product evolution _ a reverse engineering and redesign metho

8/9/2019 Product Evolution _ a Reverse Engineering and Redesign Metho

1/18

Product Evolution: A Reverse Engineering and Redesign Methodology

Kevin N. Otto 1 and Kristin L. Wood 2

1Engineering Design Research Laboratory, Massachusetts Institute of Technology, Cambridge, MA;

2Department of Mechanical

Engineering, The University of Texas, Austin, TX, USA

Abstract. New products drive business. To remain compe-titive, industry is continually searching for new methods toevolve their products. To address this need, we introduce anew reverse engineering and redesign methodology. We start by formulating the customer needs, followed by reverseengineering, creating a functional model through teardowns.The functional model leads to specications that match thecustomer needs. Depending upon required redesign scope, new features are possibly conceived, or not. Next, models of thespecications are developed and optimized. The new product form is then built and further optimized using designed experiments. An electric wok redesign provides an illustration.The methodology has had a positive impact on results by usinga systematic approach, both within design education and industrial applications.

Keywords: Redesign; Design methods; Reverseengineering

1. Introduction: Background, Motivationand Related Work

In the literature, a number of descriptive andprescriptive design methodologies have been devel-oped for general engineering design problems andengineering modeling (Pahl and Beitz, 1984; Pugh,1991; Ullman, 1992; Ulrich and Eppinger, 1994;Asimow, 1962; Altschuller, 1984; Dixon and Finger,1989; Clausing, 1994; Phadke, 1989). However, veryfew methodologies exist that focus on the class of problems known as redesign (adaptive, variant, etc.)(Sferro et al., 1993). As with original design, redesignproblems include the process steps of ‘gatheringcustomer needs’, ‘specication planning and devel-opment’, ‘benchmarking’, ‘concept generation’, ‘pro-

duct embodiment’, ‘prototype construction andtesting’ and ‘design for manufacturing’, but theyalso focus on an additional step, referred to here as‘reverse engineering’ (Ingle, 1994). Reverse engineer-ing initiates the redesign process, wherein a product ispredicted, observed, disassembled, analyzed, tested,‘experienced’, and documented in terms of itsfunctionality, form, physical principles, manufactur-ability and assemblability. The intent of this processstep is to fully understand and represent the currentinstantiation of a product. Based on the resultingrepresentation and understanding, a product may beevolved, either at the subsystem, conguration,component or parametric level.

In this paper, a new reverse engineering andredesign methodology is presented. This methodologyfocuses on the process steps needed to understand andrepresent a current product. Extensions of contem-porary techniques in engineering design are utilized at

a number of stages in the redesign process to meet thisgoal. A number of new techniques are also developedto address the unique characteristics of productevolution. The specic use of the combinedtechniques provides a novel context for applicationin industry and engineering-design education. In thiscontext, the primary target application for themethodology is developing a new product to enteran existing market. However, audiences ranging fromacademic instructors to applied engineers can use thismaterial to question and consider alternative productdevelopment steps. By considering alternative steps

and associated techniques, products may be viewedfrom new perspectives, perhaps leading to innova-tions ‘outside the current box’.

The next two sections describe our specic reverse-engineering and redesign methodology, with applica-tion to an electric wok product. Emphasis is placed onthe salient and unique features of our methodology,with relevant literature providing supplemental de-tails.

Research in Engineering Design (1998)10:226–243 1998 Springer-Verlag London Limited

Research in

EngineeringDesign

Correspondence and offprint requests to: Dr K. N. Otto,Engineering Design Research Laboratory, MIT, 77 MassachusettsAvenue, Room 3-449, Cambridge, MA 02139, USA. Email:[email protected]


2/18

2. Reverse Engineering and Redesign

To motivate the need for a redesign methodology,consider an abstraction of product evolution in themarketplace, depicted in Fig. 1. As shown in the

gure, one can determine a critical metric that isuseful for evaluating a product and its marketcompetition, and plot the value of performance foreach product as a function of the time when eachproduct was introduced. The metric values willnaturally fall as an S-curve in time (Foster, 1986;Betz, 1993). At rst, a new innovative product willenter a market domain as a new concept. The plottedcurve will remain somewhat ‘at’ for a certain periodof time, representing the time for the competition torespond. Next, a rapid profusion of innovation occurs,and many products are launched in time. The lowerleg of the ‘S’ is forming. The new technology,however, eventually tops out, physical laws of theprocess dominate, and the engineers cannot extractmore performance. The slope of the ‘S’ tops outagain, and the curve becomes atter.

Depending on the competitive environment, aproduct development team must redesign theirproduct at two levels into the product’s future. Therst level is as described along an individual S-curveand includes parametric, variant and minor adaptive

changes in the manufacturing, components, materials,geometry, assemblies, and subassemblies. Thesechanges are a direct response to customer needs andfeedback (Ashley, 1994). The second level, on theother hand, reects a discontinuous jump in theproduct characteristics. Discontinuities of this typeresult from the introduction of new technologies, newproduction processes, or a fundamental change inproduct architecture.

The bottom line of product evolution with respectto S-curves is that all products must change (bothnonlinearly along an S-curve and discontinuouslybetween them) to remain competitive. We proposethat a systematic methodology will lead to a betterunderstanding of product evolution and how toexecute effective change with reduced cycle time.The next subsection introduces the structure of such amethodology.

2.1. General Methodology

Figure 2 shows the general composition of ourreverse engineering and redesign methodology.Three distinct phases embody the methodology:reverse engineering, modeling and analysis, andredesign. The intent of the rst phase, reverseengineering, is twofold. First, a product is treated asa black box, experienced over its operating para-meters, and studied with respect to customer needsand predicted and/or hypothesized functionality,product components, and physical principles. The

second step of the reverse engineering phase is toexperience the actual product in both function andform. This subphase includes the full disassembly of the product, design for manufacturing analysis,further functional analysis, and the generation of nal design specications.

