principles of engineering design r. penny · postgraduate medicaljournal(june 1970) 46, 344-349....

Postgraduate Medical Journal (June 1970) 46, 344-349.

Principles of engineering design

R. K. PENNY

Professor of Engineering Design and Production, University of Liverpool

SummaryThe paper sets out procedures used in engineeringdesign by listing the various steps in a sequentialpattern. This pattern is not universally applicable andthe variants on it depend on the type of problem in-volved and the information available. Of criticalimportance is the way in which models-physical ormathematical-can be constructed and depending onthese, three design methods are described. These typesare illustrated by reference to a number of medicalaids which have been designed.

IntroductionEngineering design is concerned with problem

solving. While there has been much discussion aboutthe art versus the science of design it perhaps makessense to say that engineering design is concernedwith the art of using science in achieving a usefulend-product: 'art teaches us to do', 'science teachesus to know' (Webster's Dictionary being used here asan arbiter of the meaning of these words). Actually,there is little danger that all the art will be removedand that only science will be left; always in engineer-ing we shall be attempting more than we have theexact tools to help us to do, and always we shall bedealing with uncertainty.

Successful designers usually follow some sort ofmethodical approach to problem-solving. Thesemethods evolve wittingly, or unwittingly, over yearsof practice on a diversity of problems and a numberof attempts have been made to systematize a patternof events which might be classified as principles.These attempts have been successful to the extentthat a method is an improvement on intuitive andad hoc arrangements. Nevertheless the 'principles'should not be regarded as axiomatic for the verysimple reason that judgment is needed in makingdecisions in problems governed by vastly differentsets of variables for which there exist differingdegrees of lack of information.The purpose of this paper is to outline some of the

steps that take place in the design process and toillustrate the use of three basically different methodsof using them.

The nature of engineering and the role of designBefore detailing the steps in the process of design

it is worthwhile to discuss the nature of engineeringand the relationships of engineering with otheractivities. Fig. 1, taken from Dixon (1966), showshow in one dimension engineering spans from science

Politics

Sociology,psychology

Economics

Engineering Engineering EngineeringScience-- science design technology--Production

Industrialdesign

Artistic design

Art

FIG. 1. The central activity of engineering design (Dixon,1966).

to production with engineering design as the centralactivity. On the one hand the engineer has to under-stand basic physical laws and the mathematics oftheir expression, while on the other his function is toproduce to meet the needs of society. The designerwho is at the centre of this activity, has also to keephis eyes on the social and economic factors affectinghis problem as well as the aesthetic features. If hedesigns an ugly object which nobody wants or canafford he will not stay in business for long!

All engineering tasks involve the following steps:(a) Recognition and definition ofa need to be fulfilled.

Examples: The need for clean air is recognized bymost. But, how clean should it be (assuming we

on June 25, 2021 by guest. Protected by copyright.

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.46.536.344 on 1 June 1970. D

ownloaded from

http://pmj.bmj.com/

Principles of engineering design 345

can measure cleanliness), and at what point isthere to be a trade-off between cost in terms ofhuman misery as a result of polluted air and thesupply of money to provide clean air?The need for better transportation from airport

to city is becoming more obvious. What is thelikely usage over the next so many years, whatform of transport, from whence to where, and soon?These are all engineering questions and all ask

that a need be defined in quantitative terms.Answers based on intuition or bad market researchare likely to result in engineering failures and cer-tainly unplanned obsolescence (e.g. TSR 2).

(b) The design ofa system to meet the need. Design issubject to constraints involving time and moneyprincipally. Those of techniques and materials alsofigure prominently since these will certainly dic-tate the reliability of the product of the design.The engineer thus has to generate alternatives and,using the prevailing rule of his economy as a guide,must decide which alternative is 'best'. It is notsufficient to devise solutions and the distinctionbetween design and invention lies in the optimiza-tion process which is carried on after invention.

(c) Production. If a need has been demonstrated anddefined and a design has been proposed it is clearlyadvantageous to produce the device, system orprocess. This is done by organizing, directing andif necessary creating the human skills, machinesand materials needed for this task. There is clearlyanother problem in optimization involved in theProduction element.

