Additive manufacturing (AM) offers significant opportunities for product innovation in many fields provided that designers are able to recognize the potential values of AM in a given product development process. However, this may be challenging for design teams without substantial experience with the technology. Design inspiration based on past successful applications of AM may facilitate application of AM even in relatively inexperienced teams. While designs for additive manufacturing (DFAM) methods have experimented with reuse of past knowledge, they may not be sufficient to fully realize AM's innovative potential. In many instances, relevant knowledge may be hard to find, lack context, or simply unavailable. This design information is also typically divorced from the underlying logic of a products' business case. In this paper, we present a knowledge based method for AM design ideation as well as the development of a suite of modular, highly formal ontologies to capture information about innovative uses of AM. This underlying information model, the innovative capabilities of additive manufacturing (ICAM) ontology, aims to facilitate innovative use of AM by connecting a repository of a business and technical knowledge relating to past AM products with a collection of knowledge bases detailing the capabilities of various AM processes and machines. Two case studies are used to explore how this linked knowledge can be queried in the context of a new design problem to identify highly relevant examples of existing products that leveraged AM capabilities to solve similar design problems.

Introduction

Additive Manufacturing.

Additive manufacturing (AM) comprises a range of manufacturing technologies in which parts are constructed by progressively adding material layer by layer, rather than by removing material as in conventional machining. Though initially confined to plastics, additive manufacturing has expanded to support an increasing array of methods for the fabrication of parts in plastics, ceramics, metals, and even biological materials. These widening capabilities, along with several advantages conferred by layered construction, have led to the uptake of AM in multiple industries, including rapid prototyping, aerospace, and medicine.

The advantages of additive manufacturing are typically described in terms of its ability to economically produce products with several types of complexity. AM processes, for example, can create structures that have a high degree of shape, functional, and hierarchical complexity [1]. Material complexity, such as variations in material composition or properties throughout a part, has also been highlighted as a major advantage [2]. These technical advantages have significant potential to both disrupt current market supply chains and business practices [3], as well as offer exciting new opportunities for innovation in products and services [46].

While often advantageous, AM does have several notable drawbacks. These include dimensional accuracy, mechanical properties, cost, economies of scale, and workflow [7]. As a result, deployment of AM technologies requires careful consideration of both the needs and economics of a design. An additional issue is the sheer array of processes, their rapidly changing nature because of both academic research and new product development, and a lack of AM domain expertise among many engineers. Cumulatively, these factors mean that AM processes offer a significantly expanded design space, while simultaneously introducing major knowledge hurdles that may hinder its effective utilization by most designers.

Design for Additive Manufacturing.

The dramatic expansion of fabrication capabilities and the desire to use them effectively has led to a growing recognition of a need for methods and tools to fully exploit the design freedom offered by AM processes. These approaches are collectively referred to as design for additive manufacturing (DFAM) methods.

Design for additive manufacturing methods comprise a range of design activities, including decision making [6,8,9], process planning [1013], manufacturability assessment [1416], and optimization [12,1721] among others. A subset of DFAM methods focus on creative processes that help designers consider new possibilities made feasible by AM to develop innovative products that would be infeasible without AM. Several approaches have been proposed. A straightforward approach is to introduce AM design concepts through documentation or expertise. Compared to novice groups, student designers with these resources have been found to generate more and better design ideas [22]. Broad design guidelines for AM coupled with traditional design offer another approach [5]. However, these recommendations may not be sufficient to significantly alter designs from what could be obtained with traditional design methods. Functional consolidation offers a more direct mechanism to achieve product innovation. In such approaches, functional models of existing parts are used to identify places where single function features might be aggregated into multifunctional ones [23]. Though a potentially useful approach, few authors propose methods for how best to accomplish this. One proposed approach is to use functional similarly [24]. In the proposed DFAM technique, three-dimensional functional graphs corresponding to where a function is needed in the part are compared through unspecified means to a repository of such graphs. This repository can then, in theory, be used to infer design solutions [24]. However, the study does not provide details of how such knowledge bases might be implemented.

Design repositories like those in Ref. [24] have been proposed in several other forms, with various means of accessing relevant information. For example, Bin Maidin et al. [25] have proposed the use of an AM feature database, detailing four broad types of customer requirement (fit, functionality improvement, consolidation of functions, and aesthetics) and mapping these to a set of features that can be manufactured using AM [25]. However, as implemented, the database might only be able to provide broad, nonspecific information about how various features were used in the past, and what problem context led to their use. Related work coupled a similar feature database with images and explanation to form a dossier that might help inspire creativity [26]. Another study used a similar approach implemented in software [27]. A related research has proposed combining catalogues of AM features with economic evaluation, design optimization, and functional consolidation [28]. This approach was later implemented in a software platform that used a “semantic network” to link a set of AM properties relating to several types of complexity to various generic design values. These relations were then used to annotate a set of existing products [29]. While these inspiration-based DFAM methods are promising, it is unclear whether they can be easily extended to describe more complex utilization of AM capabilities, or that they can be enhanced with knowledge from additional domains.

While research to date has proposed early AM consideration, feature-based knowledge, and functional representation that may be useful for innovating in AM, the exact means by which these techniques might be used effectively remains an open question. Therefore, more sophisticated systems may be needed to allow effective ideation and feasibility checking, especially when multiple processes with varying capabilities and limitations are considered. One potential solution might be the use of information models for knowledge capture and reuse.

Knowledge Capture and Reuse in Engineering.

Knowledge management, which consists of strategies to capture and render knowledge reusable, is highly important to the engineering field. A range of approaches exist, from simple creation of documentation to the development of highly controlled, well-defined terminologies, and information models such as data schemas that capture knowledge [30]. A very versatile knowledge management approach is the use of ontologies, information models consisting of formally defined hierarchies of entity types describing some domain of interest, coupled with well-defined relations between types and axioms expressing fundamental domain knowledge. Developed properly, ontologies are shareable, extensible, and interoperable, allowing rapid reuse of past knowledge to model new domains [31].

Several ontologies have been developed specifically to manage knowledge in the engineering domain. These include standardized terminologies to describe products [32], functional representation languages, and ontological implementations of them [33,34], as well as ontologies of engineering models [35], decision-making [36], sustainable design [37], and the manufacturing domain [38]. Others have introduced modular and cross-domain ontologies for the purpose of aiding in specific engineering activities. Past efforts include the e-design framework, a modular set of ontologies that aim to coordinate information relating to engineering analysis [35], engineering optimization [39], design decisions [40], design innovation [34], engineering relationships [41], laminated composites [42], product innovation [43], biomodels [44], medical device conceptual design [45], and ergonomic design principles [46]. Additive manufacturing has also been an area of interest for knowledge management through ontologies and other information models [42]. Past efforts have included an ontology to support process planning for additive manufacturing [47], a partially realized framework in which ontologies linked to an AM-related knowledge base of features and processes [11], a DFAM framework using an ontologies to track build success and failure, and a data schema for AM [14,48].