The second stage of the methodology entails thedevelopment and execution of design models,analysis strategies, model calibration, and experi-mentation. The third and nal stage of the methodol-ogy then initiates product redesign based on theresults of the reverse engineering and modelingphases. Parametric redesign may be pursued usingoptimization analysis of the design models. Alter-natively or in concert, adaptive redesign of productcomponents and subassemblies may be pursued.

Beyond parametric or adaptive redesign, anoriginal redesign effort may be needed to satisfy thecustomer needs. An original redesign, in this context,implies that a major conict exists between thecustomer needs and the current product in the market.Because of this conict, it is deemed that an entirely

Fig. 1. Product evolution along S-curves (from Betz, 1993).

Product Evolution: A Reverse Engineering and Redesign Methodology 227


3/18

new product concept is needed. Functional analysesfrom the reverse engineering phase will direct theredesign effort.

In sum, unbiased prediction, customer-drivendesign, analysis using basic principles, and hands-onexperimentation are the philosophical underpinningsof this redesign methodology. The intent of themethodology is to be dynamic, depending on theneeded evolution for a product. The dashed line in

Fig. 2 shows the global dynamic nature of themethodology. For some product redesigns, it may beappropriate to perform adaptive or original changesbefore creating and optimizing a design model;similarly, the model development of phase two maylead to a better understanding of the product, by-passing parametric redesign and leading directly toadaptive. Alternatively, another product redesign maycall for simple parametric modications to produce

Fig. 2. Reverse engineering and redesign methodology.

228 K. N. Otto and K. L. Wood


4/18

dramatic S-curve response in quality and protmargin. The proposed methodology may be used forany of these scenarios or others.

The next section builds upon the methodology of Fig. 2. Explanations of the tools and techniques foreach step in the methodology are provided. Withoutappropriate techniques, the methodology would berelegated to a philosophic goal without a foundationfor actual use in the reality of industry or theuniversity classroom.

3. Adaptive and Parametric Redesign:Applications to an Electric Wok

Based on the redesign methodology of Fig. 2, thereare several tasks that must be completed to execute aneffective redesign of a commercial product. Webriey introduce each task below. For continuity,however, only certain tasks are embellished anddescribed in detail, following a particular theme of anelectric wok example.

Step 1: Reverse Engineering: Investigation, Predic-tion, HypothesisTwo primary goals are intrinsic to the rst step of themethodology: (1) clarify the product domain, devel-oping a rigorous statement of the customer needs, and(2) treat the product as a ‘black box’ and eitherhypothesize the ‘internal’ functions and productfeatures (solution principles), or choose the designteam’s preferences for these items. Let’s consider the

salient features of this step, particularly black boxmodeling, customer need analysis, and predictingproduct function.

Task clarication: after stating an initial problemstatement, a black box model is created, identifyingthe input and output ows of materials, energies, andsignals and the global function of the product. Theseare documented in a single input-output block model.The intent here is to understand the overall productfunction, while maintaining, guratively and literally,very little knowledge of the internal components of the product. By so doing, an ‘unbiased’ perception of

possible product evolution is maintained, in additionto avoiding psychological inertia when generatingconcepts in later stages of the methodology.Customer need analysis and diagnosis of product weaknesses: based on the problem statement andblack-box model, customer needs are gathered andorganized for the product. The voice of the customeris the essential task in forming a complete and usableproduct design specication. Several techniques exist

to gather a list of customer needs. These techniquesinclude: direct use of the product, circulatingquestionnaires, holding focus group discussions, andconducting interviews. A more complete discussionof the different methods is given in Urban and Hauser(1993).

For the purposes of our methodology, the task of

gathering customer needs involves the subtasks of interviewing an appropriate sample size of customers.Typically, nine or more customers are interviewed forsmall consumer products (Grifn and Hauser, 1993),recording the customer statements in their words fromprompted questions or from spontaneous statements(such as product likes, dislikes, and suggestions).These customer needs are interpreted into a noun-verbformat, and then ranked according to importance.After completing customer need collection forms withthis information, each interpreted need is copied ontoan index card or post-it note, listing the need,

importance rating, project title, and customer ID.The index cards are then grouped into collectivecustomer need statements, and the relative importanceof each group is assigned using the number of indexcards per group, combined with the importance rating.The result of this technique, as documented fully forthe wok example in Otto (1997), is a completecustomer need list, with both primary and secondaryneeds and weightings. This customer need list isaugmented with economic feasibility to determine thepotential return on investment (Thornton and Meeker,1995; Miller, 1995; Ulrich and Eppinger, 1994).

Functional prediction: assuming an adequate cover-age of customer needs and economic viability,functional analysis begins the methodology’s predic-tion tasks (Fig. 2). This analysis includes thedevelopment of a process description, or activitydiagram, and the forming of a function structure, assummarized in Fig. 3. An important tool for analyzingthe function of a product is to specify the ‘process’ bywhich the product being designed will be functionallyimplemented. A process or process description, in thissense, includes three phases: preparation, executionand conclusion (Hubka et al., 1988). Within each

phase, high-level user and device functions are listedto show the full cycle of a product, from purchase torecycling or disposal. After listing the high-levelfunctions in each phase, a number of productcharacteristics are chosen, including the product’ssystem boundary, parallel and sequential pathsthrough the function structure, process choices, andinteractions between user and device functions. Thesechoices are documented in a process description form.



5/18

Using the process description and customer needs,a hypothesized function structure for the product isformulated (Fig. 3). Function structure modeling(Pahl and Beitz, 1984; Miles, 1972; Hubka et al.,1988; Ullman, 1993; Otto and Wood, 1997;Shimomura et al., 1995) has historically been usedto create a form-independent expression of a productdesign. We extend common function structuremodeling to include an approach for mappingcustomer needs to subfunction sequences (calledtask listing), a method for aggregating subfunctions,and a comparison of a functional decomposition withcustomer needs.Functional modeling: the rst step is to identify majorows associated with the customer needs of theproduct activities. A ow is a physical phenomenonintrinsic to a product operation or subfunction. Forexample, an operation may be to convert electricityfor heating food in a wok. Three critical ows for thisoperation are an input electrical energy, the heatenergy produced from the conversion, and foodmaterial being operated upon. Stone and Wood(1998) summarize common ows used in functionalmodeling.