(d) Action after production. Once the device has beenproduced, and proved ready to serve its need,provision has to be made to sell it, to maintain itsreliable functioning, to replace it or repair it whenmalfunction occurs for whatever reason andfinally to plan for and implement its disposal.If an engineering task is to be performed effectively

it must be carried through in stages but as if each ofthe stages (a-d) were part of an overall plan. Failureto treat these steps as a whole results in excessivedemands on time, for added funding, unnecessarydrains on resources, late delivery, dissatisfied cus-tomers and loss of market-a story with which weare all too familiar in post-war Britain!Although the last paragraphs represent only the

skeleton of a plan, inasmuch as each element of theplan will require many sub-elements each interactingwith the other, it is apparent that a methodology isemerging from the discussion. Although the sub-elements of each part must be considered in detailfor the greatest success this is only done in this paperfor the engineering design elements-the main topicof discussion. Even in this area a brief discussiononly is possible.

The engineering design processThe engineering design process deals with basic

operations that are repeated over and over againwith a view to seeking opportunities to turn ideasinto reality. This process is usually performed with aview to producing the item needed in the shortesttime for the highest reliability at the lowest cost;other optimizing criteria may, of course, prevail.

Fig. 2 shows, in chart form, the various steps inthe design process and how these interact with eachother; it is not meant to be comprehensive or applic-able to every problem. The diagram simply enables

Identification and formulation Specificationof problem

Utilization offexi ngrat-Md-l 1ninformation-

Anlysis and Optimzeevaluation

Construct ionand test

FIG. 2. Flow diagram of basic events in the designprocess.

one to see at what points in the process decisions aremost critical in affecting other steps. Certainly de-cisions have to be made everywhere on the routebetween formulation and solution of the problemand for this reason alone so-called automated designis a very remote possibility for the future. Flow dia-grams of the type given in Fig. 2 are of paramountimportance to the programming of problems suitablefor solution by digital computation and comparisonsbetween that form of problem solution and theproblem-solving involved in engineering design havemade the modern computer seem all-powerful. How-ever, Rosenstein's predictions (1967, Table 1) ofengineer/computer interaction in the design processmake it clear that human decision is likely to remaina vital ingredient. The usefulness of the computer asan aid to decision must no longer be overlookedhowever.The first decision to be made by the engineer is in

satisfying himself that a need exists and to under-stand the objectives of his task. At this stage aspecification can be attempted; a specificationbecause it may well become apparent when the


http://pmj.bm

j.com/

Postgrad M


ownloaded from

http://pmj.bmj.com/

346 R. K. Penny

TABLE 1. Engineer/computer interaction (from Rosenstein, 1967)Performed by

Operation Present Future

I Identification of need Engineer EngineerII Information collection, organization, storage, and retrieval Engineer Computer

+ Engineer initiationIII Identification of system variables Engineer Engineer

inputs-outputsIV Establish constraints and criteria Engineer Engineer

1. Constraints all factors which express limitations + Computer aid2. Value system and criterion function

V Synthesis of alternatives Engineer Computer-t Computer aid + Engineer aid

VI Modelling of parameters and variables Engineer ?

Parameters ModelsGeometry or topology DeterministicMaterial Lumped parametersEnergy Distributed parameterPeople Probabalistic

VII Analysis-including test, evaluation and prediction Engineer Computer-- Computer aid

VUI Decision steps. Engineer ComputerComparison of analysis of alternatives against desired results + Computer aid - Engineer aid

using decision rulesIX Optimization Engineer Computer

+ Computer aidX Communication, implementation and presentation. Engineer Computer

Paper work generation + Computer aid +- Engineer aidXI Iteration

analysis, construction and testing stages are reachedthat small changes in the specification could producegreat benefits to the final outcome. The need forflexibility to accommodate change at any time is ofparamount importance. This flexibility can be builtinto the procedure by gathering information aboutpresent designs-perhaps by studying productjournals and patents-or, where no previous designexists, to utilize new ideas all with a view to synthe-sizing alternatives and conducting preliminary feasi-bility studies. The feasibility study is one of the mostimportant steps and yet one that often receives littleattention-probably because of the formidabilityof the task. It attempts to decide on physical realiza-bility of a given specification while ensuring com-patability with the production and economicproblems as well as ensuring a pay-off in terms ofprofit for the customer. Clearly there will be anatmosphere of doubt after the feasibility study be-cause of the modest investment at that stage. Ifcertainty could be achieved with only modest invest-ment further steps obviously would be redundant.The feasibility study gives the best indications ofprobability of success or failure and of the likelyorder of further investment needed. A necessary pre-requisite in it involves the synthesis of alternativesand here the inventiveness or creative ability of thedesigner is called into play; factors affecting crea-