Although a broad range of ontologies have been proposed for knowledge management in both engineering design broadly and in additive manufacturing specifically, there does not appear to be an ontology that deals comprehensively with product realization, or more specifically product innovation achieved through DFAM. Moreover, existing engineering domain ontologies often are not actually inoperable, and by extension are less extensible and reusable. Relatively little work to date has focused on how to effectively locate and use knowledge gained through past innovations in AM. Were this possible, such knowledge might be used to spur innovations in other domains, or to solve market or design problems that have features in common with past use cases. To the extent that the area has been explored, only relatively simplistic knowledge structures have been employed, resulting in potentially substantial loss of knowledge and context. Moreover, few authors have focused on both the technical and enterprise aspects of innovation.

In this paper, we present a DFAM framework to facilitate reuse of past knowledge from the manufacturing domain and enterprise domain and to enable a query-based for designing products based on past successes. It aims to provide a formal information model linking knowledge captured from the enterprise, design, and manufacturing domains so that a designer can tag a knowledge base of past additive manufacturing for detailed, expressive semantic searching.

Methods

Rationale.

We argue that in order to realize radical innovation with AM, the design process must begin to consider DFAM as early in the design process as possible. The method and realization presented in this paper are based on the insight that the innovative potential of AM stems from a set of expanded capabilities, which in turn enable value to be delivered to a product's market in ways that would be difficult, economically nonviable, or simply impossible using traditional manufacturing methods. If information relating to capabilities and how they have been leveraged previously to generate value in past contexts were easily accessible, it could then help to provide the insights needed to innovate in another context. Effective knowledge management might thus offer significant advantages.

This work uses a suite of ontologies to link knowledge bases reflecting past uses of AM with information about machines, processes, and manufacturing capabilities. The ontological approach provides a domain neutral terminology and aids in the creation of an information model that links these knowledge bases to one another to reflect a rich body of AM domain knowledge. It also provides a machine-readable formalization of knowledge that supports automated reasoning and powerful semantic queries that can be used to extract useful information from these knowledge bases. Thus, queries to prior knowledge may be used to assist the design process by using past AM knowledge to assist in ideation and problem solving. This paper details the development of an ontology that enables knowledge capture from past uses of AM that have been reported in the academic literature or have been released in the commercial market. This ontology is used to capture a knowledge base and provide a means to query said knowledge based on a designer's specific product needs.

Development of the Innovative Capabilities of Additive Manufacturing Ontology.

Throughout this section, boldface type will be used to indicate ontology classes, while bolded and italicized type will indicate relations between classes. The innovative capability of additive manufacturing (ICAM) ontology was constructed from a combination of existing and newly defined ontologies to form a suite of modular ontologies implemented in the Web Ontology Language (OWL) [49]. In addition to aligning domain ontologies, the development process also led to the creation of two ontologies representing information relating to business models and manufacturing capabilities. The former was included out of a recognition that innovation comprises both design and enterprise knowledge. Similarly, the use of AM capabilities to disrupt existing markets and supply chains was considered an important aspect of ICAM. We envision ICAM and related ontologies as providing a linkage layer between disparate design, manufacturing, and enterprise considerations so that information from these domains might be reused to foster creativity.

Identification of a Top-Level Ontology.

Upper-level ontologies provide a formal definition of a set of entities to which all other (nonrepresented) entities can be considered subtypes and so offer an abstract model of information models that utilize their classification structure. An upper-level ontology was deemed necessary to facilitate the extensibility and reusability of ICAM for future projects, as well as to impose a well-documented, formal information model on the various knowledge domains that are unified in ICAM. The basic formal ontology (BFO) [50] was chosen for this project due to its small class structure, extensive use in other scientific domains, readily available guidelines and training material, and considerable success it has enjoyed in the biomedical field. BFO breaks all entities into two types, continuants, or entities without temporal parts, and occurrents, which comprise things like processes and events, which have temporal parts that unfold in time (Fig. 1). Continuants are further divided into those that are independent and cannot inhere in others, those that must inhere in a single independent continuant such as intrinsic qualities or realizable dispositions (specifically dependent), and those that must inhere in some independent continuant that can change over time (generically dependent), such as information [51].

Use of Existing Ontologies.

Several existing ontologies were used to expedite the creation of ICAM. To that end, two BFO conformal ontologies were used to provide the higher level information model that serves as the backbone of ICAM. The first, the relations ontology (RO) [52], provides a core set of property relations between entities within BFO while defining no additional classes of entity. Though more specific properties and subproperties can be added as appropriate for domain specific relations, the property relations defined in RO can be thought of as a core set of properties through which most relations between entities will be expressed. In the context of ICAM, this is convenient, as it greatly eases querying the ontology. The second BFO conformal ontology that we reused is the information artifacts ontology (IAO) [53]. IAO provides a formal treatment of information content entities (ICEs), which comprise everything from text and figures to models, and directive expressions. As ICAM deals extensively with models that describe the basic operational structure of an enterprise as well as models of product function, these ICEs are critical to its knowledge capturing abilities.

Past research in the engineering field has resulted in the development several ontologies describing subsets of the engineering domain. Though ontologies are theoretically interoperable, achieving this is greatly facilitated using common upper-level models and a high degree of orthogonality (limited overlap) between domains. Though, few engineering ontologies meet these criteria. For this work, interoperability was achieved by redefining existing ontologies through a labor process of realignment with a common upper level, elimination of incompatible terms, and consolidation of properties under the RO. Ontologies were selected for inclusion based on several inclusion criteria: publication in a peer reviewed source, definition using OWL or a compatible language, and availability through free online ontology repositories. Where multiple alternatives exist, decisions were made based on consistency with BFO and scope relative to that of ICAM.

An OWL implementation of the NIST core product model (CPM) [32] was included, providing an information model of basic product attributes. The CPM models a product as being composed of forms consisting of materials and geometry, features composed of forms and having designed functions, which in turn ultimately build to individual artifacts and assemblies. The manufacturing service description language (MSDL) [38] was included to provide knowledge relating to manufacturing processes and services, as well as a subset of axioms relating to material performance. The MSDL consists of a hierarchy of (mostly reductive) manufacturing processes and manufacturing services, with additional information about process parameters. The semantic additive manufacturing process ontology (SAMPro) [47], an ontology of additive manufacturing processes that nests within and extends the original MSDL, was also included to provide an explicit model of additive manufacturing processes. Finally, the functional basis ontology (FBO) [34], developed based off the functional modeling terminology of the same name [33], was included to provide a set of formally defined functions for describing various parts and features. The functional basis is used to compose functional models with a limited, defined terminology. The FBO simply implements this functional modeling terminology in OWL.

Implementation of Modular Ontologies.

Innovative capabilities of additive manufacturing was created to support innovative design in additive manufacturing by capturing information relating to various fabrication capabilities of additive manufacturing in general as well for specific machines, by representing the functional purpose of these capabilities, and by facilitating a searchable knowledge base of past innovative solutions using additive manufacturing. It comprises a suite of linked ontologies covering three domains: product realization, manufacturing, and business. These are, in turn, linked by unifying application ontology (ICAM) that connects a series of otherwise disparate knowledge bases (Fig. 2). To summarize, a product knowledge base captures past uses of AM. A machine knowledge base contains information about specific manufacturer models of AM machine that are used to create these products. A process knowledge base is used to reflect shared traits of all machines used in a specific AM process. Finally, the capabilities knowledge base contains all the specific abilities of AM processes or machines that are used to fabricate products having some set of desirable characteristics.

