe desired results.
Six Sigma has two key methods: DMAIC and DMADV, both inspired by [FB05]. DMAICis used to improve an existing business process and DMADV is used to create new productor process designs. DMADV is also known as DFSS.
DMAIC The basic DMAIC method consists of the following five steps:
• Define process improvement goals that are consistent with customer demands and thebusiness strategy
• Measure key aspects of the current process and collect relevant data.
• Analyse the data to verify cause-and-effect relationships. Determine what the rela-tionships are, and attempt to ensure that all factors have been considered
• Improve or optimise the process based upon data analysis using techniques like Designof experiments
• Control to ensure that any deviations from target are corrected before they result indefects. Set up pilot runs to establish process capability, move on to production, setup control mechanisms and continuously monitor the process
DMADV The basic DMADV method consists of the following five steps:
• Define design goals that are consistent with customer demands and the enterprisestrategy
• Measure and identify CTQs (characteristics that are Critical To Quality), productcapabilities, production process capability, and risks.
• Analyse to develop and design alternatives, create a high-level design and evaluatedesign capability to select the best design.
• Design details, optimise the design, and plan for design verification. This phase mayrequire simulations.
• Verify the design, set up pilot runs, implement the production process and hand itover to the process owners.
87
Today’s complex ships require large volumes of material, parts, and equipment. As theamount of material, parts, and equipment increases, so does the possibility for defects andvariation within the shipbuilding process. This situation is further complicated by the num-ber of different processes that occur before a single assembly becomes a part of the ship. Thecost in waste and lead time is potentially very high. Likewise, the complexity of ship systemdesign and the cost to rework these defect leaves little room for error. A design defects insome scenarios can result in serious setbacks.
The integration of Lean and Six Sigma is a relatively new concept in shipbuilding andis still evolving. The DFSS methodology produces many benefits when implemented in shipdesign and production including improvement of ship production and operational capabilitiesthrough early stage intervention and consideration, improved customer satisfaction, reduceddevelopment cycle times, and reduced life cycle costs. The main goal of using DFSS is tounderstand the customers needs and requirements in order to design the correct product[INW+06].
3.3.11.2 Design for Lean Manufacturing
Lean manufacturing is a generic process management philosophy derived mostly fromthe Toyota Production System and identified as "Lean" only in the 1990s [JR91]. Leanmanufacturing is a variation on the theme of efficiency based on optimizing flow by decreasingwaste like [WJ03]:
• Transportation (moving products that are not actually required to perform the pro-cessing)
• Inventory (all components, work-in-progress and finished product not being processed)
• Motion (people or equipment moving or walking more than is required to perform theprocessing)
• Waiting (waiting for the next production step)
• Overproduction (production ahead of demand)
• Over Processing (due to poor tool or product design creating activity)
• Defects (the work involved in inspecting for and fixing defects)
The main objective of process improvement, specifically lean manufacturing, which was firstinstituted in Japanese shipbuilding, is cost reduction via the elimination of unnecessaryoperations, waiting times, and inventories. Many lower level process improvements are thedirect result of worker initiatives. Although most of these improvements are small, thecomplete effect is beneficial.
3.3.12 Conclusion
What is the primary objective of a shipyard? As every business school in the world hastaught us, the primary goal is to maximize free cash flow to investors. The free cash flow isprimarily driven by profit, so that the first objective of the shipyard become to increase profitby reducing the production cost. Many people may argue that safety, ship performance,and delivery should be the shipyard’s main goals. No doubt these are important goals.However, these are simply important requirements that must be met. Minimising life cycle
88
cost should be the goal. The ship design is a complex multidimensional space. Safety, quality,environmental, productibility, and other product attributes are constraints that must be metto some target level in order for the ship to be viable in the market. Because the ship designis a non-linear complex space, there are multiple regions of localized minimum for LCC.Some of these targets are blocked by the constraints.
Within a holistic ship design optimisation we need to mathematically understand exhaus-tive multi-objective and multi-constrained optimisation procedures. Optimisation problemsand their basic elements may be defined as the following (see Fig. 3.5):
• Design variables – This refers to a list of parameters characterizing the design beingoptimized; for ship design this includes ship’s main dimensions, unless specified bythe ship owner’s requirements and may be extended to include a ship’s hull form,arrangement of spaces, structural elements and networking elements (piping, electrical,etc), depending on the availability of the input data.
• Design objective function – A function associated with an optimisation problem whichdetermines how good a solution is, for instance, the total Life Cycle Cost of a ship.
• Design constraints – This mainly refers to a list of limits mathematically defined inorder to keep a feasible solution at the end of the optimisation process. Basically theselimits result from regulatory frameworks related to safety (stability limit, yield stressof steel, etc.) and may be expanded by the cost of materials (for ships: cost of steel,fuel, labour) and other case specific constraints.
• Optimal solution – A feasible solution that minimizes (or maximizes, if that is the goal)the objective function is called an optimal solution. For multi-criteria optimisationproblems, optimal design solutions are called Pareto front and may be selected on thebasis of trade-offs by the decision maker.
Qualityconstrainst
Safetyconstrainst
Producibilityconstrainst
Environmentalconstrainst
MinLCC
OptimalSolution
Figure 3.5: Minimizing the LCC while meeting other constraints
89
Though a holistic approach to the ship design problem appears theoretically well estab-lished, it remains for the researchers and engineers to develop and implement a long list ofapplications, addressing the complex problem of ship design for life-cycle. This is a longterm task of decades, requiring profound skills and understanding of the physics, technologyand design of ships, a clear domain of properly trained naval architects. DFX methods oftenprovide conflicting guidelines mainly because there too many issues at the same time!
Some elements to help the development of the holistic LCC optimisation of ships will beexplored in section 4.
3.4 Selection of a cost estimation method
3.4.1 Introduction
The ability of a company to compete effectively in the increasingly competitive globalmarket is influenced to a large extent by cost [SPJW02]. Cost is the expenditure necessary forthe attainment of a goal; therefore cost estimation is predicting the cost prior to undertakingthe activity. The motivation to estimate costs is to aid decision making, cost managementand budgeting. If costs are higher than the price that organisations can achieve for productsthen the products will make losses and this influences the future viability of organisations.Knowledge of the potential cost before the task is undertaken provides an opportunity tooptimise design to minimize costs or to quote a price to customers that would enable a profitto be achieved [ELM06].
Cost assessment occurs at various stages of ship design development. Economic evaluationas early as possible, in the design phase, is therefore crucial to find the best price–functioncompromise for the ship project.
After a detailed presentation of various relevant cost assessment methods (see section3.4.2), we propose a Multi Criteria Decision Aid (MCDA) method to select an appropriatecost assessment method at each stage of the ship cycle (see section 3.4.4). In order tocompare the cost performance between all alternatives, the evaluation of each alternative isperformed by PROMETHEE. A part of this research has been published in [CR09].
3.4.2 The different cost assessment methods
Cost assessment occurs at various stages of ship design development. Economic evaluationas early as possible, in the design phase, is therefore crucial to find the best price–functioncompromise for the projects or product. However, economic evaluation during the designphase is not easy. It is very different from assessment when the product/process designis complete and detailed which allows the cost of all optimisation choices to be taken intoaccount. In the design phase, the project or product is never completely defined (see section2.4). It is necessary in this phase to implement rapid and more or less precise cost estima-tion methods (depending on available data) allowing the designer to select one solution inpreference to another on economic grounds.
