Transcription
DESIGN FOR NONASSEMBLY: CURRENT STATUS AND FUTURE DIRECTIONSSangjin Jung1, Rianne E. Laureijs2, Christophe Combemale2, and Kate S. Whitefoot 3*1Departmentof Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213of Engineering and Public Policy, Carnegie Mellon University, Pittsburgh, PA 152133Department of Mechanical Engineering; Department of Engineering and Public Policy, Carnegie Mellon University, Pittsburgh, PA 152132DepartmentABSTRACTNonassembled products, which are produced from a rawmaterial and post-processed to a final form without any assemblysteps, form a large and potentially growing share of themanufacturing sector. However, the design for manufacturingliterature has largely focused on assembled products, and doesnot necessarily apply to nonassembled products. In this paper, wereview the literature on design for nonassembly (DFNA) and thebroader literature on design for manufacturing that has designguidelines and metrics applicable to nonassembled products,including both monolithic single-part products and nonassemblymechanisms. Our review focuses on guidelines that apply acrossmultiple manufacturing processes. We identify guidelines andmetrics that seek to reduce costs as well as provide differentiatedproducts across a product family. We cluster the guidelines usinglatent semantic analysis and find that existing DFNA guidelinesfall into four main categories pertaining to: (1) manufacturingprocess, (2) material, (3) tolerance, and (4) geometry. We alsoidentify existing product family metrics that can be modified fornonassembled products to measure some aspects of thesecategories. Finally, we discuss possible future research directionsto more accurately characterize the relationships between designvariables and manufacturing costs, including investigatingfactors related to the complexity of operations at particularprocess steps and across process steps.1. INTRODUCTIONNonassembled (NA) products makeup a substantial share ofthe manufacturing sector and have the potential to growconsiderably in the future. NA products are produced from a rawmaterial and post-processed to a final form without any assemblysteps [1, 2]. We estimate NA products to account forapproximately 50% of U.S. manufacturing as measured by valueof shipments [3] (detailed information is provided in AppendixA). Examples of specific NA products include: products made ofraw materials such as fabricated metal products (e.g., fasteners,brackets, fixtures, beams, wire, springs), glass products, wearparts, and other durable goods; monolithic plastic goods;ceramics; food and beverages; as well as chemicals andpharmaceuticals [3]. Moreover, advances in materials andproduction processes—such as additive manufacturing (AM),large die castings, and parts consolidation—have enabled moreproducts to be manufactured with reduced or no assembly [4-7].Therefore, one might expect nonassembled manufacturing toincrease in the future.While there is a long history of developing design formanufacturing (DFM) guidelines [5, 8-14], much of it is focusedon assembled products, and does not necessarily apply to NAproducts. For example, multiple DFM studies have recommendedusing symmetry to reduce errors and time of assembly, such asdesigning a part to be symmetric around the axis of insertion so*Corresponding Author: Kate S. Whitefootthat it cannot be assembled incorrectly [13, 15, 16]. However,different design elements of NA products may be important toreducing the costs of manufacturing steps. For example, for amolded, forged, or additively manufactured part, asymmetry maynot affect costs, but whether or not the design has overhangs cansignificantly affect costs and ease of manufacturing [17].NA products also pose unique challenges in productdifferentiation. Assembled products often achieve differentiationby incorporating modular components with common interfaces,which can be swapped out to change various product attributeswhile maintaining the same assembly step. In contrast, achievingdifferentiation in NA products requires changing inherentattributes of the entire product, such as material, geometry,tolerance, size, and treatment [18]. Unlike assembled products,achieving this differentiation of attributes in NA productsnecessarily requires a process change, which may requireadditional labor, machines, and/or tooling [18].This paper reviews the literature on design for nonassembly(DFNA) and the broader literature on DFM that has designguidelines that can apply to this domain. In this review, weexamine cost-related guidelines—including recent literature ondesign guidelines for nonassembly mechanisms—as well asguidelines and metrics related to product differentiation andproduct variety. We then synthesize the identified DFNAguidelines into four main categories using latent semantic analysisand hierarchical clustering.We find a major gap in the literature dealing with designguidelines for nonassembly: while general guidelines forassembly are widespread [13-15] and offer helpful