For each of the ows, the next step (Fig. 3) is toidentify a sequence of subfunctions that when linkedrepresent the hypothesized product functions orcustomer activities when interfacing with the product.A subfunction, in this case, is an active verb paired

with a noun that represents a product operation. Littleand Wood (1997) provides lists of appropriate verbsand nouns to use in functional analysis.

For example, two important customer needs,expressed in the customer’s voice, may exist for anelectric-wok product as ‘heats and cools quickly’ and‘temperature uniform across inner surface’. A suitableow for addressing this need is an energy ow of heatthat ultimately acts to heat a material ow of food. A

sequence of subfunctions for the energy ow may beof the form: convert electricity to radiation, heatcontainer, conduct heat, sense heat, regulate heat, heatfood, insulate from environment, etc.

Once the design team completes the functionstructure, functional modeling comes to a closurethrough a verication step. This verication entails areview of the customer needs list, where thesubfunction or sequence of subfunctions are identiedthat satisfy each customer need. Needs not covered bythe function structure require further analysis andadded functionality; sub-functions not satisfying aneed require conrmation of their incorporation.

A predicted process description and functionstructure provide a sound basis for critically analyzingan actual product and for seeking avenues to improvequality. The customer needs are related to a form-independent functional model of the product. Twoadditional hypothesis tasks are now needed forcompletion: directly matching the customer needswith product features and working physical princi-ples. Between these two matchings, a design team candetermine which product features or principles offerthe best opportunity for redesign.

In summary, the rst step of the redesignmethodology provides us with the necessary investi-gation, hypotheses, and predictions for a successfulproduct evolution. While this information is only thepreface to the redesign task, if properly wielded, itshould give us insights that far exceed our expecta-tions and the resources needed to gather anddocument the product state. Let’s consider the electricwok application to illustrate example results.

Wok redesign application – Step 1: this redesignproject was initiated through the perception of a need:the inadequacy of current electric woks to satisfy thedemands of the young urban dweller desiring to

Fig. 3. Functional modeling and analysis process.



6/18

conveniently cook authentic Chinese food (Fig. 4).The original product is a six-quart Electric Wok,shown in Fig. 5.

A competitive product is a traditional wok, usedover a gas ame and heated by convective and

radiative modes of heat transfer. Heating occursuniformly across the entire wok surface, rather than ina concentrated ring. This method thus lends itself touniformly cooking the food placed within it.

Beginning with the black box model in Fig. 4, a setof customers were interviewed as they used the wok in normal cooking operations. Interview sheets wereused to document the customers’ stated strengths andweaknesses of the product, especially compared with

a traditional wok. These sheets were then organizedinto a cumulative customer need list, recording theaggregate importance ratings as evaluated by thecustomers. Important customer needs include:temperature – ‘temperature uniform across theinner surface’, ‘heat and cool quickly’, ‘maintainuniform temperature in time’, etc.; size – ‘compactwok for countertop and storage’; cleanable – ‘preventfood from sticking’ and ‘allow the removal of cooking surface from heating unit’, etc. Figure 6summarizes the critical customer needs diagnosedfrom the current product. (Otto, 1996) provides thedetailed approach and analysis that supports this list.In the following sections, we build on these customerneeds and focus on the key threads of ‘uniformtemperature’ and ‘heats quickly’ to illustrate themethodology at a sufcient level of detail.

Step 2: Reverse Engineering: Product Teardown and ExperimentationConcrete experience now becomes the emphasis of the redesign methodology. The current productarchitecture must be understood in detail, and, mostimportantly, the customer needs must be comparedwith the current product’s functionality, solutionprinciples and choices of design parameters. Whilethis is generally understood when working to redesignan existing product, a development team needs toexecute a complete teardown approach when theyhave no product in the market and are using acompetitive product as a baseline. Also, for educationpurposes, students can apply this approach to dissect,

deduce and understand how a product was designed.Overall, repeating this approach, on several compe-titive products provides a comparison that ultimatelyleads to avenues for product evolution.

Product teardown: product disassembly (or teardown)initiates the second step of reverse engineering. Aplan is incrementally developed for the disassemblyactivity, listing the order of disassembly, componentor assembly to be removed, tool usage, accessdirection, orientation of product (to prevent compo-nents from falling out) and any expected permanentdeformation caused by the disassembly. In itsentirety, the plan provides a means for assessing theassemblability of the product, as well as a means forreturning the product to its original form. It should benoted, however, that the disassembly plan may notcorrespond to the same order the product wasoriginally assembled. It just provides one possiblearrangement for evaluation.

The disassembly is systematically executed, label-ing each component as it is removed. Distinct

Fig. 4. Wok black box model.

Fig. 5. Six quart electrical wok.



7/18

Fig. 6. Electric cooking wok: Customer needs list w/diagnosis of weaknesses.



8/18

assemblies or sub-assemblies are also noted tounderstand the product architecture. As the producttear down progresses, a Bill-Of-Materials (BOM) andan exploded view (Fig. 7) are constructed toinvestigate the abstract and concrete features of theproduct. This BOM and exploded view should list allassemblies, sub-assemblies and parts in the order theywere originally assembled. These results make DFAcost analysis easier.

In concert with product tear down, a novel

Subtract-and-Operate (SOP) procedure is executedto study the functional dependence of each productcomponent (Lefever and Wood, 1996; Lefever, 1995).A summary of the procedure for an assembly orsubassembly of a product is given below:

Disassemble (subtract) one component of theassembly.

Operate the system through its full range.

Analyze effect through visual inspection ormeasurements.

Deduce the component’s subfunction; compare toBOM.

Replace component; repeat for all parts inassembly.

Using this procedure, construction of a SOP tableoccurs, listing the part number, part description, anddetermined effect of removal. Through this proce-dure, a more rened understanding of componentfunctionality may be obtained; likewise, for thosecomponents that are recorded as having ‘no effect ondegrees-of-freedom’ when subtracted, redundantfunctionality exists in the product, implying thatpart reduction may be directly pursued. Table 1 showsthe SOP deduced functionality for the electric wok, aswell as identied components that represent avenuesfor simplication (Type 1 and 2 degree of freedomredundancies in the table).