In general, cost-estimating approaches can be broadly classified as intuitive, parametricor statistical techniques, and analytical models. However, the most accurate cost estimatesare made using the analytical approach. Among the many methods for cost estimating, atthe design stage, are those based on knowledge bases, features, operations, weight, material,
90
physical relationships, and similarity laws [SA02]. In this section we present the informationavailable to select the most appropriate cost assessment methods among the following:
• Intuitive method (IM) or Expert opinion
• Case based reasoning (CBR) or Analogy analysis
• Parametric method (PM) or Statistical analysis
• Feature-Based Costing (FBC)
• Fuzzy logic method (FLM)
• Neural networks method (NNM)
• Simulation method (SM)
3.4.2.1 Intuitive method (IM) – Expert opinion
The role of the cost engineer in the design process is to provide models which are capableof establishing a cost value from the data available at the different design stages. Costestimation is often regarded as a mysterious art as it is somewhat more of a statisticaldiscipline compared with the other engineering activities. Establishing a cost estimate atany stage of the design requires a high degree of appreciation of the processes which occurin both design and construction process [Bol07]. Detailed costing may require knowledgeof how long certain construction processes take to be carried out, for example, joining astiffener to a plate taking into account size, material and welding techniques, while costingfor a concept design will require, for example, knowledge of how the use of the differentspaces of the ship impacts on cost.
Principle and process The IM is based on the experience and the opinion of the esti-mator. The cost engineer requires both a good database of historic information on previousships and good contacts with industrial partners to forecast how technical and financialchanges may impact on construction costs. Once this information is established, the costengineer uses his expertise to identify the cost estimation models which correlate well withboth the type of vessel and the capabilities of the shipyard and his experience to enhanceconfidence in the result predicted by the model.
The added value of an experienced cost estimator is his understanding of which particularCER is applicable to each situation. Since most CERs include empirically derived factors, itis necessary to have them used by someone who appreciates and understands their implica-tions [Mir06]. This is particularly important if new technologies which have not been usedin the past are planned. Furthermore, expert opinion is very useful in providing a rationalcross-check of the data that modern, complex, computer generated equations produce. Itis therefore the appropriate combination of the value of judgement provided by the expertused in conjunction with CERs that can result in better forecasts.
Strong points and drawbacks The result of the expert opinion analysis is always de-pendent on the estimator’s knowledge. It can prove to be very effective but it can also bemisleading. Generally, low estimates are generated by people whose interests are served bylow estimates and high estimates are generated by people whose interest is served by highestimates. It may be seen that the competitiveness of a shipyard may be encapsulated inthe cost engineer’s knowledge.
91
3.4.2.2 Case based reasoning (CBR) – Analogy analysis
This method uses a direct comparison between two similar products or sub-products andis based on the experience and knowledge of the designer. The effectiveness of this methoddepends largely upon the ability of the single designer to identify differences between theintended and existing systems. Although most of the applications of this method are basedon the expertise of a designer, some well-developed applications use CBR [RG98, DC99].The implicit assumption is that similar products have similar cost. By comparing productsand adjusting for differences it is possible to achieve a valid and usable assessment. Themethod requires the tools to identify both the similarity and differences between items. Thiscan be done through the use of experience or databases of historical products.
CBR can be used to model, store, and re-use historical data, and take hold of knowledge forproblem-solving tasks. An important feature of CBR is the ability to learn from past casesor situations. A CBR system stores and organises past situations, then chooses situationssimilar to the problem at hand and adapts a solution based on the previous cases [RR00]. Anoverview of the CBR process is illustrated in Fig. 3.6. The method is mainly used with GroupTechnologies (see section 2.2.5.1), which allow a typical solution for every proposed designscenario. It was also assessed as a better alternative in the context of rapid technologicalchanges; while the other methods are hard to update, an analogical comparison is less likelyto overlook the impact of the changes.
CBR uses the solutions of past experiences to solve a problem. This kind of reasoninguses the following basic operations: the recognition of the problem, the recalling of similarexperiences and their solutions, the choice and the adaptation of one of the solutions (sourcecase) for the new problem (target case), the evaluation of the new situation and the learn-ing of the solved problem. It implements the techniques of case indexation and similaritymeasures as well as adaptation procedures from the source case solution to the target case.
Case-based reasoning finds its justification in three areas [DC99]:
1. Cognitive Psychology – Cognitive psychology has shown that most people improvetheir capacity to solve problems by experience. They have more difficulty in solvingnew problems than in solving problems that they have already seen or problems thatare similar to those that they have already seen.
2. Rule-Based Systems – CBR is considered as an alternative to expert systems. Rule-based systems can be used when one knows the area of application, so that a theory, orknowledge of, or at least production rules, exist. This capitalization and formalizationof knowledge in the form of rules is rendered complex and sometimes impossible whenthe area of application is weakly theorised. Moreover, it is accepted that the maximumnumber of rules for an expert system is around 1500. CBR systems, thanks to theircapacity of reasoning on past experience, avoid the formalization of knowledge in theform of rules. They also make it possible to call on the intuition, the judgement andthe habits of the expert, and to obtain a result or a decision, even when there is notheoretical model of the system in question. Mixed systems which integrate an expertsystem in the adaptation phase of case-based reasoning often exist. Furthermore, oneof the advantages of the CBR technique is the facility of development.
92
3. Analogical Reasoning – Case-based reasoning is a particular case of analogical rea-soning frequently used in problem resolution where source and target cases belong tothe same universe.
Principle and process The principle of CBR is based on dependence relationships thatexist between the problem and the solution. Indeed, it considers that if it is possible toestablish a link between the source problem specifications and the target problem specifica-tions, then it is possible to transpose and adapt the solution of the source case to the targetcase and to adapt the new solution by recreating the link between the target case problemand its new solution. CBR follows the elementary steps proposed by [Sla91] (see Fig. 3.6):
1. Retrieve relevant candidates from characteristics of the new cases.
2. Select the best case of the preceding extractions with the help of a similarity measure.
3. Modify, and adapt selected cases in order to propose a solution or an interpretationfor the new case.
4. Test the proposed solution, to evaluate the solution.
5. Carry out the learning by recording the new cases and release the indexation of cases.
Knowledge Base
PreviousCases
PreviousCases
PreviousCases
PreviousCases
PreviousCases
PreviousCases
PreviousCases
NewCase
Retrieve
AdaptedSolution
Revise and test
NewCase
RetrievedSimilarCase
Testedcase
Store
Learning
Adapt
Problem
Figure 3.6: Case based reasoning process (CBR)
Picking out cases from the memory banks is one of the essential elements of CBR. Thisrequires that particular attention is paid to the indexation method and similarity measures.
Adapting cases in order to propose a solution or an interpretation. In some systems, theadaptation is not useful every time. There are three possible adaptation approaches; theadaptation by substitution, the derivational adaptation and the adaptation by transforma-tion.