Since few engineering domain ontologies have been extensively vetted, it is possible that future research will yield more comprehensive or better-defined ontologies of domains included in ICAM. A modular structure, based on the spoke and wheel approach advocated in BFO style guidelines, was adopted to guard against this possibility (Fig. 3). The highest level ontology is BFO, which provides a top-level view of all entities. Just below this level of abstraction is IAO, and by extension RO, which were deemed necessary to accurately and appropriately model all subsequent domains. From this point, the ontology splits into domain specific models, with CPM providing the main model of the design domain, MSDL and SAMPro the manufacturing domain, and custom-made business model ontology capturing enterprise considerations. These are supported by a set of modular ontologies that define shared concepts, such as dimensions, various engineering material qualities, and the FBO, meaning that each domain can reference the class structure independently.

Since the engineering ontologies were not BFO conformal, they had to be significantly edited. This was done by iteratively nesting each class and property in the ontologies in BFO's class structure and RO's property structure, respectively. Automated reasoning software was used to identify cases where this created inconsistencies. The culprit axioms were then redefined to avoid these issues. Where terms were overly broad or properties used inconsistently, they were split into separate terms that could be more explicitly aligned with BFO's model. Instances where they were used were then modified to reflect the revised definitions. Overlapping terms were either assigned to a specific domain or deprecated and then replicated in separate ontologies that were then imported into the overlapping ontologies. In both cases references to overlapping terms were modified to refer to a common Internationalized Resource Identifier (IRI), which uniquely identifies a term in an ontology and its originating ontology. Doing so renders the shared terms indistinguishable from one another during reasoning and is equivalent to sharing a term from one ontology with another. For example, CPM and MSDL both contain terms for the physical features of the product. In this case, features were regarded as being more fundamentally part of the scope of CPM, and so the CPM term was used to replace the MSDL term. Other terms, such as geometric and dimensional terms, were determined to exist at a higher level of abstraction than engineering design or manufacturing. In this case, the terms were deprecated, and replicated in an aggregated treatment of geometry.

Creation of an Ontology of Business Models.

An ontological representation of business models was incorporated into ICAM to capture information relating to the market aspects of product innovation using AM. This allows ICAM to represent information on how AM facilitated the delivery of value to a customer, even in cases where value to a market that is entirely divorced from the physical embodiment of the product. Without a formal ontology to capture these details, such value propositions would be difficult to express as they rely largely on how some enterprise carries out its operations. Thus, the Business and Entrepreneurship model (BEM) ontology was developed based on existing business model development methodologies. Briefly, business models are defined as descriptive models that represent the constituent parts of a hypothetical business venture, documenting the resources, revenues streams, and value offered by a business to some customer or group of customers. Because they are information content entities, the model itself largely describes objectives, which relate various entities. Like any other model, this representation is developed with a known rationale, assumptions, and idealizations. Key to it is a set of hypotheses as to how value is delivered to some group of agents that constitute a market for the product (Fig. 4). The information model thus captures the ways in which resources are exploited, and various streams of cost and revenue are realized.

Within the broader context of ICAM, AM machines are treated as a type of resource available to a business and which in turn have a set of capabilities. Ownership or access to a machine bestows these capabilities to the business, and their realization enables the creation of products or services that have value to a target market. That value might be any number of things, and so the ontology is designed such that it can express many types of value to consumers (Fig. 5). In the case of a physical product, it might for example have dispositions or qualities that are advantageous to customer, such as a pleasing aesthetics, or some set of functionality that the consumer deems desirable. Alternatively, the product may itself enable its owner to participate in a process that has value, perhaps by altering or eliminating some task that potential customers might already undergo, such as in the case of a product that automates certain customer tasks. Third, the AM system may instead enable the business to instead sell a service in which they perform some process that a customer either cannot or is disinclined to perform on his or her own.

Identification of Additive Manufacturing Capabilities.

A necessary step in the creation of ICAM was to review the current market for additive manufacturing machines and to identify process or machine capabilities that have been previously reported. This served two purposes. First, it helped to directly identify and record instances that could serve as a basis for a knowledge base of AM products. Second, it facilitated the creation of a comprehensive list of capabilities and machines having those capabilities based on both manufacturer specification and previously reported information.

A capability in ICAM is defined as a disposition of some object to be able to participate in a process at a level of quality specified in some process plan to the benefit of it or some other agent. Put simply, a capability implies completion of a process of value to some agent at some minimal level of quality. A review of manufacturers and machines was conducted to identify both AM machines and the capabilities that they possess. It should be noted this this review was not meant to populate a comprehensive machine knowledge base but rather to identify the range of capabilities on offer from current AM machines. This review was then supplemented with a review of AM based upon published studies describing existing or proposed devices or fabricated by AM, specifically focusing on the medical device domain where use of AM is common. These capabilities were assigned specific processes or machines as appropriate.

Based on this review, a class tree (Fig. 6) was created to characterize capabilities of AM systems. To keep the model consistent in the case of expansion into other manufacturing domains, we characterize these as broad manufacturability capabilities, rather than capabilities that inhere only in AM systems. We characterize three broad types of capability found in AM: fabrication capabilities, contextual fabrication capabilities, and manufacturing process output capabilities. Fabrication capabilities include the commonly cited ability to create objects that have a high degree of shape, hierarchical and functional complexity. Contextual capabilities describe the ability to complete processes under some set of conditions, and include things like desktop printing and distributed manufacturing. Process output capabilities describe the capabilities of entities created with a manufacturing process, such as shape memory or bistability.

Enabling relations between capabilities were also incorporated to first show how various capabilities might be combined to realize new ones, and second to allow more efficient querying of machines and processes. For example, a functionally graded form fabrication capability might be enabled by an ability to fabricate multimaterial forms, a combination of shape and hierarchical complexity fabrication capabilities, or both. By tracking these enabling relations, it is possible to automatically identify machines that have ways to create various capabilities using reasoning software and semantic queries.

Innovative Capabilities of Additive Manufacturing Information Model.

Innovative capabilities of additive manufacturing implements a set of relations between the disparate domain ontologies by linking similar classes and by connecting multiple domain ontologies to knowledge bases. Details of the product are captured in a design domain, consisting of several types of information content entity that describe the basic components of a product modeled in CPM—the core product model ontology. These include specifications for various aspects of the product, such as dimensions or an intended function or behavior, as well as models, metrics, and other information that motivate various aspects of the design. This design process is ultimately driven by a set of requirements, a class shared with BEM—the Business and Entrepreneurship model ontology. These, along with problems and opportunities in some market, motivate a business model that is expressed as a set of information content entities that describe the product in terms of its proposed value, intended customers, and the basic framework for how it will be delivered to customers and used to obtain revenue.

The business model ultimately describes an agent or aggregate of agents that owns or uses some set of entities having capabilities that enable fabrication of the product or delivery of some service. In the case of AM, these would consist of machines that perform manufacturing processes from MSDL and SAMPro. The product in this case is the output of these processes. Products described in the product knowledge base are similarly modeled using terminology such as forms and features from CPM. The product, or some subpart of itself, maybe the bearer of a capability or trait that is a value to some existing individual or market. Alternatively, it might modify or enable some customer process, which in turn is of value to that customer. Certain AM capabilities enable these products to be manufactured with said traits, indicating that AM has added some value to the product.

