principles for modelling of manufacturing sequences861208/fulltext01.pdf · principles for...
TRANSCRIPT
Principles for modelling of manufacturing
sequences
MATS WERKE
DOCTORAL THESIS
STOCKHOLM, SWEDEN 2015
I
Abstract
The manufacturing sequence influence, to a large extent, component properties like fatigue
life, shape accuracy and manufacturability. By simulating the manufacturing sequence, using
numerical or empirical models, and extracting important accumulated data, like residual
stress, hardness and shape, the possibilities of early analysis of a design concept and the
associated manufacturing sequence will increase. An established methodology has the
potential of reducing physical testing and the time and costs of product design and process
planning.
This thesis proposes an algorithm to be used for setting up a framework of interconnected
process step models. With support from the algorithm, it is possible to extract a virtual
simulation sequence from a physical manufacturing sequence. Thereby, you can replicate the
aggregated effects of process steps on part key features and manufacturing features. The
algorithm will serve as a tool in process planning when establishing virtual manufacturing
sequences. The virtual sequences should be used for virtual prediction of component
properties, optimization of process parameters and evaluation of the effects of replacing,
removing or adding process steps to a manufacturing sequence
The algorithm is based on stepwise upstream selection of process steps, definition of
interconnected models and selection of interconnected datasets using breadth first search. The
algorithm completes existing procedures for data mapping and exchange of data between
models into an overall approach for establishing virtual manufacturing sequences. Other
scientific contributions are methods for modelling of deep rolling and blasting, a model
material for validation of rolling and forging simulation and principles for integration of
process simulation with CAD/CAM.
II
Acknowledgements
This work was accomplished with support from a number of people. First, I would like to
address my gratitude to my supervisor Professor Mihai Nicolescu and my co-advisor,
Professor Torsten Kjellberg for discussions and consultations.
Further, I would like to thank all the researchers, who have contributed to this work by means
of discussions, co-authoring of papers and reports and active work in the case studies. I
especially would like to thank Dr Mats Bagge for his contributions in the development of the
upstream metamodelling algorithm in Paper V, Lars Johanson, Swerea IVF for development
of the C6t2 utility, Dr Sven Haglund, Swerea KIMAB, for fatigue discussions and finite
element analyses etc in the steering arm case in Paper II, Dr Mikael Hedlind, for his
contribution concerning the ISO 10303 standard in Paper VI, Thorbjörn Hansén, Swerea
MEFOS, for the collaborative work in Paper IV. I also would like to thank Mikael Ström,
Swerea IVF, for inspiring discussions concerning research methodology.
My gratitude also goes to representatives of Scania, Volvo Trucks, Bharat Forge and Parker
Hannifin for contribution with time and resources in various research projects by means of
simulation, forging tests, material data development, residual stress measurements, fatigue
testing, co-authoring of papers and reports, discussions and knowledge contribution.
The work was made possible by financial aid from Vinnova, Sweden´s Innovation Agency,
which is gratefully acknowledged.
Finally I would like to thank my family Birgitta, Fredrik and Sofia for supporting me in all
thinkable ways and for putting up with my writing during countless weekends and vacations.
III
Content 1 Introduction ......................................................................................................................... 1
2 Appended papers ................................................................................................................. 3
3 Contributions to appended papers ....................................................................................... 4
4 Referred publications .......................................................................................................... 5
5 Definitions and abbreviations ............................................................................................. 6
6 Motivation ........................................................................................................................... 7
7 Research methodology ...................................................................................................... 10
8 Challenges ......................................................................................................................... 12
8.1 Selection of process steps to be included in the virtual MS ...................................... 12
8.2 Definition of interconnected models and datasets ..................................................... 12
8.3 Map output data on proceeding models ..................................................................... 13
8.4 Exchange of data between process step models ........................................................ 14
8.5 Validation .................................................................................................................. 14
8.6 Implementation of sequential simulation in product development ........................... 15
9 Research questions ............................................................................................................ 16
10 Summary of appended papers ...................................................................................... 17
Paper I .................................................................................................................................. 17
Paper II ................................................................................................................................. 19
Paper III ................................................................................................................................ 20
Paper IV ................................................................................................................................ 21
Paper V ................................................................................................................................. 22
Paper VI ................................................................................................................................ 24
Paper VII .............................................................................................................................. 25
11 Synthesis of research .................................................................................................... 27
12 Research results ............................................................................................................ 28
13 Scientific contributions and future research ................................................................. 29
14 References .................................................................................................................... 30
15 Appendix ...................................................................................................................... 33
1
1 Introduction
The development of the manufacturing industry relies on research into manufacturing
processes and materials and the development of new products. Global competition forces
manufacturing companies to continuously optimise process sequence performance and
efficiency in order to assure product quality and competitiveness. Manufacturing sequence
(MS) optimization refers to at least two manufacturing processes in consideration of one or
more target criteria. The objective of process chain design is a target-oriented arrangement of
adequate processes depending on certain component characteristics as defined by the design
specifications, to enable highly efficient manufacturing. The purpose of modelling and
simulation of manufacturing sequences is to improve efficiency in product design and process
planning. This will lead to the economic manufacturing of high-quality products with the
greatest possible productivity.
This thesis introduces a systematic representation methodology to be used in order to set up a
set of interconnected numerical or empirical process step models for simulation of aggregated
features from the MS. This metamodelling methodology includes procedures for selection of
process steps to be included in the virtual sequence, definition of interconnected models and
datasets, data mapping and data exchange between models. Furthermore, procedures for
implementation of MS simulation in practice and validation of plastic deformation are also
presented, see Figure 1.
Fig. 1 Procedures for metamodelling of manufacturing sequences.
The research motivation is described in chapter 6 whereas the research methodology is
described in chapter 7. The challenges are clarified in chapter 8 and the research questions in
chapter 9 are based on the identified gaps between existing and desired situations in chapter 8.
Chapter 10 summarizes the enclosed publications and the research is synthesised in chapter
11. Chapter 12 concludes the research and the scientific contributions are described in chapter
13 together with suggestions for future research.
An algorithm for metamodelling of manufacturing sequences is proposed. The algorithm is
based on stepwise upstream selection of process steps, definition of interconnected models
and selection of interconnected datasets. The algorithm supports framework modelling and is
2
applicable for both numerical and empirical models. With support from the algorithm, it is
possible to extract a virtual simulation sequence from a physical manufacturing sequence.