generalinsights to a wide range of cases, nonassembly presents uniquechallenges that have thus far been addressed only in narrowproduct or process-specific contexts, for example steel pouring [4]or metalorganic chemical vapor deposition [19]. In reviewing thebroader DFM literature, we see that many DFM guidelines focuson reduction of parts, design of part interfaces for assembly, anddesign for assembly steps that are not applicable to NA products.In this paper, we seek to draw out some of the general designprinciples underlying the presently disunified literature and tobegin developing a nonassembly equivalent to the broad designguidelines that have benefited assembled products.We find that existing design guidelines that apply to NAproducts focus on product size and shape, material, tolerances,and post-processing steps to reduce costs and increase variety. Wealso identify some existing product family metrics that can bemodified to fit NA products. In addition, we discuss the limitationof the DFNA guidelines and directions for future work associatedwith developing a set of guidelines and metrics for capturing therelationships between design variables and production complexityand cost.Paper #: MD-20-10911
2. DESCRIPTION OF DFNANA production takes raw materials as inputs, and transformsthem into a product (whether intermediate or final) throughfabrication and post-processing steps that do not include assembly[1, 2]. For the purpose of this paper, we adopt the MerriamWebster definition of “assemble” as “to fit together the parts of”[20]. NA products can either be sold as a standalone product (e.g.cutlery, glassware, plastics, pharmaceuticals) or sold to acustomer who uses it as a subcomponent of another assembledproduct (e.g. fasteners, auto body components, buildinginfrastructure including beams and joists, and other monolithicsub-components). For example, a formed plastic cup holderproduced by company A that sells the product to automotivecompany B that assembles it into a vehicle for final sale would fitthis classification (Figure 1). Examples of NA products includefabricated monolithic metal parts such as hand tools, cutlery,fasteners, brackets, fixtures, beams, wire, springs, automotive andmachinery components, and other such products; plastic injectionmolded products; glass products; ceramics; chemical andpharmaceutical products; additive manufactured products;processed food and beverage production; and other products thatdo not require assembly. Common processes applied to NAproducts include forging, stamping, pressing, grinding, sintering,extrusion, molding, and other processes that form raw materialsinto shapes, remove or add material, and treat the material tochange its properties [18].While many NA products are monolithic parts, in specialcases, some multi-component parts may fit the definition of NAproducts described above. One case is when a component isembedded into the part during fabrication (e.g., composite fibers,embedded sensors [21-23]). A second case is nonassemblymechanisms, in which distinct components are fabricated together[7, 24, 25]. In both of these cases, the components are built intothe product during the fabrication process in contrast withassembled products, in which multiple discrete components arefabricated separately and fit together.Figure 1. DFNA applies to both products that are sold as finalgoods, and products that are sold as intermediate goods and laterassembled into final goods. We review guidelines that are relevantto both of these cases. Images from [26-28].The concept of DFNA is the practice of designing a productfor fabrication and post-processing that seeks to minimize costsand optimize product attributes (including the variety of thoseattributes in a product family) to maximize profits within theconstraints of existing production technology [1, 18]. DFNA willnaturally draw on principles in the broader DFM literature,although some of these principles may apply differently to NAproducts, and result in distinct DFNA design guidelines. We useFigure 2 to illustrate the relationship between design for assembly(DFA) and DFNA. As the figure shows, there is a region of designguidelines that apply to both NA and assembled products (e.g.,using low-cost materials that meet functional requirements,minimizing tooling changeover, design for low-labor-costoperations). We review these strategies in this paper, as well anyguidelines that apply uniquely to DFNA.Figure 2. There is overlap between DFA and DFNA designguidelines.3. SCOPE OF LITERATURE REVIEWIn this paper, we review design guidelines for NA products aswell as DFM guidelines and metrics that can apply to NAproducts. In this review, we focus on guidelines that apply acrossmultiple manufacturing processes. There are many articles thatfocus on design guidelines for a single manufacturing process(e.g., injection molding or additive manufacturing), which arehighly specific to machine capabilities for the specific process.For example, Bralla [4], Boothroyd et al. [5], and