Use Process for Innovative Capabilities of Additive Manufacturing.

We propose a straightforward method for concept ideation and feasibility assessment using both the machine and use-case knowledge bases captured by the ICAM ontology (Fig. 7). The process begins with a problem or need identification stage, wherein the designer solicits customer feedback to identify potential market opportunities for new product or redesign. The AM use-case knowledge base is queried to identify products or product features that targeted similar problems, product attributes, or market opportunities based on broad product ideas or customer needs. The previous cases found by queried are intended to help inspire a product concept, and to catalogue the exact set of capabilities, materials, and specific features required to realize that concept. An initial query might, for example, seek to identify geometric features previously fabricated by AM that solved a similar problem, modified some product trait, or introduced desirable functionality. Provided these features are useful to the new product and utilize some capability that is unique to AM processes, these features might then offer a distinct in the design of a part for AM.

A second query then searches for a suitable manufacturing option that can produce a product based on these requirements. For example, the designer might want to create a product that uses a self-supporting lattice and must also be made of some high-performance metal. A query then might return a set of machines that perform direct metal laser sintering or directed energy deposition processes, but not return printers that struggle with self-supporting lattices or perform processes that do not use that metal. In this case, the designer can move to select what they believe to be the best concept and move onto a restrictive DFAM process based on the set of machines and processes returned by the query. Alternatively, they may find that there is no system that suits their needs, and so be forced to search for more product ideas.

Instantiation of Knowledge Bases.

Information uncovered during capability identification process was used as a starting point to instantiate a knowledge base of machines, capabilities, and previous use cases of AM. For each use case, a hypothetical business model was created, though in many instances, the products discussed are academic in nature and thus may be incomplete in this regard. Care was taken to note the type of AM used in each product or process and the types of capabilities used. The value of each entry was then mapped to various customers and shared concepts, such as functions, modifications of existing processes, and adjustment of qualities relevant to some customer. These were then implemented as a set of instances in the ontology as a separate knowledge base, which can be opened or closed as necessary. To aid in knowledge capture, a small set of medical domain terms were also added as classes in the ontology.

Machine-level information was obtained by finding manufacturer's information for specific machines used in products included in the product knowledge base. This information was used to identify capabilities and materials that could be used in specific machines as well as descriptive information about costs, process speed, and the like when available. These were expanded with additional machines made by the same manufacturers and by additional searches to identify competing manufacturers. All AM machines were modeled as classes, with each instance sharing asserted subclass properties and additional specifications from instances of information content entities linked to the class of machine. The process knowledge base was formed by expanding SAMPro with additional information about capabilities and allowable materials based on both the products in the product knowledge base and the machines in the machine knowledge base. This information was collected in a spreadsheet and read into the ontology as subclass axioms defining properties of specific models of machine and classes of process.

Case Studies.

Two case studies were used to better understand the knowledge capture, design support, and innovation capabilities of ICAM. The first considers a design case previously described in the literature, with the aim of evaluating knowledge capture and the ability to query ICAM to solve a simple design problem arising from the author's solution. In the second, we consider an in-house design case for a surgical instrument.

Case Study 1.

The first case study focuses on knowledge capture. Many of the proposed functions of ICAM rely on the ability to capture information about various products enabled by or manufactured with AM. Here, we considered relatively simple surgical products described previously in the literature. In both studies considered, a set of simple surgical products are fabricated via AM with minimal redesign, with the authors of the reports suggesting hypothetical value in the form of either cost savings or the elimination of a logistical delivery problem [54]. However, both reports noted issues with the use of AM for these products, such as mechanical failure under loading that could realistically be encountered during an operation. In this case study, ICAM is used to capture information relating to the proposed AM surgical tools, including design and manufacturing information, capability usage, and the underlying business cases. Once complete, ICAM's product and machine knowledge bases are then queried to identify potential solutions to the authors' reported difficulties.

Case Study 2.

In the second case study, we considered an in-house design of a novel minimally invasive surgical (MIS) instrument. MIS operations are performed through small ports with internal diameters of that are often a centimeter or less in diameter. This necessitates the creation of “tool on a stick” style devices, consisting of a handle, an elongated shaft, and a tool head remotely operated from the handle. Operations are then performed through the port, visualized with a similarly designed camera. While offering significant benefits to patients, the operational context requires a great deal of skill, and in many instances, makes finesse driven tasks such as suturing quite difficult due to loss of dexterity, tactile feedback, and visibility. Thus, in many operations, stapling devices called endocutters are frequently used to partition and seal tissues. However, their size and shape often make them poorly suited to range of operations due to small operating fields, inconveniently shaped anatomy, and significant limitations on the actual device geometry due to the port structure.

Results

Construction and Classification of Innovative Capabilities of Additive Manufacturing.

Innovative capabilities of additive manufacturing was successfully constructed in Protégé 5.2 [55]. The ontology was subsequently instantiated with existing knowledge relating to machine specific capabilities and past products using the Cellfie plugin [56] to read in spreadsheets of related data. A fully implemented version of ICAM with instantiated knowledge bases was classified with the Pellet Reasoner [57] without any inconsistencies. Manual inspection of the inferred and asserted class hierarchy showed that they were identical.

As of the running of the two case studies, ICAM's linked knowledge bases contained 46 machines annotated with capability information, 20 individual AM products drawn from the medical realm, and 15 isolated product features. Where capability, customer value, and business model information were available, this information added additional means by which to search for information in the framework. Processes were also enriched with additional information regarding process capabilities, materials, and the like.

Case Study 1. Knowledge Capture for Additive Manufacturing of a Simple Surgical Tools.

Case Study 1 Results: ICAM was used to capture information relating to the case studies. In both cases, the devices in question provided a value to their customers by eliminating the need for a customer to participate in a series of processes. For both the retractor and surgical kit case, an agent participates in some process of inventory management, which has subprocesses consisting of maintaining a stock of relevant parts, tracking that stock, and then ordering and receiving parts. As ICAM does not have an inventory management model, these are simply represented as instances of a planned process that realize various receiving, storing, and transferring functions, with readable language comments describing details of what the instances represent.

The surgical kit for space missions is demonstrative of the knowledge capture abilities of ICAM's linked ontologies. In this instance, the problem the customer has is a process that is typical of terrestrial operations (the shipping and delivery of surgical tools) that is impossible, or at least very inconvenient in the case of a space mission. Thus, the process itself is a problem, rather than a disposition it realizes. The author's proposed solution is simple to use AM to replicate existing surgical tools on site. A simple property chain axiom infers that because the service (AM printing) eliminates a process (shipping and receiving), clearly, it is a value to the customer for whom that process is a problem. This is represented in the ontology as a design (an information entity) that bears a value in the form of a manufacturing plan (a realizable entity inhering in some person possessing the design) that eliminates a process. Linked as such with a has value property, it is thus possible in future designs to locate the design by querying for entities that deliver value by eliminating processes that are themselves problematic.