Thereby, you can replicate the aggregated effects of process steps on part key features and
manufacturing features. The algorithm will serve as a tool in process planning when
establishing virtual manufacturing sequences. The virtual sequences should be used for
optimization of process parameters and evaluation of the effects of replacing, removing, or
adding process steps to a manufacturing sequence. The algorithm will also serve as a tool in
strength and shape analysis of components.
Other scientific contributions are modelling methods for deep rolling and blasting, a new
model material for validation of plastic deformation during rolling and forging and an
approach for implementation of MS simulation in practice.
The research is based on experiences from case studies of forged components using a steering
knuckle, steering arms, a bevel gear pinion, a crown wheel and a crankshaft and a casted
cylinder block.
3
2 Appended papers
Paper I Werke M, Moshfegh R, Johanesson B, Svenningstorp J, Johansson P, Stinstra E
(2005) Statistical methods for extracting and evaluating manufacturing
parameters of significant importance to component robustness, Proceeding of
QMOD conference, Palermo
Paper II Werke M, Kristoffersen H, Haglund S, Svensson L Nord A (2008) Predicting
residual stresses and hardness of a critical component using a combination of
numerical and empirical methods, Article in Research Steel International special
edition, Krakow
Paper III Werke M (2008) Methods and models for shot peening simulation, Proceedings of
SPS08, Stockholm
Paper IV Hansen T, Werke M (2012) Modelling and model validation of the deformation of
macro inclusions during hot deformation, Proceedings of 8th International
Conference on Clean Steel, Budapest
Paper V Werke M, Bagge M, Nicolescu M, Lindberg B (2015) Process modelling using
upstream analysis of manufacturing sequences, Article in International Journal of
Advanced Manufacturing Technology, DOI 10.1007/s00170-015-7076-4
Paper VI Werke M, Hedlind M, Nicolescu M (2014) Geometric distortion analysis using
CAD/CAM based manufacturing simulation, Proceedings to SPS14, Gothenburg
Paper VII Werke M, Nicolescu M, Semere D, (2015) Variation analysis of manufacturing
sequences, to be published
4
3 Contributions to appended papers
Paper I M Werke contributed to the development of the overall methodology and the deep
rolling model. E Stinstra contributed with statistical methods, P Johansson
contributed with the VMEA methodology.
Paper II M Werke contributed to the modelling methodology, using a combination of
numerical and empirical models and to the fatigue analysis. S Haglund
contributed with the methods for measurement of the surface roughness, fatigue
analysis, and load analysis of the test case using FEM.
Paper III M Werke contributed to the complete paper including the literature survey.
Paper IV M Werke contributed to the analysis of plastic deformation of the steering arm
and crankshaft. T Hansen contributed to the development of the model material
and P Ottosson, Swerea IVF, contributed to the simulation of the deformation of
inclusions during rolling.
Paper V M Werke contributed to the upstream metamodelling methodology. M Bagge
contributed with the development of the algorithm.
Paper VI M Werke contributed to the methodology and global modelling and simulation of
machining. M Hedlind contributed with knowledge concerning STEP.
Paper VII M Werke contributed to the overall MS simulation methodology. M Nicolescu and
D Semere contributed to the Monte Carlo simulations and FEA.
5
4 Referred publications
The author has also contributed to other publications with major influence on the research.
The most important publications are enlisted below
[10] Werke M, Kristoffersen H, Andersson J, Johansson K, Thoors H (2009) Virtual
verification of the shape accuracy of a powertrain component. Research report
Prosim project
[13] Werke M (2009) Simulation of manufacturing sequences for verification of
product properties. Licentiate thesis, KTH.
[18] Werke M (2003) Internet based distribution of simulation results, EVEN
conference, Trinity College, R. I. Campbell, N. O. Balc, Loughborough, pp 4 – 13
[22] Werke M, Gotte A, Sibeck L (2014) Castforging for production of components
with tailored geometry and strength, Research report, Vinnova
[28] Werke M, Ström M (2013) Konceptkonstruktion – Framtid och
forskningsmetoder. Uppdragsrapport Vinnova
[33] Wärmefjord K, Söderberg R, Ottosson P, Werke M, Lorin S, Lindkvist L,
Wandebäck F (2013) Prediction of geometrical variation of forged and stamped
parts for assembly variation simulation, Proceedings of International Deep
Drawing Research Group Conference IDDRG2013, Zurich
[34] Werke M (2009) Blasting influence of residual stresses and hardness – An
investigation with support from measurements, Swerea IVF project report 09/09
6
5 Definitions and abbreviations
CAD/CAM: Computer aided design/Computer aided manufacturing
CFD: Computer fluid dynamics
DACE: Design and analysis of computer experiments
DoE: Design of experiments
DRM: Design research methodology
DS: Descriptive study
FE/FEM/FEA: Finite element/Finite element method/Finite element analysis
FMEA: Failure mode and effect analysis
LHD: Latin hypercube design
Metamodel: A sequence or framework of interconnected process step models.
MS: Manufacturing sequence
Validation: Judgement of the validity of a model for prediction of required data, using
measurements.
OEM: Original equipment manufacturer
PS: Prescriptive study
RC: Research clarification
STEP: Standard for exchange of product model data
UNV: Ideas universal file format
VMEA: Variation mode and effect analysis
VRML: Virtual reality modelling language
VTF: View tech file format
XRD: X-ray diffraction
7
6 Motivation
Forged components like shafts, gears, knuckles and steering arms are characterized by high
strength and robust properties. The components are available in many applications such as
final drives and gearboxes in vehicles, see Figure 2, handheld tools, agricultural and forestry
machines, but also in power plants, process and mining, aviation and marine and defence
equipment. These types of components, whose function is to transfer force and torque, are
manufactured in steel and in traditional process steps such as forging, machining, heat
treatment and mechanical surface treatment.
Fig. 2 Transmission components in final drive and gearbox.
A process step is here defined as an activity that creates a shape (e.g. forging, casting, powder
methods and rapid prototyping), modifies a shape (e.g. machining, heat treatment), joins parts
(e.g. adhesives, welding, fasteners) and finalises a product (e.g. polishing, coating, painting
and mechanical surface treatment), see Ashby [1].
Component properties like strength, geometric distortions and manufacturability are not solely
dependent on single process steps but are sequentially aggregated through the MS. This is
illustrated in the example in Figure 3 where fatigue strength is improved through a sequence
of process steps.
Fig. 3 Possible improvement of fatigue- and tensile strength through a MS [2].
8
In fact the complete chain, including the melting, solidification and forming processes
influences the final product features. This is schematically illustrated with an example where
a bevel gear pinion is manufactured through several process steps, see Figure 4 and Paper V.
Fig. 4 Manufacturing sequence for a bevel gear pinion.