DRMAssociates [29] provide very detailed guidelines associated withspecific manufacturing processes (e.g., sheet metal, casting,injection molding, stamping). Anderson [6] also introducespecific design guidelines for castings and molded parts, plastics,and sheet metal. Meisel et al. [30] suggests some specific designguidelines for metal-based AM, and Booth et al. [31] providedesign guidelines and a worksheet for AM assessing complexity,functionality, material removal, unsupported features, stressconcentration, tolerances, and geometric exactness. We excludethese process-specific guidelines from our review in order tosynthesize more general guidelines for DFNA.There is a significant body of literature that discusses boththe benefits and tradeoffs associated with designing a product orsystem to be manufactured without assembly (e.g., using partsconsolidation) as opposed to assembled from multiplecomponents [32-34]. We do not review this literature here, butinstead focus on guidelines for design for manufacturing after thedecision to produce a non-assembled product has been made.There is currently only a small body of literature focused ongeneral design guidelines and methods that are specific to NAproducts [1, 2, 18]. However, a subset of existing DFM guidelinesare applicable to NA products as well as assembled products. Our2
approach to collecting and synthesizing a body of literaturerelevant to DFNA was to review not only NA-specific guidelines,but also the broader DFM literature, identifying guidelines thatare applicable to NA products.This review consisted of three stages: literature collection,identification of NA-relevant guidelines, and synthesis. For theliterature collection, we searched for articles in English on Scopusand Web of Science that have “design for manufacturing” or“design for manufacture” in their title and “guidelines,” “rules”,“design principles”, “design for manufacture”, or “design formanufacturing” in the abstract or title. This search resulted in over200 papers published between 1978-2019. The majority of thesearticles are published by authors in the United States or Europe,although articles published by authors in Asia, South America,and Australia were also represented. We then reviewed the designguidelines in these papers and reduced them to the most relevantset based on two criteria: (1) the guidelines are relevant to NAproducts, and (2) the guidelines are general and broadlyapplicable. Based on the first criteria, guidelines were eliminatedif they solely focused on conditions requiring multiplecomponents assembled together. Specifically, they wereeliminated if they focus on: (1) part count reduction, (2) design ofinterfaces for assembly, and (3) design for assembly steps. Basedon the second criteria, guidelines were eliminated if they onlyapplied to one specific NA manufacturing method—such asinjection molding or additive manufacturing—but did not applyacross any other manufacturing methods. The remaining set ofguidelines are reviewed in this paper. Finally, we synthesized theidentified guidelines using latent semantic analysis andhierarchical clustering analysis to group them according to thesimilarity of the guidelines and keywords used in them.4. COST-BASED GUIDELINES FOR DFNA4.1 General DFNA GuidelinesMany research articles and books have introduced generaland specific DFM guidelines [4-6, 13-15, 29, 35, 36]. The mainpurpose of many DFM and DFA guidelines is to modify a designto carry out the same functions while reducing production costs[37]. The guidelines that we identify as applicable to DFNA andsummarize in this section can be readily associated withminimizing production costs. For example, design for low-costlabor operations, reduce weight to reduce costs, and design forgeneral purpose tooling (because that will reduce tooling costs).The complete list of the specific DFNA-relevant guidelines areincluded in Appendix B.Stoll [15] suggests ten DFM principles, two of which areapplicable to DFNA: (1) design parts for ease of fabrication and(2) minimize handling. In addition, Kirkland [12] introducesseveral kinds of general guidelines for DFM, some of which canbe applied to NA products (e.g., optimize raw material selectionand process selection). However, the remaining design guidelines(e.g., develop a modular design, avoid separate fasteners, andminimize assembly directions) are only able to be applied to thedesign of assembled products. Adachi et al. [38] also suggestsgeneral design guidelines for DFM, some of which are applicableto NA products that are associated with the production processand facility (e.g., synchronize with development of productionfacilities, minimize impacts on production processes).Bralla [4] published the Design for ManufacturabilityHandbook, which introduces a wide range of general DFMprinciples that apply to multiple manufacturing processes andsuggests numerous design considerations and guidelines fordifferent types of parts and products. Unlike the prior research onDFM in the literature, these