Fig. 7. Wok bill of materials and exploded view.



9/18

Experimentation: the nal task of the second step is toexperiment with the overall product, its assemblies,and its components. Step 1 of the redesign methodol-ogy generates a complete list of customer needs, inaddition to predicted product features and physicalprinciples. These data, in conjunction with the BOM,are now used to direct the choice of physicalparameters that must be recorded for the product.For example, for the electric wok customer needs of

‘uniform temperature’ and ‘heat quickly’, the tear-down process reveals a heating coil, controller, coilsupports, spacers, and a wok bowl. Temperaturedistributions of the bowl (spatially and in time),geometry of components, and material properties of the coil, bowl, and supports are needed to furtherunderstand the meaning of the customer needs.Recognition of these physical parameters may notoccur at this stage of the redesign, but instead during

Table 1. Electrical wok SOP effects table

Assembly/ Part No.

Part Description Effect of Removal Degree-of-Freedom Type

Deduced Subfunction(s) &Affected Customer Needs

A1 Lid Assembly001 Cover Heat radiates to environment;

Food splattersInsulate Energy FlowContain Food

002 Cover knob No effect 2 Import Hand

A2 Bowl Assembly003 Wok bowl Food falls into base;

Food directly in contact withcoils

Contain food, Conduct Heat,Direct Food, Heat Food, EnableCleaning

004 Handles & Handle Caps Burn hands 2(handles)

1 (caps)

Import HandInsulate HandTransmit Weight

005 Wok stand Bowl cannot be oriented/ stabilized;Heat element is exposed;Temp. settings not visible

2(each

redundantfoot)

Orient Cooking, Transmit Loads,Distribute Loads, Insulate Heat,Indicate Temperature Setting,Prevent Skidding

006 Connector prongs Cannot turn on/off;Electricity doesn’t ow;Exposes electricity

1(secondprong)

Actuate On/Off Transmit EnergyProtect from Shock

007 PEM studs Bowl is not connected to base 1(>1 stud)

Transmit Loads

008 Metal Plate Exposed coils;Safety info. not displayed

Insulate HeatIndicate Visual Safety Signals

A3 Heating Assembly009 Electrical Cord No connection to source Connect to Energy

Transmit Energy010 Thermostat Temperature very high,

continuouslyRegulate Power/Temp.Sense Heat

011 Heating Coil Open circuit;Wok doesn’t heat

Convert Electrical Energy toRadiation

012 String Heating coil is not held inplace

Transmit Loads

013 Insulated Wires Temp. rise continuously;No temp. sensing

Transmit EE to Sensing

A4 Electrical HeatingSupport

014 Radiation Shield Thermostat heats up;Heat is lower in bowl

Shield Temp. ControlDirect Heat

015 Coil Supports Heating coil misaligned;Housing heats up

2(>1 support)

Insulate HeatDirect Heat

016 Element housing Heating element is freeoating;Bowl does not heat well

Direct Heat/Radiation (HeatContainer)

..... ..... ..... ..... .....



10/18

the forming of engineering metrics (Step 5), or theconstruction of design models (Step 6). In either case,the design team iterates back to the experimentationtask to obtain the required data. Results of the task,independent of when it is executed, are recorded onan updated BOM.

Wok application – Step 2: the wok tear down revealed

the cause of the failure to deliver a uniformtemperature distribution across the bowl. The electricelements were housed within a narrow circularchannel, concentrating the heat transfer. The non-uniform heating characteristics in time were revealedto stem from a bimetallic temperature controller,which also was the design driver for the poor powercontrol interface.

Step 3: Reverse Engineering: Functional AnalysisProduct disassembly provides detailed informationregarding component function, assemblability, physi-

cal parameters, manufacturing processes, and anintuitive understanding of the product as a conse-quence of the various hands-on experiences. Theseresults must now be abstracted to the level of thecustomer needs statements, developed in Step 1. Byso doing, the design team may identify and rank areasof focus for product improvement.

Functional analysis is the key instrument forbuilding the abstraction. Two primary tasks enablefunctional analysis of the actual product. The rst task is to develop energy ow diagrams of the productand/or separate product assemblies (Lefever and

Wood, 1996; Lefever, 1995). Energy ow diagramsare graphical aids that represent the transfer of energythrough a product’s components.

Actual product function: after or in concert with forceow analysis, a function structure for the actualproduct is formulated. This task entails the sameapproach as described in Step 1 (Fig. 3), except, inthis case, subfunctions are not predicted; they arederived from the actual components in the product.This time we have the benet of tearing apart theactual product. We believe this two step approachhelps a design team understand different physicalprinciples by which the product could operate. Usingthe ows for each customer need, as developed inStep 1, a useful approach to generating subfunctionsfor the actual product is to examine the exploded viewand BOM from Step 2. Each ow is traced throughthe product, recording the subfunction(s) for acomponent as it is encountered by the ow. Thisprocess creates a subfunction chain for each ow. Thenext step is to connect the parallel ow chains

together as a network, adding subfunctions and owsfor components that link ow chains. The result of this process is the desired function structure.

Wok application – Step 3: a function structure wasdeveloped by rst associating ows to each customerneed. This process produced three primary ows: owof food through the wok, ow of electricity, and heatow. These ows were then expanded into sequencesof subfunctions as the ows traced through the wok.

They were then aggregated in the structure (Fig. 8).For the customer needs of ‘uniform temperature’ and‘heats quickly’, the key function chain from Fig. 8maps the electrical energy input to the heating of thefood material. The nine functions associated withthese ows (highlighted in Fig. 8) provide a high-level model for the customer needs. They are alsotargets for generating adaptive concepts for evolvingthe wok.