93
Strong points and drawbacks CBR provides the ability to propose a solution veryrapidly. Moreover, it is a transparent process. At any given time the user knows theorigin of the solution and can correct the result. Moreover, CBR plays the role of thebusiness’ collective memory (as a knowledge management system), allowing the user to usesolutions elaborated by others. That preserves the trade knowledge for the employer whenan employee leaves the enterprise or changes position. The capacity of CBR to take intoaccount unknown data is important [DC99]. In this case, research will be carried out onlyon known parameters. The system will then propose different solutions with estimated oraveraged values for unknown data.
CBR is often used to solve problems where no obvious formalization of trade knowledgeexists, which is often the case for complex sets. Moreover, by working on the closest case toestimate the cost of a part or a product, CBR is able to solve particular cases which havealready been seen. It is also able to avoid previously committed errors. One of the majoradvantages of CBR resides in its capacity to combine several methods in the adaptationphase. Indeed, once the closest case is selected, the determination of the different costsof the project or the product depends on the kind of cost and the available data. Thus,for the determination of the setting time of a manufactured part, it is possible to use aCEF or to make a simple comparison with the closest case found. More generally, it is alsopossible to make a regression from extracted close cases. The indexation of cases allowsthe important parameters related to the cost of studied products to be investigated. It is,however, difficult to obtain general trends. For that, the user does not have any alternativeto making several successive tests and comparing the results. The use of a cost estimationmethod based on CBR, in a company, is less easy than in the case of the parametric methodbecause it is necessary to provide the indexation of cases, the similarity measurements andthe adaptation functions.
A drawback is that CBR requires a number of past cases in order to be effective. In ahighly innovative company past cases may not be available so that will therefore reduce theeffectiveness of the CBR system [RR00]. The innovation can also be slowed because thedesign is always based on past experience.
3.4.2.3 Parametric method (PM) – Statistical analysis
Also known as a top-down approach (see section 2.3.4.1), this method seeks to evaluatethe cost of a product based on certain characteristic parameters. Cost estimation relations(CER’s) and associated mathematical algorithms are developed by establishing a relationshipbetween one or more parameters that are observed to change as the cost changes. Theseparameters are typically referred to as cost drivers. The parametric models are applicableonly for a specific product type within a specific manufacturing technology. Many of thesemodels are constructed around statistical relationships, which are supposed to be universal.Since the parametric method generates cost data based on the knowledge of certain physicalcharacteristics or parameters (such as ship size, weight, horsepower, etc.), it is used whena quantitative mapping between the product parameters and its cost can be drawn in theform of mathematical expressions.
Usually, the parametric method functions are like a "black box". In this case, it is verydifficult to understand the important elements of the manufacturing and to be able to justifyresults. Also, if the context of the workshop is modified, it is then necessary to remake
94
an estimation. This method is useful only if used in combination with other methods.Parametric methods meet the criteria of precision and speed of results provided that we areworking with well-defined families of parts and that no justification of results is necessary.
Principle and process The PM use the knowledge of a certain number of physical char-acteristics or parameters such as the weight, the volume, and the number of items in orderto evaluate the cost.
To illustrate this concept more clearly the following example is presented. Typically, forship development, mass relates to the cost of production. That is, as the weight of the shipincreases, so does the cost of producing it. What’s more, this particular relationship is oftendescribed as linear, as illustrated in Fig. 3.7. In this hypothetical example the points of thegraph represent the relationship of cost to mass for different ships. The line traversing thepoints represents a linear relationship i.e. as the weight increases so does the cost. With therelationship described it is then possible to use the formula to predict the cost of a futureship based on its weight alone. Within the field of cost estimating this relationship is knownas a cost estimating relationship (CER) or cost estimate formulae (CEF).
This is a rather simplistic illustration describing the main principals of parametric esti-mating. Nonetheless, variations of this approach are a widely used method within the shipindustry to predict the cost of a product under development and throughout the life cycle.
y = 46.721x + 349013
R2
= 0.7745
0.0E+00 5.0E+03 1.0E+04 1.5E+04 2.0E+04 2.5E+04 3.0E+04 3.5E+04 4.0E+04
Weight (ton)
Co
st
Figure 3.7: Simple linear cost estimating relationship
Strong points and drawbacks The primary advantage of parametric cost assessment isthat the data reflect changing cost conditions. Parametric method estimating can be usedthroughout the product life cycle. However, it is mainly used during the early stages ofdesign and for trade studies.
The parametric method is very useful because of its rapidity of execution. It can becriticised for working like a "black box": that is to say that from the specifications the onlyresults we obtain are different costs [DC99]. We don’t know the origin of these costs, whichcan discourage users. During the design step, not all the information is available. Somespecifications needed for the CEF cannot yet be defined; but, for the model to work, the
95
user must enter all the parameters. Consequently, the designer will have to estimate missingparameters. This will cause uncertainty in the result. One of the advantages of the CEFduring the design step is to make clear the influence of parameters on the economic value ofthe product. Indeed, if designers are aware of the influence of the different parameters on thecost, they will be able to optimise its design from an economical point of view: this is a veryimportant aspect for designers. CEFs cannot solve particular cases. Even if the specificationparameters of a part are within the limits imposed by the CEF, that is not enough. Someparameters which are not taken into account in the CEF can become important in particularcases. Indeed, it can be seen in Fig. 3.7 that even if the value of a parameter is within thedefinition area of the CEF, the error in the cost can be great. Furthermore, the "black box"functioning of the CEF renders the detection of this kind of error very difficult. Finally, theparametric method based on the use of the CEF meets the designers needs thanks to itsrapidity of execution and to the information that it provides on global trends, in spite of thedisadvantages previously presented.
One disadvantage is that the available data may not be sufficient to obtain a valid costrelationship. This is important because the technique can be sensitive to inconsistencies andirregularities in the data [GD96, Smo98]. Technologies, investment in equipment, as wellas new approaches to production organization have led the shipyards to use historical datawith extreme caution. In order to obtain sufficiently accurate results it is recommended thatrecent historical data is used and only for one type of vessel.
While these parametric costing techniques can be applied to a new design fairly easily theyoften do not possess the ability to capture factors such as the introduction of new productiontechniques and process. Consequently, it is very difficult to identify how any optimisation ofproduction processes may impact on cost [Bol07]. This approach will prove very successful ifapplied to vessels which the yard has previous experience. However, if construction requiresthe introduction of production process of which the yard has no previous experience, thesetechniques can only provide a much reduced degree of confidence in the cost estimation.
Further, the expense of detailed cost data collection can be high and must be traded offagainst the potential benefits.
In summary, PM is an excellent predictor of cost when procedures are followed, data ismeaningful and accurate, and assumptions are clearly identified and carefully documented.Unfortunately, it is rarely the case in a real industrial environment.
3.4.2.4 Feature-Based Costing (FBC)
Feature-Based Costing is a method for estimating the cost of a product based on theanalysis of a series of its elementary characteristics, called product features. Products canessentially be described as a number of associated features such as holes, inner contour, outercontour, welding length, welding position, cut-outs, bevels, etc. It follows that each productfeature has cost implications during production, since the more features a product has themore manufacturing and planning it will require. The growth of CAD/CAM technology andthat of 3D modelling tools have largely influenced the development of FBC [GRR06]. Withthis approach, it is possible to evaluate the consequences including or excluding the featurewill have on the costs of a single component, but also on the system of costs of the entirelife cycle of a product consisting of several components.