DFM guidelines can be generallyemployed to the design of both assembled and NA products.These DFM guidelines are primarily focused on designsimplification, dimensioning, cost, and manufacturing processes.Boothroyd [14], Edwards [16], Swift and Booker [39], vanVliet and van Luttervelt [40], and Luo et al. [41] have suggestedmore extensive and detailed DFM guidelines associated withmaterial, cost, manufacturing process, standardization, tolerance,drafting, geometry, and size, many of which are applicable toDFNA. These guidelines have been applied to different types ofproducts in firms [14, 16, 39-41]. For example, a heater core coverfabricated using injection molding by a U.S. automotivemanufacturer was redesigned following the guideline “aim atuniform wall thickness” [14, 42]. By changing the geometry toachieve uniform wall thicknesses, the consequent cycle-time wasreduced in the injection molding process and both tooling andprocessing costs were lowered [14, 42]. In addition, the numberof cavities was reduced, and the production rate of heater corecovers increased. As a result, the total manufacturing cost of theheater covers was reduced by 33% [14, 42]. The general DFNAguidelines suggested by Edwards [16], Swift and Booker [39], vanVliet and van Luttervelt [40], and Luo et al. [41] are described inAppendix B.Similar to previous studies on DFM guidelines, Anderson [6]introduces nine kinds of high-level design guidelines for DFM,five of which are applicable to DFNA: (1) adhere to specificprocess design guidelines, (2) design for fixturing, (3) minimizetooling complexity by concurrently designing tooling, (4) specifyoptimal tolerances for a robust design, and (5) specify qualityparts from reliable sources. Anderson [6] also suggests generaldesign guidelines for fabricated parts as follows, and they can beemployed to conduct DFNA that applies across multiplemanufacturing processes: Choose the optimal processing.Design for quick, secure, and consistent work holding.Use stock dimensions whenever possible.Optimize dimensions and raw material stock choices.Design machined parts to be made in one setup.Minimize the number of cutting tools for machined parts.Avoid arbitrary decisions that require special tools and thus slowprocessing and add cost unnecessarily.Choose materials to minimize total cost with respect to postprocessing.Concurrently design and utilize versatile fixtures.Understand work-holding principles.Understand tolerance step functions.Specify the widest tolerances consistent with function, quality,reliability, safety, and so forth.Be careful about too many operations in one part.Concurrently engineer the part and processes.Do not over-specify surface finishes.Reference each dimension to the best datum.Compared to the previous studies, they contain differenttypes of guidelines associated with manufacturing process, workholding, material, dimension, cost, tolerance, and surface finish,but some of these guidelines are very similar to the guidelines inthe other studies (e.g., choose materials to minimize total cost).3
As shown in Appendix B, many similar or identical guidelines areidentified in the literature.4.2 Design for Nonassembly MechanismsEnabled by additive manufacturing (AM), nonassemblymechanisms are produced by fabricating multiple componentswith joints between them without requiring any assembly steps [7,21, 24, 25]. While the current literature exclusively focuses onguidelines for nonassembly mechanisms that are producedthrough AM, they could in theory be manufactured throughseveral other means (e.g., forming with post-processing materialremoval). Thus, we include this literature in our review, focusingon the guidelines that could apply not only to AM but otherprocesses as well.The design of NA mechanisms poses unique designchallenges for joints, and thus several studies have notedimportant considerations for joint design and its feasibility (i.e.,the product will not function if the guidelines are not followed)[7, 24, 25, 43]: Fabricated joints must have sufficient clearance to avoid fusing orcatching. Fabricated joints cannot have such great clearance that they failmechanically to connect. Minimize friction in the joint and other strain. Minimize the tolerance of each joint to improve the positionaccuracy.One advantage of NA mechanisms is their wide range ofgeometric possibilities outside the limitations of assembly. Forexample, built-in elements that are internal to the nonassemblybut would be impossible to insert into the product because of itsgeometry can be created during fabrication. However, thegeometric design of NA mechanisms also raises additionalguidelines associated with mechanism performance andmanufacturing cost [7]: Pursue geometries that minimize support structures. Pursue geometries whose support structures are easily removable. Minimize interfaces for material deposition or shaping that couldlead to trapping or deterioration of material and underminemechanism performance.With respect to the general design