The melting and solidification processes introduce crystallographic structure, voids,
inclusions and segregations, i.e. irregularities, in the material. Those material properties are
then modified during the proceeding rolling and forging processes where voids are closed,
inclusions deformed and recrystallization occurs. Anisotropy will evolve and influence fatigue
strength, see Temmel [3], and geometric distortions, see Brinksmeier [4]. The heating and
cooling processes will introduce residual stresses and the material removal during the
proceeding turning and gear cutting processes will release residual stresses and introduce
residual deformations, see Paper VI. Thus, in order to predict properties the complete
sequence should be taken into consideration and a vision is to establish a sequence of process
step models and simulate the MS. A process step model is here defined as “a representation
of a process step intended to enhance our ability to understand, predict, or control its
behaviour”, whereas simulation is defined as “the imitation of the behaviour of the process”
see Kaizer [5].
Today process simulation is used in practice in order to optimise individual process steps like
casting, rolling, forging, heat treatment, machining and mechanical surface treatment. FEM is
the most established method, see Gur et al. [6] and foundries use the simulation technique in
order to study how variations in metal fluid flow and cooling conditions influence the
solidification process as well as residual stresses and shape after casting, see Vijayaram et al.
[7]. Milling and forging shops utilise FE simulation in order to study how the rolling machine
or forging press kinematics, the tool and the temperature of the workpiece influence the shape
of the workpiece and the material filling in the tools, see Hartley et al. [8]. FE simulation of
machining, like drilling, turning and milling, may be used in order to study how the cutting
tool geometry, machining parameters and fixture design influence the workpiece shape,
temperature and residual stresses, see Mackerle [9]. Empirical modelling is also possible e.g.
Werke et al. suggest in [10] a polynomial model based on residual stress measurements where
9
the residual stresses are predicted based on feed rate and cutting depth. With heat treatment
simulation it is possible to study how variations in temperature, cooling time and cooling rate
influence the residual stresses, hardness and phase distribution, see Mackerle [11]. FE
simulation of mechanical surface treatment like e.g. deep rolling and shot peening may be
used in order to study how variations in process parameters and material influence surface
residual stresses and deformations, see Paper III.
The common denominator using FEM is the discretization of continuum problems into finite
elements that are connected with nodes and thus the model data are always expressed in terms
of nodal and elemental data. These types of data also constitute the output results from the
simulations and it is thereby possible to use output data from one process step simulation as
input data to the proceeding process steps, see Figure 5.
Fig. 5 Process step simulation data flow, including process history.
This fact introduces possibilities to connect process step simulations and predict aggregated
effects from the MS on the final or intermediate component features. Sequential simulation
has a potential to improve load- and shape analysis of components, improve optimisation of
processes through the MS and evaluation of the effects of replacing, removing or adding
process steps to a MS, see Figure 6. The motivation to develop and implement a methodology
is thereby high and will reduce development time and costs of product design and process
planning.
Fig. 6 Evaluation of alternative manufacturing routes using MS simulation.
10
7 Research methodology
The research is based on Design Research Methodology (DRM), which is described by
Blessing et al. in [12]. The methodology consists of Research Clarification (RC), Descriptive
Study (DS) and Prescriptive Study (PS), see Figure 7.
Fig. 7 Design Research Methodology according to Blessing [14].
The research is clarified based on the challenges with MS simulations according to chapter 8.
The gap between the existing and desired situation for each challenge is identified and the
research questions in chapter 9 are extracted from the gap analysis.
In the descriptive study empirical data is assembled and analysed in order to derive a better
understanding of the problem. The knowledge development is performed with support from
case studies developed in the research projects according to table 1. The results are presented
in the enclosed publications I – VII and in the licentiate thesis [13]. The descriptive study is
summarized in chapter 10.
In the prescriptive phase, which is presented in chapter 11, the developed knowledge is
synthesized into principles and algorithms for MS simulation. The results are evaluated in
descriptive study II where conclusions are drawn and potential future research is identified.
The descriptive study II is presented in chapter 12 and 13.
11
Table 1: Summary of projects, case studies and partners involved in the research.
Phase Publication Case study Projects Industrial partners Research organisations
1 I Steering
knuckle
IMS-EURobust
(EU)
Volvo Trucks Swerea IVF, Chalmers
1 II Steering
arm,
Pinion
ProSim (Vinnova) Volvo Trucks,
Scania, Bharat
Forge, Parker
Hannifin
Swerea IVF, Swerea
KIMAB
1 III Steering
arm Blasting (Vinnova)
Bharat Forge Swerea IVF
1 [13] Pinion ModArt (Vinnova) Scania, Sandvik
Coromant
Swerea IVF, KTH/IIP
2 IV Crankshaft Clean Steel
(Swerea)
Bharat Forge Swerea MEFOS, Swerea
IVF
2 V Pinion,
Steering
arm
Upstream
simulation
(Vinnova)
Volvo Trucks, GKN Swerea IVF
2 V Pinion,
Crown
wheel
ReVer (Vinnova) Scania Swerea IVF, Chalmers
KTH/IIP, Swerea KIMAB
2 VI Crown
wheel
FBOP (Vinnova) Scania KTH/IIP, Swerea IVF
2 VII Cylinder
block
Scania KTH/IIP, Swerea IVF
12
8 Challenges
The modelling and simulation of complete manufacturing sequences is a challenge which may
lead to unrealistic and time consuming modelling efforts and extensive computational
requirements. This is due to the often complex material transformations through several
consecutive process steps. Simplifications are therefore required and a systematic approach is
needed. Since the simulation of MS is limited in practice there is also the need for methods
that facilitate industrial adaption of sequential simulation. The challenges are discussed
below.
8.1 Selection of process steps to be included in the virtual MS
In order to establish a simplified virtual sequence there is a need to reduce the number of
process steps to be included in the virtual sequence, i.e. a need for a method for selecting the
first, intermediate and final process steps to be included in the virtual sequence
The selection of the first process step may be a difficult task due to the complex material
transformations that, for example, a transmission component is exposed to through a long
sequence of process steps. In order to avoid this problem, one possibility is to include the
complete physical sequence in the virtual chain. This may however lead to unnecessarily long
virtual sequences. The difficulties concerning selection of intermediate process steps may
vary. If, for instance an intermediate process step is painting it is obvious that it will not
influence e.g. the final distortions and it should be excluded from the virtual sequence. On the
other hand, e.g. a heat treatment should be included. The final process step may be the last
process that influences the part key features such as strength, geometry and dimensional
tolerances. The final process step may also be an intermediate process after which
manufacturing features, such as: machinability (e.g. hardness and grain size), hardenability
(e.g. phase composition) and forgeability (e.g. crack tendencies during plastic forming) are of
interest for the study.