Step 4: Reverse Engineering: Constraint PropagationProduct disassembly and functional analysis, as

completed in Steps 2 and 3, yield detailed designdata for product evolution. With these data, improve-ments in design function, parameter choices, etc. maybe approached knowledgeably for assemblies, sub-assemblies, or components. However, constraintsbetween product components must also be wellunderstood. With such understanding, the ramica-tions and propagation of design changes may beproperly forecasted.

Fig. 8. Wok function structure – actual product.



11/18

Morphological analysis: the rst task in this step is tocreate a morphological matrix (Pahl and Beitz, 1984;Ulrich and Eppinger, 1994) of the solution principles(components) for the signicant functions. Each rowof the matrix corresponds to a subfunction from Step3, and each column represents a solution principle.Function sharing and compatibility: subsequently,function sharing and compatibility analysis comprisethe second task for understanding the productconguration (Lefever and Wood, 1996). The designteam may identify function sharing of the productcomponents simply by scanning the morphologicalmatrix for components that are listed for more thanone subfunction. Through such identication, it ispossible to plan for design changes without violatingfunctional requirements. Likewise, compatibility of the produce components may be analyzed bydissecting each critical product interface, betweencomponents, to identify its degrees-of-freedom,access directions, relative motion with other compo-

nents, tribological properties, and tolerances (size andgeometric). A matrix listing each component versusthese data may be constructed to document theanalysis.Wok application – Step 4: a morphological matrixwas developed for the electric wok (Fig. 9). (Note:results are also included here from Step 9.). Likewise,a function-sharing and compatibility analysis for thewok (customer needs: uniform temperature and heats

quickly) shows that the heating coil, element housing(reector), and wok bowl perform six of the ninefunctions of the highlighted chain in Fig. 8. It alsoshows the size, shape, or material properties of thesecomponents are coupled to each other (especially dueto the inset ring to attach the coil to the bowl).Parametric or adaptive redesigns to satisfy thecustomer needs will thus require a change in theinterface between the components.

Step 5: Reverse Engineering: Forming EngineeringSpecicationsThe last step of reverse engineering entails theforming of specications, benchmarking, and choos-ing the product systems that will be evolved. Theintent of the rst task is to dene quantitative targetsfor the product (Otto, 1995). Having establishedorganized customer needs and a function structure forthe product, each subfunction must be associated to atleast one line item in a product Specication Sheet

(House of Quality), where each specication itemmust be quantiable and measurable.

To establish an initial set of engineering specica-tions, a design team should start by listing eachsubfunction as rows of a matrix. For each subfunction,a means to measure the input and output ows shouldbe conceived. These measures should be listed foreach subfunction row as the metrics for thesubfunction. For example, a subfunction exists to

Fig. 9. Wok morphological matrix.



12/18

‘regulate temperature’ of an electric wok (Fig. 8).Possible measurements include steady state tempera-ture error, set point temperature error, and tempera-ture variation across the material (wok bowl).

Next, the generated metrics must have target valuesassigned to each. This subtask is accomplished byexamining the relevant customer needs and perform-ing benchmarks with related products or technologies.After these tasks, the results should be collected into aHouse of Quality matrix, creating the necessary

importance rankings and relationship matrices. Theresult of this process is the forming of quantiedspecications with direct links to customer needs(Hauser and Clausing, 1985; Otto, 1995).

Quantied specications provide clear goals forevolving the current product. In turn, a completeHouse of Quality, with this information, may be usedto choose, preliminarily, which components of theproduct should be evolved rst. A question exists,however, concerning the type of redesign activity topursue. Should parametric redesign be attempted rst,or is adaptive redesign required, followed by

parametric efforts? This question may be straightfor-ward to answer when obvious component decienciesexist in a product; however, in many cases, an answermay not be easily forthcoming. For example, in theredesign of an electric wok for quick heating anduniform temperature prole, an engineering specica-tion may exist to reduce the temperature distributionacross the entire heating surface from 55 F to 10 F.Such a large reduction in temperature variation may

seem to imply that a technology change is needed inthe wok. A simple mathematical model based on heattransfer may show, however, that changes in materialproperties, thermal mass, etc. can easily meet thedesired specication. Thus, creating a parametricmodel of the current product is a needed step in theevolution process, as presented in the next section.

Wok application – Step 5: a full House of Quality(specication) was developed for the electric wok.Space limitations prevent a full listing of the results;however, Table 2 shows abbreviated House of Qualitydata for the customer needs of ‘uniform temperature’and ‘heats quickly’.

Step 6: Modeling and Analysis: Model Development Virtual and physical modeling of a product willprovide in-depth insights into its operation andpossible improvements that may be achieved para-metrically. Three tasks are needed to develop a virtualor mathematical model. Beginning with each im-portant customer need or engineering metric from theHouse of Quality, the critical product componentsshould be listed. The governing physical principlesand associated modeling assumptions must then beidentied for each component, or the group of components as a whole. For example, for an electricwok, the customer need of ‘uniform temperature’ isimportant. Components affecting uniform tempera-ture (spatially) are the heating element, the electricalenergy input, the wok bowl, the radiation reector(housing), supports, and the food. The choices for the

Table 2. Abbreviated House of Quality data for an electric wok

Customer Need Sub-function(s)(to be measured)

Metric(s) Benchmark (Consumer Reports, 1997)

Target

Uniform Regulate Electricity Steady Set Point Maxim 5 F 5 FTemperature on (w.r.t. temp.) State Temp EW-50:Inner Surface Temp. Error Excellent

Error (space) West Bend

(time) [ F] 79525: Fair[ F] (50 F) Joyce Chen

Wok 22–9940:Excellent

Heat food Temperature Maxim EW-50: Excellent Excellent:Variation Across West Bend 79525: Fair 5–10 FSurface [ F] (55 F)

Joyce Chen Wok 22–9940:Excellent

Heats Quickly Heat food Temperature Rise Maxim EW-50: Excellent Excellent:Time [sec.] West Bend 79525: Fair 1 min.

(>2 min.) Joyce Chen Wok 22–9940:

Excellent



13/18

governing physical principles might be conduction,convection, and radiation assuming steady state andlumped masses.