tivity have been studied on a broad front but theways of stimulating creativity amongst student andprofessional engineers have only started to be experi-mented with. Following the feasibility studiescriteria for choosing the most promising schemes forfurther study will have to be developed (anotherdecision). Concepts must be formed and modelsformulated. The models might be mathematical onesfor paper study or physical ones for laboratory test-ing but more usually a combination of both is used,being constructed with a view to analysing ideas andassessing basic concepts of geometry and materials.This stage utilizes the application of physicalprinciples and involves evaluating and optimizingthe results with respect to size, cost, weight and any-thing else. The performance of the design is judgedagainst the original specification and if the specifica-tion is met (it rarely will be, so that another decisionis needed) the engineer must then translate his solu-tion into production terms. If the specification is notmet, and it is decided to seek new alternatives that domeet it, the cycle of events can be repeated until anacceptable compromise on the specification isreached or the project is stopped.DecisionsThroughout this discussion of a design procedure

accent has been placed on the abilities to invent and to


http://pmj.bm

j.com/

Postgrad M


ownloaded from

http://pmj.bmj.com/


analyse. In addition the need for decisions to be madein passing through the various stages between prob-lem statement and solution is emphasized. A designermust make decisions based on compromises in-volving innumerable factors-economic, human,technical and so on-and his final decision is onewhich optimizes all of these considerations. Unfor-tunately there is a widely held belief that decision-making is an art-an art practised by those willing toguess but unwilling to investigate alternatives con-sidered rationally and simultaneously. There arethose occasions when the scientific theories ofdecision-making through probability, statistics andoptimization are ideally suited to the design methodbut a vital starting point is the quantification of allfeatures of the compromise. At other times non-quantifiable factors and judgments of values renderdecision-making difficult but by no means insuper-able. In fact mathematical methods can be used inthese circumstances again by listing alternatives andisolating the most important variables. Such schemeswere pioneered in many military applications inplanning strategies and these have been extended topolitics and business and have been given what isperhaps an appropriate name-game theory. Anotheraid to rational decision-making is through criticalpath methods which endeavour to determine theeffects of different factors on planning, schedulingand controlling a project.

Clearly within the confines of this brief paper it isnot possible to discuss the available methods ofdecision-making. However, the overriding guidelinethat might be given is always to list alternatives asclearly as possible. The alternative not to decide is,unfortunately, all too frequent!

Some examplesThe work of the Liverpool University Design

Study Unit has involved a number of studies ofmedical aids; some of these are listed in Table 2.Although these have not contained the vastness and

TABLE 2. Some medical aids.

DesignFunction feature Design type

Operating Surgery on Geometry Design bytable large synthesis

animals

Patient Diagnosis Electronic Productmonitor in humans logic evolution

circuitry

Bone joints Repair Mechanics Design byanalysis

Page turner Therapeutic Mechanism Productevolution

complexity of detail more frequent in large engineer-ing systems they present interesting examples ofdifferent design types serving completely differentfunctions and using different design features in theirsolution.For some devices it is difficult to establish mathe-

matical models for the purpose of analysis. Physicalmodels can be built upon ideas often from standardparts, the model tested experimentally and thenchanged to make a new model. The process isrepeated until a satisfactory approximation to theoriginal specification is met. This is often termedproduct evolution and the development of the patientmonitor and page turner, listed in Table 2, are goodexamples of this form of design. The design methoddiscussed earlier was followed basically but the con-ception of a device made of standard parts wasfairly straightforward from the beginning. The roleof experiment as an aid to design (one feature ofexperimentation which is badly underrated) was vitalin both of these examples.

Contrast between these two examples of evolu-tionary design are worth further discussion. Bothwere aimed, essentially, at simplifications of existingtrends. The patient monitor was intended to sense anumber of signals, to analyse them in prescribedcombination and to present the analysis in an audioand/or visual manner according to preset limits. Theprescription for the monitor was made by Stewart(1969) in order to draw the attention of clinicianstowards the possibilities of a simple and cheapportable device as opposed to complex and expensivesystems which often provide unnecessary data. Thisprescription was made possible by the selection ofwhat were considered to be the most importantpatient measurements from which followed therequired analysis of the critical combinations. Themanner of display was then a matter of choice forconvenience-in this case by minimization of dials,meters and recording devices to those that wereessential. The clear specification provided by Stewartmade the designer's task an almost routine one inproviding a workable solution. The monitor con-structed with relays has successfully survived initialtests from analogue signals simulating critical condi-tions and from signals generated from animals. It isnow in the refinement stage of solid state conversionand further trials. Some views of the early unit areshown in Fig. 3. The requirements for the pageturner were much simpler and yet less precise.Existing designs were few, the best not easily trans-portable and too expensive. A visit to a local ortho-paedic ward soon justified the need where the painfulattempts by limbless patients to perform what isnormally the everyday task of reading a book wereseen. The device to be designed was to accept areasonable range of book types, sizes, paper texture


http://pmj.bm

j.com/

Postgrad M


ownloaded from

http://pmj.bmj.com/

348 R. K. Penny

........