The selection of process steps may be performed with support from experiments and e.g.
Brinksmeyer et al. [4] use physical experiments according to Design of Experiments (DoE) in
order to analyze the influence on geometric distortions from different process steps like
casting, forming, cutting and case hardening on discs. The experimental approach is however
time- and cost consuming and there is a need for more efficient guidance when selecting
process steps. No systematic approach for selecting process steps has been found in literature.
This thesis discusses 3 major approaches, namely selection of the most important single
process step, selection of a subset of process steps downstream through the sequence and
selection of process steps upstream through the sequence.
8.2 Definition of interconnected models and datasets
The description of generic methods for definition of interconnected models and selection of
interconnected datasets is limited in literature. Govik et al. describe in [14] a special case for
simulation of forming, assembly and spotwelding of stamped parts where simplifications of
interconnected models and simulations are described. The model simplifications are here
based on knowledge and no systematic method is used. Denkena et al. propose a methodology
13
where the interdependencies between processes and interconnected datasets are stored in
“Technological interfaces”. The authors demonstrate the use of technological interfaces in
conjunction to the manufacturing of helical car-gearings in [15] and in conjunction to the
manufacturing of pinion shafts in [16], including precision forging, hardening, hard turning
and grinding. The publications describe how to store datasets but do not present how to select
the datasets to be included in the technological interfaces. This thesis discusses two major
approaches namely downstream modelling and upstream modelling of the MS.
8.3 Map output data on proceeding models
Different process steps require various, mesh topologies and element types depending on the
physical phenomena that the models are intended to reflect, see examples in Figure 8.
Fig. 8 Different models and mesh topologies for different processes.
In the case hardening model the mesh has high density in surface regions in order to reflect
various carbon contents during decarburisation, see lic. thesis [13]. In deep rolling or
shotpeening simulation it may be sufficient with a local model of a small substrate of the
material near the surface, see Paper I and Paper III. For process steps that lead to large
plastic deformations, like e.g. forging, a complete global model that reflects the global
geometry is required. Further, for these types of process steps there is a need for automatic
remeshing during the execution in order to solve problems with large element distortions.
This may lead to a high density mesh in areas exposed to large plastic deformation and lower
density in regions with lower deformations. The mesh may here differ from the mesh that is
created with support from a CAD model to be used in a structural load analysis, see Paper II.
Thus when simulating a sequence of process steps the mesh and geometry requirement may
vary between the process step models concerning model geometry and mesh topology. The
requirement of the element type may also vary for different process steps concerning element
shape (brick, tetrahedron, 2D or 3D etc) and number of integrations points. Mapping
methodologies exists and e.g. Afazov et al. [17] present a computer based algorithm for
mapping of data from an output mesh onto a new input mesh. The algorithm is however
14
limited to mapping of data between models developed in Abaqus, Ansys, MSC Marc, Deform
and Morfeo.
8.4 Exchange of data between process step models
It is possible to exchange data between a majority of the most common simulation softwares
in native ASCII-formats or in generic formats according to standards like VRML-, VTF- and
UNV- formats or STEP. VRML is an open standard for visualisation of 3D models on the
Internet. It is possible to exchange the surface mesh and display various results with colour
representations, see Werke [18], but the possibilities for exchange of other types of simulation
data is limited. The VTF format was originally developed by the company Ceetron (former
ViewTech) and is a standard file format for post-processor applications like Glview Inova and
Glview Express (used by e.g. the simulation software Forge). The UNV format was originally
developed by the Structural Dynamics Research Corporation (SDRC) in order to facilitate
data transfer between CAD, computer aided test (CAT) and CAE. These representations are
generic but the implementations in practice are limited. Another possibility is to utilize the
generic ISO 10303 (STEP) standard for data communication. STEP includes all aspects
concerning data transfer between different engineering disciplines including CAD, CAM and
CAE. STEP contains several applications protocols (AP:s) dedicated to specific engineering
tasks. AP 209 (Composite and metallic structural analysis and related design) is especially
suitable for FEA and has possibilities to communicate both CAD models and FE models, see
[19]. One disadvantage is that the standard is limited to linear FEA.
Another option is to reformat output data into input data to the next simulation code using
dedicated reformatting codes. Afazov presents in [20] a data translation and mapping software
with the ability to communicate data between simulation software like Abaqus, Ansys,
MSC.Marc, Deform and Morfeo. A similar system is presented by Åström [21] and Werke et
al. describe in [22] how to reformat output mesh and porosity data from casting simulation in
ProCast to input data for forging simulation in Deform. The data exchange is performed using
a tailored data translation software. The purpose is here to simulate castforging which is an
innovative manufacturing method where pre-casted components are forged into final shape in
order to increase the ductility and fatigue strength. One disadvantage with the described
approaches is however that CAD/CAM data is not included in the data exchange.
8.5 Validation
Modelling and simulation of process steps are connected with risks for errors. The reasons for
simulation errors may originate from insufficient accuracy in material- and process data,
inconsistency in material models and mesh topology. The assembly of several process step
models into a sequence of models will also imply risks for inaccuracies connected with
mapping- and data exchange due to inconsistencies between different material models. Since
errors from one process step simulation will aggregate as input to proceeding process step
simulations it is important to validate the virtual sequence. Validation is here defined as “The
process of determining the degree to which a model is an accurate representation of the real
world from the perspective of the intended uses of the model”, see Obercampf et al. [23].
Validation provides evidence of how accurately the computational model simulates the real
15
world for system responses of interest. This is performed by the assessment of the model
accuracy by comparison with experimental data and the decision of model accuracy for the
intended use.
The possibilities to validate macrostructural models, with support from measurements, are
well described in literature. Obercampf et al. [24] propose a generic validation methodology
that decomposes and simplifies complex physical phenomena into four levels; system level,
subsystem level, benchmark level and unit problem level and where each level is validated
using measurements. This is demonstrated by Ahlberg [25] through the validation of welding
and heat treatment simulation in a sequence. The description of how to validate
microstructural models is however limited in literature. Milesi et al. [26] demonstrate
possibilities to simulate the deformation of inclusions during forging using microstructural
models. The paper does however not explain how to validate the simulations.