High-level model: after identifying the physicalprinciples and assumptions for each customer need,a balance relationship is created to document a high-level physical model. As in Step 1, a shbone

diagram, or a failure modes and effects diagram(Clausing, 1994), may be used to document thebalance relationship, where the ‘effect’ in the diagramis the customer need, and the ‘causes’ are the physicalprinciples. The ultimate goal is to rene the ‘causes’to the level of physical parameters. Balance relationships: the last task in formulatingmathematical models for the product is to convert thebalance relationships into a set of mathematicalequations. Basic engineering principles may be usedto choose the appropriate scaling law relationships(Miller, 1995). Parameter ranges from the product’s

bill-of-materials should be used to augment themathematical relationships with appropriate para-

meter values. If such parameter values are unavail-able, the ‘Experiment with Product Components’ task of Step 2 should be executed.

Physical prototype models: in some cases, cycle-time,economic, or product-complexity considerations mayprevent the development of a mathematical model.The creation of a physical prototype can be used as an

alternative modeling approach. Prototype modelsshould be designed in such circumstances. Theintent here is to create a bench-top or otherexperiment (not an entire product prototype) for acustomer need, focusing on the effected productcomponents and variables. For instance, a customerneed may exist for an electric wok lid handle to ‘tcomfortably in the user’s hand during operation’. Amathematical model of this need is not directlyapparent; however, physical prototypes that varyshape, size, and texture of the handle may be designedfor analysis and testing.

Wok application – Step 6: mass, steady-statetemperature uniformity, temperature rise time, and

Fig. 10. Wok heat transfer model.



14/18

volume equations were derived for the electric wok,in addition to handle temperature and weight. Anexample high-level model and spatial and balancerelationship for steady-state temperature uniformityare shown in Fig. 10.

Step 7: Modeling and Analysis: Analysis Strategies

Models are developed for each of the importantcustomer needs in Step 6. The next step is to developanalysis strategies for solving the models. The rsttask for developing such strategies is to calibrate themodel for each customer need.

Model calibration and formulation: model calibrationentails the ‘matching’ of the model to the actualperformance of the product. Two update proceduresare usually required for calibration: either changingestimates of model parameters, or adding terms to themodel to capture important physical effects. Once themodel is calibrated, it must be converted into a formthat is conducive for solution. Example formulationsinclude objective functions and constraints foroptimization or constraint propagation, simultaneousequations for spreadsheets, state space equations forsimulation, or combinations of these.

Physical prototypes: an alternative approach to thevirtual model formulations is the creation of aphysical prototype strategy. In this case, design of experiments must be developed, including the numberof control factors, identication of noise factors,choice of response(s), number of experiments, and theexperimental matrix. For statistical validity, it is alsoimportant to choose the number of replicates, the in-between tests for residual analysis, blocking, and therandom run order for the tests.

Wok application – Step 7: based on the model fromStep 6, experiments were executed to determine wok time constants and heat transfer coefcients. Theseexperiments included temperature measurements withthermocouples attached to the inner surface of thebowl (consistent with the specication metrics, Table1). Figure 11 shows example experimental results at a10–20% power level. The results were used tocalibrate the metric equations and to solve anoptimization formulation:

||T || (temp. variation)t r t t (rise time)|T (t r ) T ss | (temp. at SS)T .(t ) + f (t ) = 0 (transients)

W W (weight)T c T c (min. center T)T h T h (max. handle T)

Steps 8–10: Redesign: Parametric or AdaptiveThe data from Steps 1–7 provide the design team withimmeasurable knowledge and capability. Customerneeds are gathered and organized. Product function ispredicted, experienced and abstracted. Physicalprinciples are hypothesized, tested, and modeled.Engineering specications are quantied and ranked.The design team must now employ these data in asuccessful redesign effort, i.e. parametric, adaptive,and/or original.

Parametric redesign: parametric redesign requires amodel of the original product or a model of a newproduct conguration following adaptive redesign.The model should be calibrated and reformulatedaccording to the chosen solution technique. Optimi-zation, spread sheets, and/or simulation is thenapplied to obtain new product parameters. Verica-tion of the results then commences using sensitivityanalysis or tolerance analysis, such as statisticaltolerance stackup (Ullman, 1992) or Taguchitolerance design (Phadke, 1989). If satisfactoryresults are obtained from the parametric redesign,fabrication and experimental testing of a betaprototype commences to verify the recommendeddesign improvements. Adaptive redesign: in the case of adaptive redesign,the design team seeks to create alternative solutionprinciples to chosen product subsystems, replacesubfunctions of the product, or add new subfunctionsto the product.

The rst task, then, is to update the functionaldescription of the product, by either comparing thepredicted function structure to the actual, or by simplyadding new functions prescribed by the customerneeds. Based on the new function structure, adaptive

Fig. 11. Wok calibration.



15/18

solutions may be obtained in a number of ways. Forexample, the morphological matrix of Step 4 may beexpanded to include a wide variety of solutionprinciples for the important subfunctions. Brainstorming techniques and discursive bias are directlyapplicable for augmenting the morphological matrix.Alternatively, or in concert, the Theory of InventiveProblem Solving (TIPS) may be applied (Altschuller,1984). This theory employs documented physicaleffects (over 1000), design principles (40), andprevailing trends in technology (eight so-called lawsof evolution) to seek alternative solutions. Forexample, the customer needs, functional models,and specication of an electric wok might show aconict between ‘washable’ and ‘safety’ (doesn’tburn or heat parts excessively). Separating the baseand bowl in time and space (exposed heating elementwhen bowl is attached vs. covered element when bowlis removed for washing) is a possible solutionsuggested by the TIPS’ principles. Alternatively,TIPS recommends covering the heating elementswith a radiative-transparent material that does notretain heat as it transmits radiation.