of multi-articulated NAproducts, Cuellar et al. [44] suggest ten guidelines that fall intothree broad types: General: integration of parts functionality and facilitating supportsand interfaces in production. Play: minimizing lost motion. Stress: distributing and managing applied load.5. PRODUCT DIFFERENTIATION-BASED DFNAThe guidelines discussed above are motivated by reducingthe costs of producing individual products. In addition tominimizing costs, many firms also seek to increase their marketshare and profits by providing differentiated products [45, 46].Product family design, defined by Jiao et al. (2007) as “aconceptual structure and overall logical organization ofgenerating a family of products by providing a generic umbrellato capture and utilize commonality”, provides a strategy forreducing costs while increasing product differentiation [47]. Inassembled products, this differentiation can be achieved throughthe interchangeability of unique components or modules viashared interfaces [45-48]. Interchangeability, however, does notserve to reduce costs for NA products because they either have nosub-modules or components, or—such as in the case ofnonassembly mechanisms—subcomponents are fabricatedtogether simultaneously. Therefore, different strategies arerequired to minimize costs of a differentiated NA product family[18]. We review product family design strategies for NA productfamilies, and metrics that may be applied to NA product portfoliosto reduce costs while preserving differentiation. We then focus oncharacterizing NA product attributes that may impact NA productfamilies.5.1 Product Family Design StrategiesWhile most product family literature focuses on sharingcommon components [49-54], which will not necessarily serve toreduce costs for NA product families, there are some existingproduct family guidelines that can apply to NA products. First,both Robertson and Ulrich as well as Meyer and Dalal have notedthat products that can share common production assets, processes,and systems can help facilitate production flexibility [1, 55]. Inthe context of a product family, differentiated NA products can bedeveloped based on platforming a set of common elements (e.g.,materials) while creating variants using other elements (e.g.,treatments). For instance, the manufacturing process of integralfilms requires coatings of fifteen different fluid layers such as acidpolymer layer, image receiving layer, and blue, green, and redsensitive emulsions, and the family of integral film products shareraw materials for each layer and multiple steps within their entireproduction process [1].In addition, multiple authors propose scale-based strategiesfor creating product families [56-58], which may also be appliedto NA products. Scale-based strategies define parameters of theproduct design (e.g., key dimensions) which can be scaled up ordown in magnitude to achieve multiple product variants within afamily [58]. Simpson [57] defines a scale-based product family asone which is based on stretching or shrinking along a dimensionto accommodate product variety within product platforms.Simpson et al. [56] also suggest a metric for developing scalebased product platforms, the Product Platform ConceptExploration Method (PPCEM), which takes in market factors,design parameters, and scaling variables and develops a scalebased product platform output.Our review yielded only two studies that propose strategiesspecifically for NA product family design. Meyer and Dalal focuson engineers’ evaluation of shared process and technologiesacross NA products, finding platforms where there is sharing andreuse [1]. Furthermore, they suggest that the best way to measureproduct platforms for NA products is through platform efficiency,a measure which they propose as being primarily based onmanufacturing, tooling, and engineering costs of various productgenerations [1]. Moving beyond shared components andresources entirely, Laureijs et al. [18] seek to define the buildingblocks of commonality specifically for NA products, and proposea theory of NA product family design that focuses on commondesign variables across NA products. These common NA designvariables are: material, geometry, tolerance, size, and postprocessing treatment steps (e.g., material coatings, or heattreatment). One example is a product family based on commonproduct geometry and size but with varying materials andcoatings. Diversity in these variables may affect the flexibility ofthe line, thus impacting total manufacturing costs from producing4