8.6 Implementation of sequential simulation in product
development
Virtual methods invoking CAD models and tools for finite element analysis of component
strength and stiffness provide well established support for efficient product development and
are compulsory in a majority of the mechanical design offices when designing and optimising
components. It is a well-known fact that the manufacturing processes have a large influence
on product properties like fatigue strength, shape accuracy and surface quality. The influence
of product properties induced by the manufacturing processes is not yet included in the virtual
product development methodology and the design analysis often starts with virgin material
data. This means that the engineering development includes several loops of prototyping and
physical testing in order to ensure that the required properties are reached. This uncertainty
also results in a conservative dimensioning philosophy, unnecessarily heavy and over-
dimensioned components and thereby unnecessarily high safety factors. A virtual design
philosophy invoking the effects on component properties from the manufacturing sequence
may have good potential to increase efficiency in the product development process and
increase the quality of the component and manufacturing processes. Product development, i.e.
“Total design” according to Pugh [27], includes conceptual and detail design activities, see
Figure 9.
Fig. 9 Illustration of product development according to Pugh [29].
In the conceptual design phase the design engineer makes drafts of various concepts of the
component that are evaluated against each other, see Werke et al. [28]. FEA is then
conducted, based on the CAD models, where the stresses and deformations due to the applied
loads are predicted. If the stress- or deformation levels are too high the designs are modified
and new feasibility tests are performed using FEA. One or several concepts are selected for
further development in the detail design phase. In this phase the weak spots in the design are
16
eliminated, the concept is refined and the final decision concerning geometry is made. A
virtual design philosophy invoking the effects on component properties from the
manufacturing sequence may have good potential to increase the accuracy in the design and
thereby minimize the need for costly prototypes and physical tests, see Figure 10.
Conceptual
designSpec.
CAD
modelFEA
Design OK?
Poc. plan OK?Detail
designTestPrototype Test OK? Manufacture
Process
planExisting situation
Desired situation
Yes
No
Yes
No
Conceptual
designSpec.
CAD
modelFEA
Design OK?
Poc. plan OK?
Detail
designManufacture
Process
plan
Yes
No
MS
simulation
Fig. 10 Illustration of existing (top) and desired (bottom) situation in virtual product
development.
Process planning activities may also be supported with MS simulation for procedures like
selection of process steps, to be included in the MS and optimisation of process parameters.
The challenge is to integrate sequential simulation with established CAD tools in order to
improve efficiency in both geometrical design and process planning.
Cunningham [29] describes a module within the Deform simulation software to be used by
process planners in order to select and optimise process steps. The module consists of
workpiece and process attributes. The workpiece object attributes include geometry, which
can be supplied from a 2D CAD-file, material data and mesh density. The process attributes
include process parameters, tooling and fixtures etc. and by assembling these objects a
sequence of operations can be built up and simulated, beginning with workpieces in their
initial state which are then subjected to successive processes, each of which determine the
new attributes of the workpiece. However, no systematic approach concerning how to
integrate simulation of MS with CAD has been found.
9 Research questions
Based on the previous discussion it is concluded that there are gaps between the existing and
desired situation concerning principles for metamodelling, i.e. principles for selection of
process steps, definition of interconnected models and selection of datasets. Other gaps have
been identified concerning validation of inclusion models. It is also concluded that the use of
sequential simulation is limited and in order to implement MS modelling and simulation in
practice it is beneficial to adapt modelling and simulation methodology to existing tools for
product development and process. The gaps between the existing and the desired situation are
summarized in Figure 11. Thus, in order to establish a complete MS modelling and simulation
17
methodology the existing principles for dataexchange and mapping should be complemented
with metamodelling principles and validation techniques.
Fig. 11 Gaps between existing and desired situations concerning modelling of MS.
The research questions are formulated as:
1. “How to define a simplified metamodel?”
2. “How to validate modelling of inclusions and segregations?”
3. “How to integrate MS simulation with CAD?”
10 Summary of appended papers
Paper I
A method for simulation on how the MS influence component robustness is proposed. The
method is a combination of engineering knowledge, statistical methods and simulation.
The first step is to select the most important process step using FMEA, see Lundberg [30].
The second step is to select the most important process parameters from the selected process
step using VMEA. VMEA is a method for analysis of variations and their effect on properties,
see Johansson et al. [31]. As a third step a process model is defined and variations in material
and process parameters are simulated. The strategy is illustrated in Figure 12.
18
Fig. 12 Single process step selection and modelling method.
The method is demonstrated on a component, a forged steering knuckle, which is an
important component in the steering mechanism of a truck. The main steps in the
manufacturing sequence are described in Figure 13.
Fig. 13 Steering knuckle with manufacturing sequence.
The deep rolling process is selected as the most important process step using FMEA. Deep
rolling is used in order to strengthen the component, for example in zones exposed to high
loads and the steering knuckle is exposed to deep rolling in a critical fillet radius. A spherical
ball is pressed against the surface to be treated, see Figure 14. The process introduces
compressive residual stresses, cold deformation and reduces microcracks.
Fig. 14 Deep rolling process.
With support from VMEA the most influential material- and process parameters on fatigue
strength are selected, i.e. spindle velocity, feed rate and hydraulic pressure, Young’s modulus,
yield strength and tensile strength. Based on this information a FE model of deep rolling is
developed together with a plan for computer simulation experiments with several test points
between the low and high values. A computer based test plan is developed with support from
Latin Hypercube Design, see Stehouwer et al. [32]. Computer based test plans according to
19
DoE are also possible to use, see Wärmefjord et al. [33]. The effect on variations in surface
residual stresses from variations in process- and material parameters is predicted and the
component robustness is evaluated.
Paper II
A method for prediction of the influence from the MS on fatigue life is proposed. It is based
on downstream modelling and simulation of a subset of the MS. The method is demonstrated
on a micro alloyed steering arm, an essential component in the steering mechanism of a truck.
The main steps in the manufacturing sequence are described in Figure 15.
Fig. 15 Downstream simulation of steering arm.
The subset is selected, based on knowledge and the modelling starts with a mesh of the input
bar. Hot shearing, hot rolling, roughing blow, finishing blow and controlled cooling are
simulated in a sequence using a forging simulation code. The output mesh, describing the
geometry after the final forging step, including nodal temperatures from the forging
simulations, is used as input to a cooling simulation using thermo-metallurgical-mechanical
simulation. During blasting the components are bombarded with small steel balls in order to
remove slag from the surface. Other effects are increased compressive residual stresses and
cold work deformation, factors that increase fatigue strength. Based on a literature study in
Paper III it is concluded that the existing FE models for shot peening do not manage to
describe the complex blasting process where decarburized parts are tumbling against each
other at the same time as they are exposed to blasting. An approach with a combination of
empirical and numerical models for blasting is therefore proposed. The hardness and residual
stresses are predicted using measurements and a new method for analysing the influence of
the blasted surface texture on the stress concentration factor using a combination of
measurements and FEA is proposed, see Figure 16.