Going further, it can be seen that several capabilities of the printer used (its ability to realize certain material qualities, be run by low skill users and desktop printing ability) enable the very process that solves the problem. Thus, the use of AM in this case offers advantages in the form of the manufacturing context, rather than the specific product being created. In the study, the authors did additionally note that the AM material strength was not reliable at the given thickness, and so they increased thickness. This is reflected in the instantiated case as a problem with one output (a thin surgical tool), having to do with its ultimate tensile strength (indicated by membership of the problem instance in said class), which is solved by a manufacturing plan, which realizes an increasing function that affects thickness. The thickness of the new design is connected to the problem disposition via the “has solution” property, indicating that the thickness change fixed the strength problem.

In the retractor application, each of these subprocesses realize costs, with the receiving portion being an output of a process that includes shipping, which itself realizes yet another cost. These costs are a problem to customer. The described plan, in which the retractor is instead printed on site from a three-dimensional CAD, eliminates the shipping and receiving process, and in doing so, eliminates the cost. As in the previous case, a property chain infers then that the AM printing service is of value to the customer (in this case a hospital). The information model moreover captures that the key AM contributions were the realization of several capabilities: distributed manufacturing, low skill manufacturing, and desktop manufacturing, all of which combined enable the service in the first place.

Suppose, however, the mechanical performance of the device was insufficient, like the surgical kit. In this case, the output of the printing service (an AM retractor) would be as having a problem with a has problem relation pointing to an individual of the class ultimate tensile strength, which is realized in a failure process. This provides the basis for a search of ICAM's broader AM use case knowledge base to identify a suitable design change and or process substitute that will improve strength of the part while still permitting the delivery of the printing service. As the entire plan hinges upon being able to economically distribute manufacturing, it is likely safe to assume that any solution must be available to a user with relatively low skill. Moreover, given original product was meant to be manufactured in a sterile environment, desktop printing capabilities are also likely to be necessary. A query of the knowledge base can be used to find an appropriate solution subject to these constraints upon the manufacturing system:

(bearer of some ((is solution to or realization affects) some ultimate tensile strength) and is specified output of some (performed by some (bearer of some (desktop manufacturing capability and bearer of some low skill manufacturing capability))

The first statement looks for entities that have a disposition (presumably strength) that solves a problem stemming from an ultimate tensile strength. These would presumably be cases where a new material was used, or reinforcement was added. The second narrows the search to cases where the part is manufactured by a system with the same capabilities that enabled the retractor service in the first place. Analyzing the results, the surgical kit considered previously of course matches these constraints, and is returned when the knowledge base is queried. The knowledge base returns other options as well. Another result came directly from the printer knowledge base, as it contains a desktop printer that produces parts reinforced with fiber, resulting stronger parts. Similarly, printers that adjust part density with internal lattices were also noted, as adjustment of lattice infill can affect part strength.

Discussion Case Study 1.

The results of the first case study show two key findings about the use of ICAM to capture knowledge and then reason upon it. In both cases, the model was sufficient to capture information relating to each case relatively unambiguously. Because of the information-driven approach used to express relations in ICAM using primarily the RO, the knowledge from the first case was made reusable. It is difficult to see otherwise how an issue relating to mechanical failure because of relatively poor mechanical strength would otherwise have been identified. ICAM's proposed use case also appears to have been shown to be reasonable. In addition to the first case where simple geometric changes resolved a device failure, the second case used a query that consolidated the ideation and feasibility checking aspects of ICAM. A past solution used a printer with the necessary capabilities, and thus the query returned a result. However, this also implies that there exists a printer within the knowledge base that has the full set of required capabilities in addition to those needed to implement the reinforcement solution concept. The second major takeaway, however, is that ICAM is limited largely by the scope of its knowledge model. It is impossible to look for cases where a “shipping process” was disrupted in the current version of ICAM. This suggests that further extension with domain specific knowledge might be of value.

Case Study 2: Minimally Invasive Surgical Tool.

Case Study 2 Results: Unlike the first case study, no instantiation of ICAM was required as it is being used to investigate a device concept. In this case, the key concern is ideation to find ways to reduce the cross section of the proposed surgical tool. To do so, a query must be formulated to identify relevant instances from the AM application knowledge base. Several approaches might be feasible and so can be combined into a single query using multiple “And” statements. First, it should be noted that the endocutter tool is modeled in ICAM as a design specification, having some dimension specification that exceeds the corresponding specification in its requirements. From the problem, the designer might infer that he or she is looking for solutions that either solve problems having to do with other members of the class area, or alternatively realize a reducing function that affects area, expressed using the function hierarchy from the FBO. To identify potential design directions from the AM application knowledge base, a query such as the following is used:

is solution to some area or realizes some (reducing function and affects some area) or has function some (reducing function and realized in some (affects some area))

As with the first case study, the first statement in the query identifies cases where some aspect of the application instance was a solution to a problem with an area. The second part looks for AM use instances where a product or service realizes a function that has the effect of reducing area. The third is similar, looking for instances where the product itself is specifically designed to have the function of reducing some area. In each case, the queries refer to classes rather than individuals within the ontology.

A query to search the AM application knowledge base instantiated in ICAM yields two results representing ways of reducing area. The first result returned (Fig. 8) represents a device manufactured by a fused deposition modeling process that can be folded into two stable configurations: one with a small cross section and one that can be used for grasping. The fold pattern then has some function, which may be realized to reduce area. In both instances, however, some information is lost as ICAM does not have a medicine specific information model to draw upon, so the results are not searchable by knowledge specific to the medical field.

The second instance identified (Fig. 9) is quite similar. Rather than using a designed folding structure, it instead relies upon a shape memory material (polylactic acid (PLA)) and a geometry that can be stretched into a cable. In this instance then, the shape memory allows a deformation process that substantially changes the overall shape of the product, reducing the area such that it can be introduced via an endoscopic port (Fig. 9). Once introduced, the shape memory of the material causes it to unwind, blocking off a vessel.

A third product instance uses a combination of these approaches by combining shape memory hinges with an implant so that it can be unfolded during delivery and self-folds in vivo. Captured in ICAM, this case simply combines aspects of the two prior ones. The hinge structure allows the delivery of a nitinol spinal implant through a cannula as a set of straightened links. The design thus utilizes one of the AM features in the knowledge base, namely a living hinge feature. This has a substantially smaller cross-sectional area compared to its folded configuration, which matches the shape a traditional implant. The business terminology correspondingly captures information about a value to patients achieved by enabling an alternative procedure to more invasive alternatives that are problematic, or at least less preferable.

These options can thus inform ideation for a solution to the area problem. The rigid structures of endocutter prohibit large scale deformation as in the case of the shape memory approach. However, foldable structure was deemed to be a viable approach, especially if hinged structures are used to allow an otherwise box-shaped endocutter to be delivered. From this, a concept for an endocutter, which has a hinge structure in its distal and proximal sides, was developed. The hinge is bent during introduction, leading to an elongated longitudinal dimension but a reduced transverse width, reducing the overall size of the stapler (Fig. 10).