No matter what adaptive redesign technique isemployed, new solution principles result. Thesesolutions must be veried with the function sharingand compatibility analysis of Step 4 of the methodol-ogy, and checked against current patents. In addition,modeling and testing of a physical prototype mustoccur to verify the design changes, usually after adesign model is executed.Product integration: as shown in Fig. 2, the nal task in the reverse and redesign methodology (prior todetail design and beta prototyping) is to perform aproduct integration of the new concepts developedfrom the parametric and adaptive redesign steps. Thistask, at a fundamental level, entails a decisionprocess of whether or not to implement theindividual concept ideas into the product. Manyfactors must be considered to perform this analysis:importance level of the customer need, compatibilityof the concept with other being considered, how wellthe design concept satises the target values, thecurrent performance of competitors’ products, the

capability of the company to fabricate the newconcept within the current manufacturing process,and the cost impact of the new concept on theproduct. A Pugh chart, trade-off table or detailedperformance/cost analysis is needed to execute thisdecision process.Wok application – Steps 8–9: the model developed inSteps 6 and 7 was used to perform a parametricredesign of the wok bowl and heating element. This

model was solved using the EXCEL Solver, resultingin a new bowl geometry (increased thickness for agreater thermal mass), greater aspect ratio from theheating element, modied handle geometry thatremains cool, etc. Figure 12 shows a result of runningthe optimization. Notice that for a bowl thickness of 0.0035 m, the uniformity of temperature (spatially in

Fig. 12. Example optimization results for the parametric analysis.

Fig. 13. Wok DOE experiments.



16/18

the heating area, from center to mid-point) decreasedto 6 F, satisfying the target value shown in Table 1. Adesigned experiment was subsequently performedabout the new model optimum (Fig. 13).

In addition to the parametric redesign, adaptiveconcepts were chosen based on the morphologicalconcepts developed in Fig. 9. A Pugh chart was

developed to determine the preferred choices of theseconcepts, based on the specications and House of Quality in Step 5. Preferred choices of concepts arehighlighted in Fig. 9.

Based on the results of parametric and adaptiveredesign, product integration of the new concepts wasconsidered. Table 3 shows an abbreviated trade-off table for the new wok concepts, focusing only on thecustomer needs of ‘uniform temperature’ and ‘heatquickly’. Concepts were discarded if they did notsignicantly increase the product performance, if theyincreased cost signicantly, or if they raised new

customer need issues, such as reliability, etc.The nal alpha prototype for the redesigned wok (Fig. 14) represents the result of the productintegration shown in Table 3. It includes manyadvances: a removable bowl for washing, a largehandle, an on/off switch, removable cord, simple/ visible power control, uniform power control intime, compactable volume for storage, and a wide

view radiant surface. No electric wok on the market yet incorporates all of these customer-requested features.

Table 3. Abbreviated product integration trade-offs

Concept Customer Need Customer Need Signicant % Capability Cost IntegrateRating (Fig. 6) Toward Y/N)?

Targets (Table 1)

Thicker Bowl(Parametric)

Uniform Temp. 7,/4, Y/Y Y Y Low Y

Bowl shape(Parametric)

Uniform Temp. 7,/4, Y/Y Y Y Low Y

New Bowl Mat’l,(Parametric)

Uniform Temp.Heat Quickly

7,/4, Y/Y6, Y

Y Y Low Y

Ceramic HeatingElement(Adaptive)

Uniform Temp. 7,/4, Y/Y N Y Medium N

Pirex Heating

System(Adaptive)

Uniform Temp. 7,/4, Y/Y Y Y High N

Power Controller:Dimmer Switch(Adaptive)

Uniform Temp. 7,/4, Y/Y Y Y None Y

Dish-ShapedReector(Parametric &Adaptive)

Uniform Temp.Heat Quickly

7,/4, Y/Y6, Y

Y Y Low Y

.....

Fig. 14. Final alpha prototype.



17/18

4. Conclusions: Introspection of DesignEducation and Industry

This paper presents a novel redesign methodology,compared with related literature, such as the work of Sheppard and Tsai (1992), Brereton and Leifer (1993)and Gabriele (1994). It is based on a ten-step process

with three primary phases. Each phase provides aclear set of tasks to seek new product congurations,combining contemporary design techniques with newapplications and extensions developed by the authors.These tasks provide a unique forum for signicantlyevolving all types of consumer products, as illustratedby the electric wok example. Over the last four years,the methodology has been applied to over 150products at MIT and UT (Otto et al., 1998; Jensonet al., 1998).

4.1. Design Education

We have conducted these methods both on industrialapplications and taught these techniques for ve yearsrunning at MIT, UT and the United States Air ForceAcademy: MIT as a single class project graduatecourse, and UT/USAFA as a two-person projectfreshman course, a 3–5 person project sophomorecourse, a ve-person project senior course, and asingle-person project graduate course. In general, wend the exercise of a structured design process hasmany benets, including concrete experiences withhands-on products; applications of contemporary

technologies; realistic and fruitful applications of applied mathematics and science principles; conceptgeneration from function to morphological matricesto new effects, studies of systematic experimentation;exploration of the boundaries of design methodology,and decision making for real product development.

At MIT, the instruction focuses on exploring thestrengths and deciencies of various structured designmethods that have been developed in the community,as detailed above. The student population includes alarge fraction of persons returning to graduate schoolafter having worked in industry. They nd that, forthe most part, they did not have a full appreciation forthe theory and power behind many of the methods.Many are surprised at the quality and comprehensivenature of results that the methods provide, andbecome believers in a systems approach to a wellposed design process.

At UT, the results are much the same as at MIT.The most popular project in the freshman-levelintroductory course to mechanical engineering is thereverse engineering of a children’s toy or simple

mechanical device. Students resonate with theopportunity to learn the basics of design and applyprinciples from their freshman physics and mathe-matics courses. Senior-level students express theirdesire for even more hands-on experiences. They alsomake it clear that the redesign of a consumer productis a great forum for learning design methodology andfor preparing for their capstone design class thatfocuses on design projects sponsored by industry.

4.2. Industrial Relevance

A common design methodology criticism is that it isexplanatory and thought provoking, but not relevantto actual practice. This criticism is becoming anantiquated view. We nd leading industrial compa-nies to be constantly seeking more structuredapproaches to their product development processes,especially in the era of concurrent development,

intensive computing, and downsizing of workforces.An effective structured method allows not just oneexpert to understand and complete a task, but manyothers as well.