Fig. 16 Surface topology measurement with a confocal microscope and surface stress
concentration factor from blasting grooves analysis using a FE model.
20
The paper concludes that the blasting should be modelled using a combination of empirical
and numerical models. The knowledge is further developed by Werke [34]. Here
measurements of residual stresses and hardness before and after blasting during various
blasting times are systematically stored in a database. It is concluded that the surface
conditions that decarburisation effects on stresses and strains should be involved in the
empirical models.
Paper III
A literature survey is presented where 10 shot peening models are evaluated. Shot peening
improves fatigue strength with support from bombardment of the surface with small steel
balls, see Figure 17. Shot peening increases the surface strength through the introduction of
compressive stresses, increased hardness through cold work deformation and reduction of
microcracks.
Fig. 17 Shot peening process and schematic illustration of residual stress curve.
The results from a shot peening simulation may give indications of the residual stress- and
plastic strain profile below the surface as well as indentation depth. Significant for the
presented FE models is the fact that the studied geometric target volume represents a local
substrate below the surface. The state of the art concerning FE models for shot peening is still
on a basic level with one ball shot or a pattern of impacts against a substrate of an ideal
material. The investigated numerical models may be used effectively in qualitative studies of
how e.g. variations in shot velocity, hardness, shape and size of shot influence surface
residual stresses. However, the described models are not sufficient to describe the effects from
the machine, like the tumbling of pieces against each other at the same time as the shot
peening. Further, the physical surface conditions may be far from ideal including e.g.
decarburisation effects induced from previous forging and heat treatment processes. Thus in
order to perform quantitative studies concerning the effect of shot peening or blasting on
residual stresses and hardness a parametric model based on measurements is an alternative
approach, see Figure 18.
21
Fig. 18 Parametric shot peening model approach.
Paper IV
A new physical model material for validation of plastic deformation after rolling and forging
simulation is proposed. The model material consists of a combination of two steel grades with
slightly different composition. With support from Hot Isostatic Pressing (HIP) the model
material is shaped as bars with quadratic shaped cross section, see Figure 19.
Fig. 19 Model material shaped as a round bar.
The usability is demonstrated on rolled bars, a forged crankshaft and a forged steering arm.
The model material is here complemented with engrafted inclusions, in order to demonstrate
how to validate deformations of inclusions, and a centre round steel wire (4 mm) material in
order to validate the simulation of deformations on the kernel material. The model material is
rolled and forged and the output workpieces are visually inspected and compared to
simulations. The agreements between the position of the wire in the model material and
simulation results are good, see Figure 20.
22
Fig. 20 Comparison between simulation and etched measurements of plastic deformations.
Paper V
A new method for metamodelling of the MS is proposed. The method is based on upstream
selection of process steps, definition of interconnected models and selection of datasets. The
paper compares the advantages and disadvantages of upstream and downstream
metamodelling, see Figure 21.
Fig. 21 Comparison of downstream (left) and upstream (right) metamodelling.
In the downstream approach the first task is to analyse the first process step. If it is concluded
that that it will influence final results a model is defined. The output data, which may be used
as input data to proceeding models, are then used as guidelines for proceeding process step
selection and model definition and the approach is repeated stepwise forward until the
interconnected models for the final process step are defined.
23
In the upstream approach the user starts with analysing the last process step and defines a
model of this process step. The model input data that carries a history from previous processes
is then detected and defines the required output data from previous process step models. If the
previous process step influences the required output data, it is included in the virtual chain
and a model is defined, otherwise it is excluded. The process step selection and modelling
actions are repeated stepwise upstream. If no historical input data are needed for a defined
model the first process step in the metamodel is defined and the complete virtual sequence is
established.
Thus the upstream approach has better possibilities to reduce the number of process steps to
be included. Further, in the upstream approach the definition of models is guided by the
requested key features, which are stepwise extracted to specifications of output data needed
for previous models upstream through the sequence. This guidance from key features is
missing in the downstream approach.
When there is a need for several models to represent one process step the sequence of models
will be expanded into a framework of interconnected models. Two search algorithms, to be
used in order to navigate in the framework modelling procedure, are evaluated namely depth
first search and breadth first search, see Cormen et al. [35] and Figure 22. When using depth
first search, one branch is modelled to the end before modelling a new branch. When using
breadth first search, each process step is analysed and modelled completely before moving
stepwise upstream or downstream to the next process steps.
Fig. 22 Depth first search (top) and breadth first search (bottom) navigation.
In a depth first search scenario, all the previous process steps in a branch are analysed and
modelled before jumping back and modelling a new branch. This fragmentation in modelling
is avoided using breadth first search and the paper concludes that breadth first search is more
24
efficient as compared to depth first search. Based on those conclusions an algorithm for
metamodeling of manufacturing sequences is proposed. The algorithm is based on stepwise
upstream selection of process steps and definition of interconnected models using breadth first
search. The algorithm is enclosed in the paper in Scilab code and is demonstrated with
support from two test cases: a bevel gear pinion and steering arm, see Figure 23.
Fig. 23 Crown wheel (left), bevel gear pinion (middle) and steering arm (right) for heavy
trucks.
Paper VI
A CAD/CAM and FEM based methodology is proposed to be used in order to control
geometric distortions after machining. The method is demonstrated in the process planning of
a crown wheel according to the above Figure 23.
The component is designed in a CAD system and intermediate machining workpiece models
are developed. The CAD model of the component is used in order to create CAD models of
the forging dies and the billet to be used for process simulation of forging and cooling in a
sequence. The output from the cooling simulation is used as input to the proceeding
machining simulation. The models of the fixtures, machining features and the input raw piece
for machining are assembled in the CAD/CAM system and translated to machining
simulation. Several steps of virtual machining are then performed where the output residual
deformations and residual stresses from one machining setup are used as input to the
proceeding machining operation. The process parameters, material data and component
geometry are adjusted until the level of residual deformations are accepted. The data
exchange between CAD/CAM and simulations is performed using the STEP AP209. The
activities are illustrated in Figure 24.
25
Fig. 24 CAD/Cam based methodology.
Paper VII
In order to eliminate the sources of variations, at the process planning stage, it is important to
consider these variations in the developed models.
This paper investigates the dimensional variation propagation and the interactions among
machining operations in a process sequence. The main reason of variation propagation in
machining processes is that the previously machined features are used as the datum in the
subsequent machining operations. The method presented in this paper simulates the MS and
evaluate the effect on the geometry of machined part in terms of set-up and machining
deviations. The computational model estimates the dimensional characteristics of the
machined component by combining Monte Carlo technique with FEA. The first set-up
variation considered is caused by the orientation of the workpiece on the machine tool table.