This slight fold, coupled with a modified internal structure to allow transverse movement, which would resulting a sufficient reduction in area to introduce the tool through the standard ports used for the target surgery. Coupled with further design cases, ICAM and its linked knowledge base might then be queried to introduce novel functionality into the device.

The question remains how the folding device might be manufactured in practice. The shape in Fig. 10 requires a hinge structure, and a means to actuate it. From the knowledge already queried, a nitinol or PLA structure might be suitable, so any manufacturing option would need to be able to use one of these materials. Second, an ability to construct a living hinge is desirable. The AM machine and process knowledge base can thus be queried. Doing so reveals that fused deposition modeling very commonly supports PLA printing, and can manufacture living hinges (as inferred by use in a case. Similarly, SLM has been used to print nitinol (though is not used as a standard material for any machine in the knowledge base), and indeed was used to manufacture the hinged device returned by querying the product knowledge base. Thus, two manufacturing options were determined for variants on the concept.

Case Study 2 Discussion.

Case study 2: Demonstrates the use of ICAM for a less simplistic device. Even for this more complex device, the querying needed to identify the device was nonetheless relatively straightforward. Because ICAM contains a subontology dealing with various design dimensions and material properties, it can be queried using these in combination with functional information. In doing so, ICAM enabled queries, which found two directly related devices that employed potential solutions from a related field. While ICAM has a limited knowledge base of AM use cases at present, this could be expanded further, potentially opening the door to solutions from outside contexts. This points to both a potential advantage and a potential problem. In the latter case, ICAM lacks domain-specific models to capture full context. This means that as the number of cases increases, the percent that are ill suited might increase in turn. However, this concern might be mitigated by simply importing domain models into ICAM to support various domain-specific case sets. The advantage, however, is that this same property might allow a great degree of cross domain reasoning, whether domain knowledge is included or not.

Discussion

Several approaches have been proposed for the use of creativity in DFAM, many of which utilize past AM successes to inspire new design directions. On its own, this approach is potent, but somewhat limited. Without a robust way selecting past successes that are directly relevant to specific design, a designer is left to sift through a large mass of disparate data that may or may not be relevant. Moreover, this relevance may not be immediately obvious from pictures and may be labor intensive to associate with a design based on plain text descriptions. With the introduction of ICAM, we propose an extension of past methods. Rather than using information about past successful deployment of AM, we instead propose to capture the knowledge from those past experiences and model it in such a way that it can be easily, very specifically retrieved, and analyzed by a designer to aid in ideation. Moreover, this approach does so from multiple perspectives such as engineering design and enterprise. Though ICAM has only been implemented as a less easily usable information model, we assert that this model could serve as a basis for a highly effective design ideation tool.

As seen in both case studies, ICAM has the potential for innovative design ideation. In the first, a simple problem is presented in the case of a retractor that is not necessarily strong enough. The knowledge base is queried to find a correspondingly simple set of solutions. Realistically, an intelligent designer might have come up with these by himself or herself based on first principles. Less likely would be that the designer would know that a reinforcement option might be swapped directly into the original model of distributed manufacturing envisioned in both cases. While a person highly familiar with the AM domain might know of printers that had the necessary combination of capabilities, he or she would be forced to rely on recall. A person without this expertise might be utterly unable to make such an association. More importantly, a broader knowledge of operations and applications of additive manufacturing might be needed to subsequently identify other cases where tools might be manufactured on site, rather than centrally manufactured. This combination of technically useful knowledge and representation of cases that might have been creative for other reasons such as distribution can be of significant aid to the designer above and beyond what might be done with a less expansive approach. Thus, the capture of knowledge in this case has potentially significant value in even a straightforward design case.

By comparison, the instances returned in the second case study require a broader and more sophisticated knowledge of AM fabrication capabilities. That understanding might not be particularly easy to reach or bring to bear in a specific design context absent a knowledge rich design aid. Text-based descriptions could capture similar information, but identifying relevant text would nonetheless be challenging. A database lacking a knowledge model might capture the use of foldable or shape memory structure in AM. However, its application to the specific problem would again be difficult to surmise without significant mental effort. Picture-based systems might be sufficient to indicate shape memory to reduce area, but the dynamics of folding would be difficult to represent to say the least and yet again difficult to associate with a specific design challenge. In this case then, ICAM appears to offer a significant benefit to the designer. Rather than browse through random and or semicomplete use cases, they can instead limit their reuse of past knowledge to cases that have some desired similarity to the problem at hand. If expanded to include many cases from many domains, this might make ICAM a very powerful tool for design ideation.

Though we focus primarily on the usefulness of ICAM for ideation, it should be noted that ICAM has its roots in a highly formal knowledge model, designed to support interoperability and reusability of information. The usefulness of this knowledge intensive approach is seen in both case studies. Because the ICAM supports a knowledge driven approach, ICAM can be used to identify highly specific information, such as specific functionality or alteration of specific attributes of some entity. This means that every instance added to its knowledge base makes ICAM more powerful overall, able to search a wider array of products and along more and more value generation pathways. Case study one shows how two related instances might be used to gain knowledge about each other, and how an only tangentially related case might be used to solve a problem common to both AM applications. As ICAM is expanded, these tangential solutions might become more numerous, and ICAM more utile. This is not necessarily the case in solutions that lack a formal knowledge layer. More design instances might mean there are more cumulative design directions available to the designer in the entire framework, but their chances of finding them might become increasingly slim as a database expands.

The demonstrative case studies, however, did unveil some limitations in this approach. First, the quality and number of relevant products returned from a given query are very much dependent on the contents of ICAM's knowledge bases. While the first case study identified a design having potentially useful features based on a very specific query, a less well instantiated version of ICAM's knowledge base might have failed to do so. In a simpler case, an incomplete machine database might erroneously lead a designer to conclude that no available system has the set of capabilities they require to realize some product concept. On top of this, instantiation of knowledge is fairly involved. One must map out several aspects of a product. In this study, we accomplished this via an ontology software plugin, but this approach requires extensive knowledge of both the plugin and the underlying ontology, something that makes mass use deployment difficult. More work may thus need to focus on tools built upon the ontology to render it more usable and less labor intensive to operate.

A second, minor limitation points to a potential direction for future research with ICAM. In both cases, lack of a specific domain model led to a loss of information that might have been useful in future querying. In projects where the products discussed in case study one are relevant, it may be useful to search for solutions that simply eliminate certain types of process. Similarly, in case study two, both solutions were from the same domain (minimally invasive surgery) as the proposed product under investigation. Being able to look at devices in the knowledge base used in minimally invasive surgery could be useful for both design ideation (as it would quickly yield a laundry list of small mechanisms used in these surgeries) and for the analyzing the current market to identify innovative new business models. This suggests that domain-specific expansions of ICAM might support highly detailed reasoning based on nonengineering domain knowledge.

The potential for expansion points to one of several strengths in this approach. The multidomain model in ICAM is clearly engineering focused, but supports expansion into other domains using a similar process to the one used to re-align engineering ontologies. This might be used to further define existing knowledge within the knowledge bases, and expand upon for use elsewhere. Absent a formal information model, this would be difficult to do in a replicable way. Moreover, the approach taken in this works means that the resulting ontology should interoperate easily with existing BFO conformal ontologies, as well as new ones developed in the future. The use of enterprise information is also a major strength. Though case study one focused largely on knowledge capture and reuse exercise, it is notable that the retractor design points to a highly unorthodox business model for the medical space. Distributed production of various medical products is uncommon, but application of this principle elsewhere could yield interesting new innovations. Without such an approach to analyze both manufacturing and economic factors, these data would likely be lost.