a product portfolio on the line [18]. Figure 3 recreates theproposed theory of how variety in each of these design variablesimpacts manufacturing operations that affect costs.Load/Unload/SetupMaterial PriceScrapYieldCycle TimeFixtures/ToolingBatch SizeAs youIncreaseVariety In:EquipmentPrice/QuantityYou may affect the process in the following re 3. As proposed by Laureijs et al. [18], when varietyincreases in each of the proposed NA product attributes, variousparameters of the production process are affected in a way thatcan influence costs.5.2 Product Family Metrics for DFNASeveral metrics have been created to allow engineers toevaluate the variety and commonality within a product family inorder to reduce costs while maintaining differentiation [1, 49-53,59-64]. Most of these metrics focus on the commonality ofcomponents in assembled products [49-53, 59-64]. For example,the Percent Commonality Index quantifies the percentage ofshared components, connections between components, andassembly stations across a product family [61].Lager [2] suggests a conceptual platform-based designframework which integrates product platform, process platform,and raw material platform for NA product families, and theframework can help identify the commonality related to designrequirements, functionalities, production processes, and rawmaterials. The conceptual framework does not provide detaileddesign methods or metrics to evaluate the characteristics of NAproduct families such as platform efficiency and commonality.One article measured platform efficiencies in a case of NAproducts [1], which focuses on product R&D, manufacturing, andtooling costs across product generations. The metric used iscalculated as:𝐸𝐶 𝑀𝑅𝐶𝑀𝑅(1)where p is the index of a single derivative product; g is thegeneration of the product line; C is the product engineering costsattributable to architecture and platform development, orderivative product development based on these platforms withina product family; M is the manufacturing engineering costs; andR is the retooling and related capital costs for manufacturingequipment. A smaller value of Ep means higher platformefficiency. Ep only considers costs to capture the degree ofplatform efficiency within NA product families. The metric doesnot map costs to specific design variables (e.g., choices oftolerances that increase yield losses, machine set-up, orcalibration time) to inform design decisions.Despite the lack of design-relevant metrics for NA productfamilies, some of the existing metrics for assembled products canbe used or adapted to evaluate the degree of product variety orcommonality in NA product families. Metrics such as the Non-Commonality Index (NCI) [62] and the Product Family PenaltyFunction (PFPF) [63] have been developed to evaluate thedispersion of a product family’s design variables for scale-basedproduct families. These can be applied to NA product families tocapture commonality or dispersion of non-categorical designattributes such as size, tolerance, and certain geometric measures.In theory, these metrics could be extended to binary indicatorvariables of other attributes such as material or treatment type.Two other existing product-family metrics were identifiedthat could be modified for use with NA products. The first is theProduct Line Commonality Index (PCI) developed by Kota et al.[53]. This metric considers the size, shape, material, andmanufacturing processes for each product. PCI also evaluates thecommonality on assembly and fastening schemes, but thiscommonality factor can be removed for use in DFNA. PCI wasoriginally utilized to evaluate the commonality of multiplenumbers of parts in a product family, but can be adjusted tomeasure the commonality across NA products as follows:PCI𝑓𝑓𝑓11𝑛1𝑛100(2)where n is the number of NA products in the product family; f1 isthe size and shape factor; f2 is the materials factor; and f3 is themanufacturing processes factor. For example, if the value of PCIis close to 100, it means that the NA product family has highcommonality in terms of size and shape, materials, andmanufacturing processes. In addition, it can be possible to adddifferent types of commonality factors (e.g., tolerances) in designand fabrication processes for NA products. The second is theComprehensive Metric for Commonality (CMC) developed byThevenot and Simpson [49], which considers size, shape,material, assembly and fastening schemes, manufacturingprocess, and total cost within a product family. The greater valueof CMC represents higher commonality (e.g., when a productfamily has more common size, shape, material, assemblyschemes, and manufacturing process, and when the total costs foreach product are lower). Like PCI, this commonality metric canbe adjusted to capture the degree of commonality for NA productfamilies by dropping the commonality factor on assembly andfastening schemes.The existing product family metrics in the literature have notbeen developed for NA products, so there are limitations whenapplying them to NA products. Existing metrics may not be ableto capture the key differentiating characteristics across an NAproduct family such as the degrees of comm
specific manufacturing processes (e.g., sheet metal, casting, injection molding, stamping). Anderson [6] also introduce specific design guidelines for castings and molded parts, plastics, and sheet metal. Meisel et al. [30] suggests some specific design guidelines for metal-based AM, and Booth et al. [31] provide design guidelines and a .