The second set-up variation is due to elastic deformation of the fixture. Therefore,
dimensional variations are introduced at each process step by the setup and machine errors.
The deflection after face milling of the head face and the milling of the cylinder bores on a
cylinder block are simulated, see Figure 25.
26
Fig. 25 MS simulation concept.
Further, a method for prediction of the effects from variations in surface roughness on stress
variations is presented, using a combination of Monte Carlo simulation and FEA. A
population of 5000 different surfaces was used to simulate the contact line between a cylinder
and an engine piston. The simulation was performed by superimposing several random
sinusoidal curves of different wavelengths and phases. The wavelength, amplitude, and phase
of the sinusoids used to create this population were generated from normal and uniform
distributions. The contact stress was then computed by FE method, see Fig. 26.
Fig. 26 The simulation of the surface roughness on the contact line between cylinder and
piston (left) and the stress distribution on the contact line (right).
27
11 Synthesis of research
A flowchart which illustrates how the knowledge and modelling principles evolve through the
publications is presented in Figure 27.
Fig. 27 Relations between the appended publications.
a. In Paper I a method for single process step selection and modelling is proposed using
FMEA and VMEA. The approach has limitations and in order to derive an overview
of how the manufacturing sequence influence key features several process steps must
be included in the virtual MS. The methodology is therefore expanded into a
modelling and simulation of a subset of the process steps. A downstream modelling
approach is presented in Paper II.
b. The obtained knowledge concerning deep rolling and shot peening simulation,
presented in Paper I and Paper III, lead to the methodology for blasting simulation in
Paper II using a combination of numerical and empirical models, i.e. a framework
modelling approach.
c. The experiences from Paper II concerning downstream modelling, framework
modelling and the utilisation of both numerical and empirical models lead to the
suggestion of stepwise upstream framework modelling in Paper V and the proposed
algorithm in Scilab code.
28
d. A new model material presented in Paper IV, to be used as reference data when
validating modelling and simulation of inclusions and segregations.
e. In the licentiate thesis [13] different possibilities to communicate data between
models are discussed and an algorithm for translation and mapping of data, C6t2, for
tailored data exchange is presented. The C6t2 utility is written in C++ and runs at a
command prompt or from a script file. The utility is a complement to previous
described work in [17], [20] and [21]. During the development of the C6t2 algorithm
several pitfalls concerning mapping emerged. Several issues varied in the input and
output formats like node and integration point ordering, distribution of integration
points, results described in either a local or global coordinate system etc. The utility
also has the ability to transform a sheet metal forming mesh into a welding mesh with
solid elements and map output data from the shell elements to the solid elements, see
Figure 28.
Fig. 28 Data transfer functionality in C6t2 (left). Transformation of shell elements to
solid elements and mapping of stresses (Von Mises) on the new solid mesh (right).
f. In the licentiate thesis [13] STEP AP209 is identified as a suitable standard for
communication of data such as mesh, stresses and deformations in connection to
product development and process planning in practice, using CAD/CAM. This
knowledge was used in the development of the CAD/CAM based method for virtual
process planning that is proposed in Paper VI.
g. The simulation of variations in one process step in Paper I is expanded to simulation
of variations through a sequence of process steps in Paper VII.
12 Research results
The research results are summarized below:
“How to define a simplified metamodel?”
The stepwise upstream metamodelling approach is more effective, as compared to the
downstream approach in the efforts to establish a simplified metamodel. It is also concluded
that the breadth first search procedure is advantageous, as compared to the depth first search,
when establishing a framework of models stepwise upstream.
“How to validate modelling of inclusions and segregations?”
A new model material for validation of plastic deformations on both macro- and microscale
during rolling and forging is proposed.
29
“How to integrate MS simulation with CAD?”
STEP AP209 is sufficient in process simulation for communication of datasets such as mesh,
stresses, strains and deformations in connection to communication of CAD models. It has
been shown that the standard may be sufficient for limited MS applications such as prediction
of global machining distortions due to clamping.
13 Scientific contributions and future research
Metamodelling algorithm: A new approach with an algorithm for upstream metamodelling of
manufacturing sequences is proposed. The algorithm is feasible for the establishment of a
framework of models that contains both numerical and empirical models. The algorithm
completes existing methods and algorithms for mapping of data and exchange of data into an
overall set of principles for modelling manufacturing sequences. The upstream metamodelling
approach may be used in conjunction with methods proposed by Bagge [36] where the
process planner defines processes to be included in the sequence by asking: “What is the
required final outcome of the sequence?”, after which upstream selection of process steps is
conducted as shown in Figure 29.
Fig. 29 Process planning methodology as compared to manufacturing sequence.
Validation of models for inclusions: The new model material for validation of deformations in
microstructure modelling is a scientific contribution.
Mechanical surface treatment: Scientific contributions are a new model for deep rolling,
taking the surface roughness prior to deep rolling into consideration and blasting simulation,
using a combination of numerical and empirical models. Blasting is however a complex
process where the workpieces are tumbled against each other and at the same time exposed to
blasting. The surface decarburisation during forging and cooling will also have a major effect
on the surface residual stresses before and after blasting. Those effects are not fully
investigated, and this is recommended for future research.
Implementation in product development: The scientific contribution is the proposed
methodology for adaption of MS simulation in practice using CAD/CAM and STEP. The
development of STEP for comprehensive exchange of nonlinear simulation data together with
CAD is a subject for future research.
30
14 References
[1] Ashby M.F (2005) Materials selection in Mechanical design. Elsevier, Third
edition, Oxford
[2] SMUTAB (1998) Smide för hög utmattningshållfasthet. IVF-skrift 98813,
Gothenburg, Sweden
[3] Temmel C (2007) Fatigue anisotropy in forged components. PhD thesis,
Chalmers, Gothenburg, Sweden.