Funding Data

• National Science Foundation (NSF) (Grant No. 1439683).

References

References
1.
Gibson
,
I.
,
Goenka
,
G.
, and
Narasimhan
,
R.
,
2010
, “
Design Rules for Additive Manufacture
,”
International Solid Free Form Fabrication Symposium
, Austin, TX, Aug. 9–11, pp. 705–716.
2.
Rosen
,
D.
,
2014
, “
Design for Additive Manufacturing: Past, Present, and Future Directions
,”
ASME J. Mech. Des.
,
136
(
9
), p.
090301
.
3.
Kietzmann
,
J.
,
Pitt
,
L.
, and
Berthon
,
P.
,
2015
, “
Disruptions, Decisions, and Destinations: Enter the Age of 3-D Printing and Additive Manufacturing
,”
Bus. Horiz.
,
58
(
2
), pp.
209
215
.
4.
Weller
,
C.
,
Kleer
,
R.
, and
Piller
,
F. T.
,
2015
, “
Economic Implications of 3D Printing: Market Structure Models in Light of Additive Manufacturing Revisited
,”
Int. J. Prod. Econ.
,
164
, pp.
43
56
.
5.
Rodrigue
,
H.
, and
Rivette
,
M.
,
2010
, “
An Assembly-Level Design for Additive Manufacturing Methodology
,”
IDMME-Virtual Concept
, Bordeaux, France, Oct.
6.
Ruffo
,
M.
,
Tuck
,
C.
, and
Hague
,
R.
,
2006
, “
Cost Estimation for Rapid Manufacturing-Laser Sintering Production for Low to Medium Volumes
,”
Proc. Inst. Mech. Eng., Part B: J. Eng. Manuf.
,
220
(
9
), pp.
1417
1427
.
7.
Martelli
,
N.
,
Serrano
,
C.
, and
van den Brink
,
H.
,
2016
, “
Advantages and Disadvantages of 3-Dimensional Printing in Surgery: A Systematic Review
,”
Surgery
,
159
(
6
), pp.
1485
1500
.
8.
Eddy
,
D.
,
Calderara
,
J.
, and
Price
,
M.
,
2016
, “
Approach Towards a Decision Support System for Additive Manufacturing
,”
ASME
Paper No. DETC2016-60507.
9.
Atzeni
,
E.
, and
Salmi
,
A.
,
2012
, “
Economics of Additive Manufacturing for End-Usable Metal Parts
,”
Int. J. Adv. Manuf. Technol.
,
62
(
9–12
), pp.
1147
1155
.
10.
Yim
,
S.
, and
Rosen
,
D.
,
2012
, “
Build Time and Cost Models for Additive Manufacturing Process Selection
,”
ASME
Paper No. DETC2012-70940.
11.
Liu
,
X.
, and
Rosen
,
D. W.
,
2010
, “
Ontology Based Knowledge Modeling and Reuse Approach of Supporting Process Planning in Layer-Based Additive Manufacturing
,”
International Conference on Manufacturing Automation
(
ICMA
), Hong Kong, China, Dec. 13–15, pp.
261
266
.
12.
Rosen
,
D. W.
,
2007
, “
Design for Additive Manufacturing: A Method to Explore Unexplored Regions of the Design Space
,”
Eighteenth Annual Solid Freeform Fabrication Symposium
(
SFF
), Austin, TX, Aug. 1–3, pp.
402
415
.https://sffsymposium.engr.utexas.edu/Manuscripts/2007/2007-34-Rosen.pdf
13.
Lynn-Charney
,
C.
, and
Rosen
,
D. W.
,
2000
, “
Usage of Accuracy Models in Stereolithography Process Planning
,”
Rapid Prototyping J.
,
6
(
2
), pp.
77
87
.
14.
Dinar
,
M.
, and
Rosen
,
D. W.
,
2017
, “
A Design for Additive Manufacturing Ontology
,”
ASME J. Comput. Inf. Sci. Eng.
,
17
(
2
), p.
021013
.
15.
Ranjan
,
R.
,
Samant
,
R.
, and
Anand
,
S.
,
2015
, “
Design for Manufacturability in Additive Manufacturing Using a Graph Based Approach
,”
ASME
Paper No. MSEC2015-9448.
16.
Nelaturi
,
S.
,
Kim
,
W.
, and
Kurtoglu
,
T.
,
2015
, “
Manufacturability Feedback and Model Correction for Additive Manufacturing
,”
ASME J. Manuf. Sci. Eng.
,
137
(
2
), p.
021015
.
17.
Chu
,
C.
,
Graf
,
G.
, and
Rosen
,
D. W.
,
2008
, “
Design for Additive Manufacturing of Cellular Structures
,”
Comput.-Aided Des. Appl.
,
5
(
5
), pp.
686
696
.
18.
Bendsøe
,
M. P.
, and
Kikuchi
,
N.
,
1988
, “
Generating Optimal Topologies in Structural Design Using a Homogenization Method
,”
Comput. Methods Appl. Mech. Eng.
,
71
(
2
), pp.
197
224
.
19.
Langelaar
,
M.
,
2016
, “
Topology Optimization of 3D Self-Supporting Structures for Additive Manufacturing
,”
Addit. Manuf.
,
12
(Pt. A), pp.
60
70
.
20.
Gaynor
,
A. T.
, and
Guest
,
J. K.
,
2014
, “
Topology Optimization for Additive Manufacturing: Considering Maximum Overhang Constraint
,”
AIAA
Paper No. 2014-2036.
21.
Eschenauer
,
H. A.
, and
Olhoff
,
N.
,
2001
, “
Topology Optimization of Continuum Structures: A Review
,”
ASME Appl. Mech. Rev.
,
54
(
4
), pp.
331
390
.
22.
Laverne
,
F.
,
Segonds
,
F.
, and
Anwer
,
N.
,
2015
, “
Assembly Based Methods to Support Product Innovation in Design for Additive Manufacturing: An Exploratory Case Study
,”
ASME J. Mech. Des.
,
137
(
12
), p.
121701
.
23.
Yang
,
S.
, and
Zhao
,
Y.
,
2015
, “
Additive Manufacturing-Enabled Design Theory and Methodology: A Critical Review
,”
Int. J. Adv. Manuf. Technol.
,
80
(
1–4
), pp. 327–342.
24.
Boyard
,
N.
,
Rivette
,
M.
, and
Christmann
,
O.
,
2013
, “
A Design Methodology for Parts Using Additive Manufacturing
,”
Sixth International Conference on Advanced Research in Virtual and Rapid Prototyping
, Leiria, Portugal, Oct. 1–5, pp.
399
404
.https://www.researchgate.net/publication/282863835_A_design_methodology_for_parts_using_Additive_Manufacturing
25.
Bin Maidin
,
S.
,
Campbell
,
I.
, and
Pei
,
E.
,
2012
, “
Development of a Design Feature Database to Support Design for Additive Manufacturing