96
For instance Fig. 3.8 shows the 3D view of a small ship section on the right side and itsrelated hierarchical properties such as assembly strategy and SWBS on the left side.
For general costing analysis there was a tendency for companies to use a computer-basedtool at the detailed manufacturing cost estimation level [RK03]. The results produced fromthese analyses seemed to be fairly accurate.
Principle and process This approach, also called bottom-up approach (see section 2.3.4.2),allows the evaluation of cost from a breaking down of the required work into elementary tasksand relies on detailed engineering analysis and calculations. To apply this approach, the costanalyst needs detailed design and configuration information for system components and ac-counting information for all materials, equipment, and labour. This method assumes itsusefulness when costing information for workshop processes is readily available. Given suffi-cient design detail, this method can make very accurate cost estimates. However, it is verytime-consuming and does require detailed knowledge about the product being designed andthe relevant processes.
One of the prerequisites of this approach is that the product model needs to be detailedenough to allow materials and production labour to be established. This means that thestructural definition, systems and equipment need to be defined and may rule out thisapproach being used in the earliest stages of design [Bol07]. However, as ship design toolsare continuously improved, it is becoming easier to add preliminary production details atthe start of the design so that production considerations can be incorporated in the designprocess. Consequently, this technique may be employed shortly after the initiation of adesign project.
Figure 3.8: Example of ship section with assembly and design features [Ave09]
97
Strong points and drawbacks The advantages of this approach are evident [GRR06,RR00, Bol07]:
• a clear link between the design choices and their implications in term of cost, and asa consequence an increase of the potential capacity for correcting and optimising thedesign;
• easy to use compared to other approaches, by virtue of a simplification of data collectionto calculate the cost of a product;
• the transverse nature of the main feature typologies and the resulting possibility ofapplying the method even when no similar studies exist and no previous data areavailable.
Other reasons for using FBC are that the same features appear in many different partsand products; therefore, the basic cost information prepared for a class of features can beused comparatively often (for instance for welding features). Furthermore, manufacturerswill have numerous past geometric data that can be related to features. Another reasondevelopers explore whether costs should be assigned to individual design features is that itwould provide the designer with a tool to visualise the relation between costs, and aspectsof the design that can influence the production in real time.
Moreover, this approach will capture enough details to allow the effectiveness of productionprocesses to be evaluated and potentially optimised. In the past, extracting the informationfrom the design to perform this kind of analysis would have been very laborious because thecost engineer would have to measure production details directly from plans. However, withmodern ship product modelling software, the identification of parts and junctions can beautomated, providing the cost engineer with a full breakdown.
3.4.2.5 Fuzzy logic method (FLM)
Fuzzy logic is an essentially pragmatic, effective and generic approach [CG98]. For exam-ple, when experts do have rules of thumb for the variation of costs as a function of specificdesign features, it is allows for a systematisation of empirical knowledge and which is thushard to control. The theory of fuzzy sets offers a suitable method that is easy to imple-ment in real time applications, and enables the knowledge of designers and operators to betranscribed into dynamic control systems.
While most conventional methods of cost estimation are deterministic in nature, the designprocess is characterized by intrinsic uncertainties. Fuzzy logic has been used to address theissue of uncertainty in some design applications [PNS99, SA02]. In this approach, a mappingbetween the characteristic parameters of a design and the cost function is achieved througha set of "if-then" rules that incorporates fuzzy logic in terms of varied degrees of membershipof the parameters in the cost function.
Principle and process The more complex a system (see section 4.4), the more difficultit is to make precise statements about its behaviour. The following points are naturallydeduced from these observations:
• rather than modelling the system, it is often more useful to model the behaviour/knowledgeof a human operator used to control the system;
98
• rather than using equations, the operation can be described qualitatively with anappropriate quantitative translation.
The theory of fuzzy sets is based on the notion of partial membership: each elementbelongs partially or gradually to the fuzzy sets that have been defined. The outlines of eachfuzzy set (see Fig. 3.95 ) are not "crisp", but "fuzzy" or "gradual".
(a) Conventional set (b) Fuzzy set
Figure 3.9: Comparison of a conventional set and a fuzzy set [CG98]
A fuzzy set is defined by its membership function. A common example is shown in Fig. 3.10where we define a set of people of "medium height". In classical logic, we would agree forexample that people of medium height are those between 1.60 m and 1.80 m tall. Thecharacteristic function of the set (see Fig 3.10(a)) gives 0 for heights outside the range [1.60m, 1.80 m] and 1 for heights in that range. The fuzzy set of people of "medium height" willbe defined by a membership function which differs from a characteristic function in that itcan assume any value in the range [0,1]. Each possible height will be assigned a degree ofmembership to the fuzzy set of "medium heights" (see Fig. 3.10(b)) between 0 and 1.
A number of fuzzy sets can be defined on the same variable, for example the sets "smallheight", "medium height" and "tall height", each notion being explained by a membershipfunction (see Fig. 3.11). The variable (for example: height) as well as the terms (small,medium, tall) defined by the membership functions, are known as linguistic variable andlinguistic term respectively. Both linguistic variables and terms can be used directly inrules. Fuzzification enables a real value (horizontal axis) to be converted into a fuzzy one(vertical axis).
(a) Classical logic (b) Fuzzy logic
Figure 3.10: Fuzzy membership function [CG98]
5x belongs neither to A nor B; y belongs completely to A; z belongs completely to B; t belongs partiallyto B
99
Fuzzy systems provide a non-linear mapping between crisp input variables and crisp outputvariables and allow the use of linguistic expressions for the rules which define the input-outputrelationship. A fuzzy system consists of four steps:
1. Fuzzification of the crisp input parameters
2. Activation of the appropriate fuzzy rules
3. The use of the fuzzy inference
4. The defuzzification to produce crisp output
The schematic diagram of a fuzzy system is shown in Fig. 3.12.In a fuzzy system, fuzzification is the process of transposing crisp input values to truth
values within relevant fuzzy input sets. Fig. 3.11 shows a simple example with the height ofhuman beings.
The next step in the fuzzy process is to establish the fuzzy rules. Fuzzy rules are in theform of linguistic expressions which interpret the linguistic input information and providelinguistic output information. Fuzzy rules are used in parallel and have the form of "IFpredicate THEN conclusion WITH weighting factor". A predicate is a combination of inputand output parameters by AND, OR, NOT operators. A rule is activated when the value ofthe input variable falls totally or partially in a fuzzy set.
The fuzzy rules can be optionally affected by a weighting factor. This factor may varybetween [0, 1] and states the degree of importance, credibility or confidence of a linguisticrule. The weighting factor shall reduce the membership degree of the conclusion by multi-plication of the result with the weighting factor. In order to manipulate the fuzzy controlapplication parameters externally the weighting factor can be a variable. It is thereforepossible to modify the output of the fuzzy software just by acting on the weighting factor.