[4] Brinksmeier E,Lubben T, Fritsching U, Cui C,Rentsch R, Sölter J (2011)
Distortion minimization of disks for gear manufacture. International Journal of
Machine Tools & Manufacture 51:331-338
[5] Kaizer J.S, Heller A. K, Oberkampf W.L (2015) Scientific computer simulation
review. Reliability Engineering and system safety, 138:210 – 218
doi:10.1016/j.ress.2015.01.020
[6] Gur C.H, Pan J (2009) Handbook of Thermal Process Modelling of Steels. CRC
Press, London
[7] Vijayaram T.R, Sulaiman S, Hamouda A.M.S, Ahmad M.H.M (2006) Numerical
simulation of casting solidification in permanent metallic molds. J Mater Process
Technol 178:29-33. doi:10.1016/j.jmatprotec.2005.09.025
[8] Hartley P, Pillinger I (2006) Numerical simulation of the forging process. Comput
Methods Appl Mech Engrg 195(48-49):6676:66766
doi:10.1016/j.cma.2005.03.013
[9] Mackerle J (1999) Finite Element analysis and simulation of machining: a
bibliography (1976 – 1996). J Mater Process Technol 86(1-3):171-3)
doi:10.1016/S0924-0136(98)00227-1
[10] Werke M, Kristoffersen H, Andersson J, Johansson K, Thoors H (2009) Virtual
verification of the shape accuracy of a powertrain component. Research report
Prosim project
[11] Mackerle J (2003) Finite element analysis and simulation of quenching and other
heat treatment processes, a bibliography (1976-2001). Comput Mater Sci 27:313-
332, doi:10.1016/S0927-0256(03)00038-7
[12] Blessing L, Chakrabarti A (2009) DRM, a Design Research Methodology.
Springer, London, UK
[13] Werke M (2009) Simulation of manufacturing sequences for verification of
product properties. Licentiate thesis, KTH Royal Institute of Technology
31
[14] Govik A, Nilsson L, Moshfegh R (2012) Finite element simulation of the
manufacturing process chain of a sheet metal assembly. Journal of Materials
Processing Technology 212:1453-1462, doi:10.1016/j.jmatprotec.2012.02.012
[15] Denkena B, Rudzio H, Brandes A (2006) Methodology for Dimensioning
Technological Interfaces of Manufacturing Process Chains. CIRP Ann-Manuf
Technol 55:497-500. doi:10.1016/S0007-8506(07)60467-3
[16] Denkena B, Henjes J, Henning H (2011) Simulation-based dimensioning of
manufacturing chains. CIRP Journal of Manufacturing Science and Technology
4:9-14. doi:10.1016/j.cirpj.2011.06.015
[17] Afazov SM (2013) Modelling and simulation of manufacturing process chains.
CIRP Journal of Manufacturing Science and Technology 6(1):70–77.
doi:10.1016/j.cirpj.2012.10.005
[18] M. Werke (2003) Internet based distribution of simulation results. EVEN
conference, Trinity College, R. I. Campbell, N. O. Balc, Loughborough, pp 4 – 13
[19] International Organisation for Standardisation (2001) Application protocol -
Composite and metallic structural analysis, ISO 10303-224
[20] S.M. Afazov , A.A. Becker, T.H. Hyde (2012) Development of a Finite Element
Data Exchange System for chain simulation of manufacturing processes.
Advances in Engineering Software 47:104-113
[21] Åström P (2004) Simulation of Manufacturing processes in Product
Development. Thesis, Luleå University of Technology
[22] Werke M, Gotte A, Sibeck L (2014) Castforging for production of components
with tailored geometry and strength. Research report, Vinnova
[23] Oberkampf W.L, Trucano T.G (2008) Verification and validation benchmarks,
Nuclear Engineering and design. 238:716 – 743,
doi:10.1016/j.nucengdes.2007.02.032
[24] Oberkampf W.L, Trucano T (2002) Verification and validation in computational
fluid dynamics. Progress in Aerospace Sciences 38:209–272.
[25] Ahblerg H (2005) Simulation of welding and heat treatment. Thesis, Luleå
University of Technology
[26] Milesi M, Chastel Y, Bernacki M, Loge R, Bouchard P.O (2007) Explicit
microscopic fatigue life analysis of forged components. Computer methods in
materials science 7:383-388
[27] Pugh S (1996) Creating innovative products using Total Design. Addison-Wesly
Publishing Company Inc.
32
[28] Werke M, Ström M (2013) Konceptkonstruktion – Framtid och
forskningsmetoder. Uppdragsrapport Vinnova
[29] Cunningham J (2010) Process modeling improve aero engine design.
www.eurekamagazine.co.uk
[30] Lundberg M (1991) Feleffektanalys (FMEA) av smidesprocessen. IVF-skrift
91815, IVF Mölndal
[31] Johansson P et al. (2005) Using VMEA to facilitate the preparation of robust
design experimentation. Palermo, QMOD
[32] Stehouwer, Hertog (1999) Simulation-based design optimization: methodology
and applications. Proceedings of the first ASMO UK/ISSMO Conference in
Engineering Design Optimization, Ilkley, UK
[33] Wärmefjord K, Söderberg R, Ottosson P, Werke M, Lorin S, Lindkvist L,
Wandebäck F (2013) Prediction of geometrical variation of forged and stamped
parts for assembly variation simulation. Proceedings of International Deep
Drawing Research Group Conference IDDRG2013, Zurich
[34] Werke M (2009) Blasting influence of residual stresses and hardnessess an
investigation based on measurements. Swerea IVF project report 09/09
[35] Cormen T.H, Leiserson CE, Rivest R.L, Stein C (2001) Introduction to
Algorithms. 2nd edn. MIT Press, Cambridge
[36] Bagge M (2009) An approach for systematic process planning of gear
transmission parts. Licentiate thesis, KTH Royal Institute of Technology
33
15 Appendix
Paper I
Statistical methods for extracting and evaluating manufacturing parameters of
significant importance to component robustness
Werke M, Moshfegh R, Johanesson B, Svenningstorp J, Johansson P, Stinstra E
Presented at QMOD conference, Palermo 2005
Paper II
Predicting residual stresses and hardness of a critical component using a combination of
numerical and empirical methods
Werke M, Kristoffersen H, Haglund S, Svensson L Nord A (2008)
Article in Research Steel International special edition, Krakow 2008
Paper III
Methods and models for shot peening simulation
Werke M
Presented at SPS8 conference, Stockholm 2008
Paper IV
Modelling and model validation of the deformation of macro inclusions during hot
deformation
Hansen T, Werke M
Presented at 8th
International Conference on Clean Steel, Budapest 2012
Paper V
Process modelling using upstream analysis of manufacturing sequences
Werke M, Bagge M, Nicolescu M, Lindberg B
Article in International Journal of Advanced Manufacturing Technology, 2015
Paper VI
Geometric distortion analysis using CAD/CAM based manufacturing simulation
Werke M, Hedlind M, Nicolescu M
Presented at SPS14 conference, Gothenburg 2014
Paper VII
Variation analysis of manufacturing sequences
Werke M, Nicolescu M, Semere D
To be published