These observations have led to most companiesdeveloping internal product development processes,which they also tend to guard as proprietary knowl-edge. Our experience is that, for the most part, there islittle intellectually unique within these processes. Yet,as our industrial partners have remarked, productdevelopment remains the key market battleground,and simply the high stakes involved lead companiesto these measures of guarding structured designmethods.

Acknowledgements

The research reported in this document was made possible,in part, by a Faculty Early Career Development Award andYoung Investigator Awards from the National ScienceFoundation. The authors also wish to acknowledge thesupport of the Leaders for Manufacturing Program, acollaboration between MIT and U.S. industry, in addition toFord Motor Company, Texas Instruments, DesktopManufacturing Corporation, and the UT June and GeneGillis Endowed Faculty Fellow. Any opinions, ndings,conclusions, or recommendations are those of the authorsand do not necessarily reect the views of the sponsors.

References

Altshuller, G. Creativity as an Exact Science, Gordon &Breach Publishers, NY, 1984.



18/18

Asimow, M. Introduction to Design, Prentice Hall, EnglewoodCliffs, NJ, 1962.

Ashley, S. If at rst you don’t succeed, Redesign, MechanicalEngineering (3), 1994.

Betz, F. Strategic Technology Management. McGraw Hill,New York, 1993.

Brereton, M., Leifer, L. et al. An exploration of engineeringlearning, Proceedings of the ASME Design Theory andMethodology Conference, Albuquerque, NM, 1993, pp

195–206.Clausing, D. Total Quality Development, ASME Press, NewYork, NY, 1994.

Consumer Reports, A Publication of Consumers Union, 1995Consumer Reports, Electric Frying, December, 1997.Dixon, J. and Finger, S. Research in Engineering Design 1(2),

1989.Foster, R. Innovation: The Attacker’s Advantage. Summit

Books, New York, 1986.Gabriele, G. Employing reverse engineering projects in a

capstone design course, Department of MechanicalEngineering, Aero. Eng. and Mechanics, RPI, Troy, NewYork, 1994.

Grifn, A. and Hauser, J. R., The voice of the customer,Marketing Science, 12(1), 1993.

Hubka, V. et al. Practical Studies in Systematic Design,Butterworths, London, 1988.Ingle, K. Reverse Engineering, McGraw-Hill, New York,

1994.Ishikawa, K. Guide to Quality Control. 10th ed, Asian

Productivity Organization, Hong Kong, 1992.Jensen, D., Wood, K. L. and Murphy, M. D. Evaluation and

renement of a restructured introduction to engineeringdesign course using student surveys and MBTI data,ASEE Annual Conference, Seattle, WA, June 1998, (inpress).

Lefever, D. Integration of design for assembly techniques intoreverse engineering, Masters’ Report, Department of Mechanical Engineering, The University of Texas,August 1995.

Lefever, D. and Wood, K. Design for assembly techniques inreverse engineering and redesign, 1996 ASME DesignTheory and Methodology Conference, Paper No.DETC96/DTM–1507.

Little, A. and Wood, K. L. Functional analysis: A fundamentalempirical study for reverse engineering, benchmarkingand redesign, 1997 ASME Design Theory and Methodol-ogy Conference, Paper No. DETC97/DTM-3879.

Miller, J. New shades of green, MS Thesis, The University of Texas, Austin, TX, 1995.

Miles, L. Techniques of Value Analysis and Engineering,McGraw-Hill, New York, 1972.

Otto, K. Eine Methode zur Denition technischer Produktan-forderungen, Konstruktion, 49, 1997, pp. 19–25. (Seealso Forming Product Design Specications, 1996 ASMEDesign Theory and Methodology Conference, Paper No.DETC96/DTM–1517).

Otto, K. and Wood, K. L. Conceptual and conguration designof products and assemblies, ASM Handbook, Materials

Selection and Design, Vol. 20, ASM International,Materials Park, OH, 1997, pp 15–32.

Otto, K., Wood, K., Bezdek, J., Murphy, M. and Jensen, D.Building better mousetrap builders: Courses to incremen-tally and systematically teach design, 1998 ASEE AnnualConference, ASEE, June 1998.

Pahl, G. and Beitz, W. Engineering Design, The DesignCouncil, London, 1984.

Papalambros, P. and Wilde, D. Principles of Optimal Design,Cambridge University Press, 1988.

Phadke, M. Quality Engineering Using Robust Design,Prentice Hall, Englewood Cliffs, NJ, 1989

Pugh, S. Total Design, Addison-Wesley, New York, NY, 1991Sferro, P., Bolling, G. and. Crawford, R. Omni-engineer,

Manufacturing Engineering, June 1993, pp 60–63.Sheppard, S. and Tsai, J. A note on mechanical dissection withpre-college students, Department of Mechanical Engineer-ing, Design Division, Stanford University, 1992.

Shimomura, Y. et al. Representation of design object based onthe functional evolution process model, Proceedings of theASME DTM Conference, Boston, MA, September 1995.

Stone, R. B., Wood, K. L. and Crawford, R. H., A heuristicmethod to identify modules from a functional descriptionof a product, ASME Design Theory and MethodologyConference, Atlanta, GA, 1998.

Thorton, A. and Meeker, D., Benchmarking within the ProductDevelopment Process, Proceedings of the ASME DesignTheory and Methodology Conf., Boston, MA, September1995, pp 639–648.

Ullman, D. The Mechanical Design Process, McGraw-Hill,NY, 1992.

Ullman, D. A new view of function modeling, InternationalConference on Engineering Design, ICED ’93, TheHague, August 17–19 1993.

Ulrich, K. and Eppinger, S. Product Design and Development,McGraw-Hill, NY, 1994.

Urban, G. and Hauser, J. Design and Marketing of NewProducts, Prentice-Hall, Englewood Cliffs, NJ, 1993.


product evolution _ a reverse engineering and redesign metho

Documents