FIG. 3. Views of the early patient monitor.

'V'pulleysPerspex pressure pad rubber tyreddrive wheels

...|, "Cwerpowered drive shaft runningfull width of base board

linkage powl and ratchetFIG. 4. Page turner mechanism.

and thickness and, in order to be available to manypatients in the hospital ward or in his home, easilytransportable. The primary features of the finaldevice centred on the design of a simple mechanismwhich drove a pair of friction wheels in either of twodirections, Fig. 4, and which could be actuated by apressure pad operated by low force micro-switches.The book size requirement was easily overcome by asimple arrangement ofspring-loaded, parallel movingretaining bars. The whole contraption collapses intoa case measuring 17" x 15" x 4-"and weighing 5 lb.After initial set-up the page turner is easily operatedby any but the completely rigid.A somewhat more sophisticated form of design is

design by repeated analysis. In this a mathematicalmodel is constructed and from this the relevantdesign parameters can be extracted. By selecting setsof parameters, analysis of the design permits com-parison with a specification. Analysis of differentsets of parameters permits different design alterna-tives to be assessed confidently prior to constructionof what is thought to be the most suitable. This formof design procedure is ideally suited to solution bycomputer, when the tedium of repeated analysis inorder to seek optimum designs is much relieved.A good example of design by repeated analysis

from the selection in Table 2 is that of the bonerepair illustrated schematically in Fig. 5 (Osborne-

Load.~~

FIG. 5. Osborne-Ball Osteotomy Plate (patented*).


http://pmj.bm

j.com/

Postgrad M


ownloaded from

http://pmj.bmj.com/


Ball Osteotomy Plate, patented*). This device wasdesigned to provide continuity between leg bonefragments while preventing undue strain from anyleg movement. The problem is one in mechanics inwhich it is necessary to provide equilibrium betweenthe weight of the upper body and the resistance pro-vided by the leg and at the same time ensuring fixa-tion against rotation between the two parts at thejoint. Construction of the mathematical modelrepresenting these requirements enabled the varia-tions in the cross-section of an angled metal plateand pin arrangement to be calculated and optimizedwithin the anatomical and surgical limits imposed bythe osteotomy operation.A process which is superior to either product

evolution or design by repeated analysis is that ofdesign by synthesis. Here the specification is clear-cut and the designer is able to proceed to a solutionwithout iteration. A necessary condition here is that acomplete mathematical model is possible and thatthe design parameters are simply related. Such casesare rare but that of the operating table for largeanimals presents an example. Here the requirementwas to lift a horse through a given travel fromground-level on a plane which could be tilted throughany other plane. Since any three points uniquelydefine a plane the design was provided by a three

* Osborne-Ball Osteotomy Plate. Patent Pending, StainlessSteel B.S. EN58J, A.I.S. 316.

point support system controlled by linear hydraulicactuators reacting with the foundation. In this waysuitable choice of the actuators gives the threerequired degrees of freedom-vertical lift and tiltsabout the longitudinal axes of the table. With a con-trol system to provide lifting by all actuators actingtogether and tilting by individual actuators withclose tolerance on tilt angles for limits and levels thesolution procedure was quite unambiguous andcapable of mathematical modelling. Sizing of partswithin the model and also to commercial availabilitywas then a direct process.

ConclusionsThe steps in which a designer proceeds from prob-

lem statement to solution can be arranged in alogical manner. This arrangement is not alwaysapplicable to all design problems but sequentialplanning is always advisable. This can take the formof product evolution, or design by repeated analysisor synthesis.

ReferencesDIXON, J.R. (1966) Design Engineering. McGraw-Hill, New

York.ROSENSTEIN, A.B. (1967) The modern view of the design pro-

cess. National Congress of the Society of AutomotiveEngineers, Chicago.

STEWART, J.S.S. (1969) Meaningful monitoring. Lancet, i,1305-1307.


http://pmj.bm

j.com/

Postgrad M


ownloaded from
http://pmj.bmj.com/

principles of engineering design r. penny · postgraduate medicaljournal(june 1970) 46, 344-349....

Documents