Fuzzy inference is the process that determines the activating level of the consequentoutput fuzzy set. The degree of activation of a rule is the evaluation of the predicateof each rule by logic combination of the predicate proposals. The "AND" is performedby achieving the minimum between the degrees of truth of the proposals. The degree ofactivation of the rule is used to determine the conclusion of the rule: this operation iscalled the implication. There are several implication operators (see appendix), but the mostcommon is the "minimum" operator.
The final step, deffuzyfication, provides the final crisp output. A number of methods canbe used, the most common of which is calculation of the "centre of gravity" of the fuzzy set.
Strong points and drawbacks A vital condition for the use of fuzzy rules is the existenceof human expertise and know-how. Fuzzy rule bases cannot provide a solution when no-oneknows how the system operates or people are unable to manually control it.
Figure 3.11: Membership function, variable and linguistic term [CG98]
100
When such know-how exists and can be transcribed in the form of fuzzy rules, fuzzy logicsimplifies its implementation, and the operation is then easily understood by the user.
If human expertise exists, then fuzzy rules can be used, particularly when system knowl-edge is tainted by imperfections, when the system is complex (see section 4.4) and hard tomodel and when the method used requires a global view of various aspects.
Last but not least, fuzzy logic don’t require a large learning database in order to be effec-tive, which would suit industries that produce limited product ranges, such as the shipbuild-ing industry. Thus, this producibility method can cope easily with novelty or innovation.
3.4.2.6 Neural networks method (NNM)
Neural networks [SM96] are a form of artificial intelligence that are used to simulate humanthought processes and thus can be used as a method of linking historic cost information witha proposed design model. For cost estimating purposes [RK03, SPJW02, CLA+07], the basicidea of using NNM is to make a computer program that learns the effect of product-relatedattributes on cost. That is, to provide data to a computer so that it can learn which productattributes mostly influence the final cost. This is achieved by training the system withdata from past case examples. The artificial neural network (ANN) then approximates thefunctional relationship between the attribute values and the cost during the training. Oncetrained, the attribute values of a product under development are supplied to the network,which applies the approximated function obtained from the training data and computes aprospective cost. Neural models can be developed and used for estimating all the stages of aproduct life cycle provided the data is available for training. A great advantage that a neuralnetwork has compared to parametric costing is that it is able to detect hidden relationshipsamong data.
Principle and process Artificial Neural Networks have become more common in use inresearch in recent times due to the effective manner in which they manage complex, multipleinput situations and provide a single output. In shipbuilding research, the ANN is also moreand more used as well for hull resistance prediction [CMM+94], safety prediction [Ger05],maneuverability prediction [EAM06], that for freight rate prediction [BM06] or propulsionprediction [RHF06] [Kru07]. One of their key advantages is their ability to easily modelcomplex, non-linear systems, a feature which is not true of statistical regression methodswhere an appropriate non-linear function must first be found. An advantage of artificialneural networks over statistical methods is their ability to adapt to new data. Once artifi-cial neural network architecture has been designed it can be quickly retrained as new databecomes available.
1 – Fuzzification
2 – Fuzzy Rule DB
4 – Defuzzification
3 – Fuzzy Inference
Crisp inputs Crisp outputs
Activated fuzzy rules
Membership function Aggregated fuzzy function
Figure 3.12: Schematic diagram of a fuzzy logic system
101
Essentially, a complex set of data can be modelled using ANN which is ideal for estab-lishing patterns in such systems.
Figure 3.13: Artificial Neuronal Network Architecture
An Artificial Neural Network (ANN) is a numerical mapping between inputs and outputsthat is modelled on the networks of neurons in biological systems [Kos92]. ANN is usedessentially in non-linear regression but also in non-linear classification problems. The modelpredicts one attribute (the output) using a set of input attributes and a non-linear func-tion. It supports only numerical inputs and symbolic or numerical output depending on theproblem.
The basic architecture (see Fig. 3.13) of an ANN encompasses the following componentswithin a hierarchical structure:
• Input layer: receives signals from the environment
• Output layer: expresses signals to the environment
• One or more hidden layers: retain some input and output signals within the network
Each layer has a number of simple processing elements called neurons. The structure ofone neuron is shown in Fig. 3.14. Signal paths with multiplicative weights w interconnectsthe neurons. A neuron receives its inputs (x) either from the outside of the network or fromthe other neurons. Each neuron computes its output (z) by its activation function and sendsthis as input to other neurons or as the final output from the system.
Figure 3.14: Neuronal structure
102
Commonly in the ANN, the signal flow is only in the forward direction from one layerto the next, from the input to the output. The ANN are train by a supervised learningalgorithm called back propagation. Back propagation uses gradient descent technique toadjust the weight (w) and biases (b) of the ANN in a backwards, layer by layer manner.It adjusts the weights and biases until the vector of the neural network outputs for thecorresponding vectors of training inputs approaches the required vector of training outputsin a minimum root mean square error sense [Lam03b]. The neurons in the input and outputlayer usually have simple linear activating functions that add all weighted inputs and add theassociated biases to produce their output signals. The neurons in the hidden layer usuallyhave non-linear activate functions with sigmoidal or hyperbolic tangent forms.
Neuron j with bias bj and n inputs each with signal xij and weight wij will have a linearlycombined activation signal yj as shown in Eq. 3.3.
zj =n∑
i=1
Wijxi (3.3)
A linear input or output neuron would just have this zj as its output. The most commonnon-linear hidden layer transfer functions use the hyperbolic tangent function as shown inEq. 3.4. This equation provides continuous, differentiable non-linear transfer functions withsigmoid shape.
yj = tanh(zj) =(ezj − e−zj)
(ezj + e−zj)(3.4)
Strong points and drawbacks The main strength of ANN is its universal approximationcapability. One of the most important characteristics of ANN is that it can learn from theirtraining experience. Learning provides an adaptive capability that can extract non-linearparametric relationships from the input and the output vectors.
It is probably the most accurate Data Mining (DM) method among the available data-driven prediction techniques. Unfortunately, from the point of view of interpretability it isperceived as a "black box". This is not satisfactory if customers require a detailed list of thereasons and assumptions behind the cost estimate [RK03]. It is heavy in terms of CPU timeduring the training stage and may become cumbersome for highly dimensioned input spaces.That is why it is advised to use it in conjunction with other methods that first reduce theinput space, like decision (or regression) trees or dendrograms.
Neural networks require a large learning database (DB) in order to be effective, whichwould not suit industries that produce limited product ranges. In addition, the DB mustcomprise similar products, and new products need to be of a similar nature, in order forthe cost estimate to be effective. Thus, neural networks cannot cope easily with novelty orinnovation.
The last issue involves a trade-off between over training and under training. Optimumtraining will capture the essential information in the training data without being overlysensitive to noise.
3.4.2.7 Simulation method (SM)
Since we cannot expect new knowledge of the measure on the heads of series, the produc-tion simulation is an interesting technique in shipbuilding (see section 2.3.5.8). Production
103
simulation or Virtual Manufacturing (VM) enables the modelling and simulation of produc-tion systems and processes to ensure, before the start of production, that they operate atpeak efficiency. Simulation is a key new technology of the millennium with considerableexpected growth rates per year [Ste03, Bai09].
Principle and process Production simulation is the process of designing a model of a realor imagined product and conducting experiments with that model. The purpose of simulationexperiments is to understand the behaviour of the product and to evaluate strategies forthe production/operation of the product. Assumptions are made about this system andmathematical algorithms and relationships are derived to describe these assumptions – thisconstitutes a "model" that can reveal how the system works.
The complexity of the system often cannot be represented by a simple mathematicalmodel. In this case, the behaviour of the system must be estimated with a simulation. Exactrepresentation is rarely possible in a model, constraining us to approximations concerninga degree of accuracy that is acceptable for the purposes of the study. In each one of theseenvironments, a model of the system has proved to be more cost-effective, less dangerous,faster, or more practical than experimenting with the real system.
Simulation can be used in very different fields with a lot of different methods. The mostcommon method used to model the production and fabrication of a product is called DiscreteEvent Simulation (DES). DES only takes points in time (events) into consideration. Suchevents may, for example, be a part entering a station or leaving it, or moving on to anothermachine. Any movements in between have little interest for the simulation itself. It is onlyimportant that the entrance and the exit events are displayed correctly. When a part entersa material flow object, the software calculates the time until it exits that object. Finally, thesimulation software makes a list of all the important events where each event is programmedand executed step by step. The total time needed to run a simulation is really faster and isonly limited by the computer speed.
Nowadays, DES tools like Plant Simulation, Arena or CATIA with DELMIA helps ship-yards to increase efficiency and workshop productivity, and to give computer-supportedanswers to the major questions: when and where to produce what and with which resourcesdepending on the availability and restrictions of resources and materials.
DES programs allows the mobilization of virtual plants like shipyards where product datacontains all geometrical and methodical information about the ship while the simulationmodel includes all parameters describing the production facilities, resources (machines, hu-mans, etc.) and processes. One of the major advantages of the production simulation is thatit is possible to integrate the operating rules of each workshop and simulate the complexinteractions between the different actors (human and material resources, transportation,machinery and tools, etc.). The production simulation is particularly effective to tackle phe-nomena such as the surface management, transport management, flow management (iden-tification of bottlenecks), management of failures and hazards, etc. that a simple analyticworkload simulation cannot integrate.
The cost assessment of a product starting from a simulation model is a quite easy task.Indeed, all individual process times of the manufacturing tasks are a result of the simulation
104
and linked to various resources. To assess the cost of the process, we can just multiply theoperating time of each resource by its dedicated cost rate (e/hour).
Strong points and drawbacks The use of simulation-based design and virtual realitytechnologies facilitates higher efficiency in terms of work strategy planning, and offers, as aresult, significant productivity gains.
Several strong points can be underlined:
• We can easily vary the parameters of a system such as product features, operating pro-cedures, decision rules, information flows, organizational procedures and then evaluatethe impacts of these changes.
• Hypotheses about how or why certain phenomena occur can be tested for feasibilityso that the errors are made on the model instead of in reality.
• Simulation models often have a visual Graphical User Interface (GUI), sometimes withgraphic animations. This makes them more reliable and understandable for the usersand managers.
Nevertheless, some limitations must be highlighted here [Smi99]:
• Development of a simulation model requires time and very significant resources.
• Model building requires special training. It is an art that is learned over time andthrough experience. The ability to create a model that accurately represents the realityto be simulated is not immediately apparent. Real systems are extremely complex andsome assumptions and omissions must be made about the details that will be capturedin the model.
• Another limitation is the availability of data to describe the behaviour of the system.It is common for a model to require input data that is unavailable.
3.4.3 Survey
In order to get a better understanding of how the shipbuilding industry and researchersforesee the difficulties of cost assessment, we have implemented an on-line survey enti-tled:"Survey about life cycle cost management". The purpose of this survey was to determinewhat are the main methods and tools used to evaluate/assess/control costs during the lifecycle of a ship (design, manufacturing, operation and retirement). Then we tried to identifythe major requirements of this sector as well as the critical points. The survey form andassociated detailed results are available in appendix B.
The twenty questions were gathered in 6 groups on the same topic in order to make thequestionnaire easier to answer. The survey was structured as follows:
• Introduction presenting the objectives of the survey.
• 3 questions about the company activity, company location and industrial sector.
• 2 questions about Concurrent Engineering (CE) methods and tools.
• 9 questions about cost evaluation methods and tools.
• 6 optional individual questions.
105
The survey was sent to 1250 people in the international maritime sector (shipyards, shipowners and research centres) from about 500 different companies. Only 2% (25) of respon-dents sent a reply. Possible causes are:
• Objective causes – the company follows a confidentiality policy, the person doesn’thave time, etc.
• Subjective causes – the person does not have the information, the person does not feelconcerned by the survey, etc.
Subsequently, this investigation is unfortunately not a representative survey of inter-national industrial requirements regarding life cycle cost assessment. A wider distributionmight have increased the sample size, but the reliability of the results would probably beenreduced. The main outcome of the survey is a guide and new ideas for the development ofmethods and tools within the present and future PhD studies.
The main interesting results are6 :
• The target population is mainly researchers or universities (48%), within the navalsector (72%), in Europe (68%)
• Both design for safety (48%) and design for production/manufacturing (44%) are Con-current Engineering (CE) tools already implemented within the companies surveyed
• Both design for life cycle (55%) and design for safety (28%) are CE tools that thesurveyed companies consider as the most promising for the future
• Both intuitive method (55%) and feature based costing (32%) are the most used costassessment methods during the concept design stage (before contract)
• Both parametric method (35%) and intuitive method (32%) are the most used costassessment methods during the basic design stage (after contract)
• Both feature based costing (32%) and intuitive method (28%) are the most used costassessment methods during the manufacturing of the ship
• Both feature based costing (28%) and case based reasoning (20%) are the most usedcost assessment methods during operation of the ship
• (Multi-) linear regression method (40%) is the most used tool to extract Cost Estimat-ing Relationship (CER)
• Only 20% of companies surveyed use commercial costing software
• 55% of companies surveyed said that their cost assessment methods is moderatelyeffective or totally ineffective
• The four best qualities of a good cost assessment method are considered to be:
– Good ability to reflect design changes (40%)
– Ease of use (36%)
– Reusable (36%)
6People may select more than one check-box, so percentages may add up to more than 100%
106
– Quick computation time (32%), high accuracy of results (32%), good logic visi-bility (32%)
• 16% of companies surveyed said that they failed to answer some questions for confi-dentiality reasons
3.4.4 Selection of cost assessment method
In the design phase, not all the information is available at the moment of the economicevaluation, and evaluation speed is a key element (the main design activity is not to performcost assessment).
Depending on the stage of analysis, the level of expected detail, the extent of availableinformation, different cost modelling techniques can be employed for the cost estimation atthe design stage [IH06, DC99]. These methods cannot be used during the whole life cycle(see Tab. 3.3 and Fig. 3.16). Some methods are better than others depending on thecontext and design maturity (see Tab. 3.2 and Fig. 3.15). When data are available, allthe methods could be used. But different estimation methods provide different projectionsof the anticipated costs. The projected differences in cost could have a significant impacton the overall viability of a project or the selection of the optimum design for a product orprocess.
Methods Advantages Drawbacks and limitations
IM – Quick to produce – Susceptible to bias
Intuitive – Flexible – Unstructured
– Different experts use different mechanisms
CBR – Can offer a solution rapidly – Need a reliable case base
Case – Very good logic visibility – Doesn’t handle innovative solutions
– Avoid previously committed errors
– Stores the knowledge of the company
PM – Makes clear the influence of parameters on cost – Parameters not included may be important
Parametric – Repeatable and objective – Simplistic
– Quick to produce – Logic not visible (black box)
FBC – Enables integration of CAD/CAM with cost information – Requires large resources to implement
Feature – Could be automated – No consensus on what features are
– Clear link between design choice and cost
FLM – Very good logic visibility – Need of human expertise and know-how
Fuzy – Integration of the imperfection of the model (fuzzy sets)
NNM – Accurate estimates possible – Logic not visible (black box)
Neural – Can be updated and retrained (adaptive capability) – Complex
– Requires a large and reliable historical database
– Doesn’t handle innovative solutions
SM – Good logic visibility (GUI) – Requires time and very important resources
Simulation – Can easily vary the product and – Model building requires special training
organizational parameters & experience
Table 3.2: Advantages and limitations of cost assessment methods
107
3.4.5 Multi-criteria decision analysis
In the majority of practical decision problems there is no alternative that fits perfectlyall the criteria. In fact, each alternative offers both strengths and weaknesses, which mustbe counterbalanced. Therefore, Multi-Criteria Analysis (MCA), also called Multiple Crite-ria Decision Making (MCDM), approaches have been developed to support decision makingproblems, formalising the trade-offs between the alternatives and fostering the transparencyof the decision. Multi-criteria analysis is an especially important approach for the interpre-tation of the results of a comparative analysis of technological alternatives and for addressingthe relevance of the different parameters of interest. Although MCDM models have beenused in many applications in engineering science [CNT97, TSGR05], only a very few of thesemodels can be found in the field of the shipbuilding industry.
We have chosen the PROMETHEE (Preference Ranking Organization METHod for En-richment Evaluations) method in order to perform the MCDM of the cost assessment tech-niques. This method has been developped by [BVM88] and is one of the best outrankingmethods for multiple criteria problems. The method and its applications have been describedin more detail in annex A. The PROMETHEE method ranks the alternatives, once all theparameters and the values have been presented.
3.4.5.1 Definition of alternatives
The outcome of any decision making model depends on the information at its disposal andthe type of this information may vary according to the context in which one is operating,therefore it is useful for decision making models to consider all the information as a whole.In MCDM the decision procedure is normally carried out by choosing between differentelements that the decision maker has to examine and to assess using a set of criteria. Theseelements are called alternatives.
For this study, we have used all the alternatives presented in section 3.4.2: IntuitiveMethod (IM), the Case Based Reasoning (CBR), the Parametric Method (PM), Feature-Based Costing (FBC), the Fuzzy logic method (FLM), the Neural Networks Method (NNM)and the Simulation method (SM).
3.4.5.2 Definition of criterion
The criterion represents the tools which enable alternatives to be compared from a specificpoint of view. It must be remembered that the selection of criteria is of prime importancein the resolution of a given problem, meaning that it is vital to identify a coherent family ofcriteria. The number of criteria is heavily dependent on the availability of both quantitativeand qualitative information and data. Tab. 3.3 and 3.4 summarise the results of the previousanalysis of the different cost assessment methods considered in this study, following 17 qual-itative criteria. These criteria were gathered into 6 families (design applicability, accuracy,data needs, usability and cost) and are listed in Tab. 3.5.
A preference "level" function has been added for each criteria, based on 3 coefficients P , Qand S (Appendix Fig. A.1). The selection was made because the criterions have a qualitativeform. The unit column of Tab. 3.5 shows the different retain qualities for each alternatives.It was felt that:
108
• preference level function with a P value of 0.5 and Q value of 1.5 is appropriate for aqualitative scale with 2 or 3 different qualities,
• preference level function with a P value of 1.5 and Q value of 2.5 is appropriate for aqualitative scale with 4 or 5 different qualities.
Cost assessment methods
Transparent box
Detailed
SMFBCCBRFLM
Black box
IM
Statistical
PMNNM
Figure 3.15: Tree classification of cost methods
IM CBR PM FBC FLM NNM SM
Innovation • • • •
Trade Studies • • • ◦ • • ◦
Early Design Stage ◦ • • ◦ • • ◦
Basic Design Stage • • • • • •
Detailed Design Stage ◦ • ◦ • •
Production • ◦ ◦ •
Table 3.3: Cost assessment methods versus design stages (• applicable, ◦ hardly applicable)
BasicDesign
ProductionDesign Production
DetailedDesign
Concept
StrategicalDecision
TacticalDecision
OperationalDecision
DesignMaturity
ProjectDecision
CostAssessment
Methodology
IM
PM
FBC
ActualCost
Figure 3.16: Cost assessment methodology versus time line of the project
3.4.5.3 Definition of weights and scenarios
The results of multi-criteria analysis hinge on the weighting allocated and thresholds set.The weights express the importance of each criterion and obviously may deeply influencethe final outcome of the entire calculation procedure. For some authors, the problem of howto determine the weights to assign is still unresolved since the different outranking methodsdo not lay down any standard procedures or guidelines for determining them.
109
Methods Logic Reusable for Accuracy Ability to reflect Ability to reflect
Visibility other applications Production changes Design changes
IM No No Low Possible Possible
CBR Yes No Fair No Possible
PM No No Fair No Possible
FBC Yes No Very High Yes Yes
FLM Yes Yes High Possible Possible
NNM No Yes High Yes No
SM Yes Yes Very High Yes Yes
Methods Historical data Cost DB Development Computation Ease of Compatibility with
Need Size Cost Time Use Other IT software
IM No Low Low Quick Moderate Low
CBR Yes Some Large Moderate Moderate Possible
PM Yes Some Low Quick High Possible
FBC No Large VeryLarge Slow High High
FLM No Low Moderate Quick Moderate High
NNM Yes Large Moderate Quick Moderate High
SM No Large VeryLarge Moderate Low High
Table 3.4: Effectiveness of various types of cost assessment methods
Familly Criteria Min Preferences Units
Max Type P Q
Design applicability Trade studies Max Level 0.5 1.5 1:no;2:hard;3:applicable
Early design stage Max Level 0.5 1.5 1:no;2:hard;3:applicable
Basic design stage Max Level 0.5 1.5 1:no;2:hard;3:applicable
Detailed design stage Max Level 0.5 1.5 1:no;2:hard;3:applicable
Production Max Level 0.5 1.5 1:no;2:hard;3:applicable
Innovation Max Level 0.5 1.5 1:no;2:hard;3:applicable
Accuracy Accuracy Max Level 1.5 2.5 1:low;2:fair;3:high;4:veryhigh
Ability to reflect prod. changes Max Level 0.5 1.5 1:no;2:possible;3:yes
Ability to reflect design changes Max Level 0.5 1.5 1:no;2:possible;3:yes
Data needs Historical data need Min Level 0.5 1.5 1:no;2:yes
Cost DB size Min Level 0.5 1.5 1:low;2:some;3:large
Usability Logic visibility Max Level 0.5 1.5 1:no;2:yes
Reusable for other applications Max Level 0.5 1.5 1:no;2:yes
Ease of use Max Level 0.5 1.5 1:low;2:moderate;3:high
Compatibility with IT software Max Level 0.5 1.5 1:low;2:possible;3:high
Cost Development cost Min Level 1.5 2.5 1:low;2:moderate;3:large;4:veryl
Computation time Max Level 0.5 1.5 1:slow;2:moderate;3:quick
Table 3.5: Definition of preference functions
110
In this study, 5 scenarios with 5 different weight vectors were formulated to circumventthis problem (see Tab. 3.6):
1. The first scenario W1, representing the base-case, was calculated by placing the focuson equal weights to all family of criterion.
2. The second scenario W2 was calculated by placing the focus on cost (35%), usability(30%) and accuracy (16%).
3. The third scenario W3 was calculated by placing the focus on accuracy (35%), usability(30%) and design applicability (16%).
4. The fourth scenario W4 was calculated by placing the focus on design applicability(40%), cost (30%) and usability (20%).
5. In order to finalize the set of scenarios that should be employed while evaluating thecost assessment methods, suitable surveys were designed (see section 3.4.3 and annexB) and dispatched to a large number of shipyards, ship owners and research centres.Based on a majority opinion of question 14 of the survey, the weight to be adopted forthe evaluation of cost assessment methods in scenario W5 has been defined by focusingon usability (36%), accuracy (29%) and cost (18%).
Familly Criteria Weight distribution of scenarios
W1 W2 W3 W4 W5
Design applicability Trade studies 3% 20% 2% 9% 3% 15% 7% 40% 2% 9%
Early design stage 3% 2% 3% 7% 2%
Basic design stage 3% 2% 3% 7% 2%
Detailed design stage 3% 2% 3% 7% 2%
Production 3% 2% 3% 7% 2%
Innovation 3% 2% 3% 7% 2%
Accuracy Accuracy 7% 20% 8% 16% 20% 35% 3% 5% 11% 29%
Ability to reflect prod. changes 7% 4% 8% 1% 5%
Ability to reflect design changes 7% 4% 8% 1% 14%
Data needs Historical data need 10% 20% 5% 10% 5% 10% 3% 5% 3% 8%
Cost DB size 10% 5% 5% 3% 5%
Usability Logic visibility 5% 20% 10% 30% 10% 30% 6% 20% 11% 36%
Reusable for other applications 5% 5% 5% 4% 12%
Ease of use 5% 10% 10% 6% 11%
Compatibility with IT software 5% 5% 5% 4% 3%
Cost Development cost 10% 20% 20% 35% 7% 10% 20% 30% 6% 18%
Computation time 10% 15% 3% 10% 12%
Table 3.6: Definition of scenarios
3.4.5.4 Results
Fig. 3.17 presents the results of multi-criteria decision analysis regarding preferences (pos-itive outranking flow φ+, negative outranking flow φ− and global outranking flow φ, seesection A.3 appendix A) of the various alternatives expressed numerically. The higher theglobal outranking flow the better alternative. The small φ− flow for the alternative FLM
111
indicates that is has a strong performance on most criteria, whereas the small φ+ flow ofalternative CBR is a sign that this alternative is weak in most attribution values whateverthe scenario. This result is confirmed by the spider diagram of the net flows of each criterion(see Fig. 3.18 which shows that FLM is the strongest alternative (maximization of the spidersurface) and CBR is the weakest alternative (minimization of the spider surface). Hence,also a change of the weight of the different criteria will show the FLM as the outstandingalternative (see Fig. 3.17(f)). However, this assessment technique absolutely requires theexistence of human expertise and know-how. Thus, we can not use it in all applicationcases. Furthermore, the selection of the preference function does not influence the rankingin its first position. This could be demonstrated by a sensitivity analysis of all possiblecombinations of preference functions within PROMETHEE.
3.4.5.5 GAIA visualisation
We also perform a GAIA visualisation which provides a graphical representation of thevarious alternatives for different criteria and a π decision axis (see annex A) in which directionthe best alternative is located according to the weight distribution. The GAIA plane isobtained by a projection of the information in the criteria space on a plane. The best planeis obtained by the Principal Components Analysis (PCA) technique. Through this projectionsome information is lost but most of the information is preserved. In the present case thepreserved information amounts to δ = 86%.
The GAIA plane given in Fig. 3.19 clearly confirms the previous results. Indeed, we canobserve the following characteristics:
• Required data, cost and accuracy are more discriminating than usability and designapplicability
• Design applicability and usability express a similar preference
• Cost and accuracy express a conflicting preference
• Cost and accuracy are independent regarding data needs
• FBC and SM are strong for accuracy, usability and design application but weak forcost
• IM is strong for data needs and cost but weak for design applicability and usability
In order to study the behaviour of the decision model, we implemented different scenarioswith different weights. For all weight distributions, the n decision vector remains orientedtowards the same sector of the diagram. Such variation in weights can easily be handledand visualized on the GAIA plane. It can be noticed that the alternatives FLM and FBCare still the best choice whatever the scenario.
3.4.5.6 Sensitivity analysis
Sensitivity analysis is carried out to study the subjective weights assigned to the criteria.The results are shown in Fig. 3.20 where the weight stability intervals for each scenarioare presented. The weight stability intervals give for each criterion family the limits withinwhich its weight can be modified without changing the complete ranking φ. The stabilityintervals are valid only when a single weight is modified at a time and all the other weightsdon’t change.
112
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
CBR PM IM NNM FBC SM FLM
phi+ phi- phi
(a) W1
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
CBR FBC SM IM NNM PM FLM
phi+ phi- phi
(b) W2
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
IM CBR PM NNM SM FLM FBC
phi+ phi- phi
(c) W3
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
CBR IM FBC SM NNM PM FLM
phi+ phi- phi
(d) W4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
CBR IM NNM PM FBC SM FLM
phi+ phi- phi
(e) W5 (f) All
Figure 3.17: Aggregated outranking flows of the alternatives
113
(a) IM (b) CBR
(c) PM (d) FBC
(e) FLM (f) NNM
(g) SM
Figure 3.18: Spider representation of ranking matrix for each alternative
114
Figure 3.19: Gaia view of criterion, alternatives and scenarios (δ = 86%; N:alternatives;•:scenarios; �:criterion)
Figure 3.20: Weight sensitivity analysis for each scenario
115
3.4.6 Conclusions about the selection of a cost estimation method
This study has highlighted the complex nature of decision making (PROMETHEE) in-volved in the selection of the cost assessment methods for the shipbuilding industry. Theanalysis identifies the best cost assessment methods and particularly for uncertain decisionmaking environments (see scenarios). The sensitivity analysis reveals the relative robustnessof the different scenarios.
Multi-criteria analysis, as this section demonstrates, can provide a technical-scientificdecision making support tool that is able to justify its choices clearly and consistently,especially in the shipbuilding sector.
3.5 Conclusions
In this chapter the methodology has been decided on. The design method will be a designfor X, where the priority is given to the right "X" depending on the design phase. Costassessment methods are available and must be chosen accordingly to give to the designs theright measurements supporting their decisions.
116