,”
Assem. Autom.
,
32
(
3
), pp.
235
244
.
26.
Rias
,
A.
,
Segonds
,
F.
, and
Bouchard
,
C.
,
2017
, “
Towards Additive Manufacturing of Intermediate Objects (AMIO) for Concepts Generation
,”
Int. J. Interact. Des. Manuf. (IJIDeM)
,
11
(
2
), pp.
301
315
.
27.
D'Antonio
,
G.
,
Segonds
,
F.
, and
Laverne
,
F.
,
2017
, “
A Framework for Manufacturing Execution System Deployment in an Advanced Additive Manufacturing Process
,”
Int. J. Prod. Lifecycle Manage.
,
10
(
1
), pp.
1
19
.
28.
Kumke
,
M.
,
Watschke
,
H.
, and
Vietor
,
T.
,
2016
, “
A New Methodological Framework for Design for Additive Manufacturing
,”
Virtual Phys. Prototyping
,
11
(
1
), pp.
3
19
.
29.
Kumke
,
M.
,
Watschke
,
H.
, and
Hartogh
,
P.
,
2017
, “
Methods and Tools for Identifying and Leveraging Additive Manufacturing Design Potentials
,”
Int. J. Interact. Des. Manuf. (IJIDeM)
,
42
(22), pp.
1
13
.
30.
Chandrasegaran
,
S. K.
,
Ramani
,
K.
, and
Sriram
,
R. D.
,
2013
, “
The Evolution, Challenges, and Future of Knowledge Representation in Product Design Systems
,”
Comput.-Aided Des.
,
45
(
2
), pp.
204
228
.
31.
Gruber
,
T. R.
,
1993
, “
A Translation Approach to Portable Ontology Specifications
,”
Knowl. Acquis.
,
5
(
2
), pp.
199
220
.
32.
Fenves, S. J.
,
Foufou, S.
,
Bock, C.
, and
Sriram, R. D.
, 2008, “
CPM2: A Core Model for Product Data
,”
ASME J. Comput. Inf. Sci. Eng.
,
8
(1), p. 014501.
33.
Hirtz
,
J. M.
,
Stone
,
R. B.
, and
McAdams
,
D.
,
2001
, “
Evolving a Functional Basis for Engineering Design
,”
ASME
Paper No. DETC01/DTM-21688.https://www.researchgate.net/profile/Kristin_Wood/publication/268323233_EVOLVING_A_FUNCTIONAL_BASIS_FOR_ENGINEERING_DESIGN/links/5479d6040cf205d1687fa99e/EVOLVING-A-FUNCTIONAL-BASIS-FOR-ENGINEERING-DESIGN.pdf
34.
Fernandes
,
R.
,
Grosse
,
I.
, and
Krishnamurty
,
S.
,
2007
, “
Design and Innovative Methodologies in a Semantic Framework
,”
ASME
Paper No. DETC2007-35446.
35.
Grosse
,
I. R.
,
Milton–Benoit
,
J. M.
, and
Wileden
,
J. C.
,
2005
, “
Ontologies for Supporting Engineering Analysis Models
,”
AIE Edam
,
19
(
1
), pp.
1
18
.
36.
Rockwell
,
J.
,
Grosse
,
I. R.
, and
Krishnamurty
,
S.
,
2009
, “
A Decision Support Ontology for Collaborative Decision Making in Engineering Design
,”
International Symposium on Collaborative Technologies and Systems
(
CTS'09
), Baltimore, MD, May 18–22, pp.
1
9
.
37.
Eddy
,
D.
,
Krishnamurty
,
S.
, and
Grosse
,
I.
,
2014
, “
An Integrated Approach to Information Modeling for the Sustainable Design of Products
,”
ASME J. Comput. Inf. Sci. Eng.
,
14
(
2
), p.
021011
.
38.
Ameri
,
F.
, and
Summers
,
J. D.
,
2008
, “
An Ontology for Representation of Fixture Design Knowledge
,”
Comput.-Aided Des. Appl.
,
5
(
5
), pp.
601
611
.
39.
Witherell
,
P.
,
Krishnamurty
,
S.
, and
Grosse
,
I. R.
,
2007
, “
Ontologies for Supporting Engineering Design Optimization
,”
ASME J. Comput. Inf. Sci. Eng.
,
7
(
2
), pp.
141
150
.
40.
Kornyshova, E.
, and
Deneckère, R.
, 2010, “
Decision-Making Ontology for Information System Engineering
,”
International Conference on Conceptual Modeling
, Vancouver, BC, Canada, Nov. 1–4.https://hal-paris1.archives-ouvertes.fr/hal-00662930/document
41.
Witherell
,
P.
,
Grosse
,
I. R.
, and
Krishnamurty
,
S.
,
2013
, “
AIERO: An Algorithm for Identifying Engineering Relationships in Ontologies
,”
Adv. Eng. Inf.
,
27
(
4
), pp.
555
565
.
42.
Premkumar
,
V.
,
Krishnamurty
,
S.
, and
Wileden
,
J. C.
,
2014
, “
A Semantic Knowledge Management System for Laminated Composites
,”
Adv. Eng. Inf.
,
28
(
1
), pp.
91
101
.
43.
Eddy
,
D.
,
Krishnamurty
,
S.
, and
Grosse
,
I.
,
2011
, “
Support of Product Innovation With a Modular Framework for Knowledge Management: A Case Study
,”
ASME
Paper No. DETC2011-48346.
44.
McPherson
,
J. D.
,
Grosse
,
I. R.
, and
Krishnamurty
,
S.
,
2013
, “
Integrating Biological and Engineering Ontologies
,”
ASME
Paper No. DETC2013-13527.
45.
Hagedorn
,
T. J.
,
Grosse
,
I. R.
, and
Krishnamurty
,
S.
,
2015
, “
A Concept Ideation Framework for Medical Device Design
,”
J. Biomed. Inf.
,
55
, pp.
218
230
.
46.
Hagedorn
,
T. J.
,
Krishnamurty
,
S.
, and
Grosse
,
I. R.
,
2016
, “
An Information Model to Support User-Centered Design of Medical Devices
,”
J. Biomed. Inf.
,
62
, pp.
181
194
.
47.
Eddy
,
D.
,
Krishnamurty
,
S.
, and
Grosse
,
I.
,
2015
, “
Knowledge Management With an Intelligent Tool for Additive Manufacturing
,”
ASME
Paper No. DETC2015-46615.
48.
Lu
,
Y.
,
Choi
,
S.
, and
Witherell
,
P.
,
2015
, “
Towards an Integrated Data Schema Design for Additive Manufacturing: Conceptual Modeling
,”
ASME
Paper No. DETC2015-47802.
49.
W3C Owl Working Group
,
2009
, “
{OWL} 2 Web Ontology Language Document Overview
,” W3C Owl Working Group, accessed Mar. 9, 2018, https://www.w3.org/TR/owl2-overview/
50.
Arp
,
R.
,
Smith
,
B.
, and
Spear
,
A. D.
,
2015
,
Building Ontologies with Basic Formal Ontology
, MIT Press, Cambridge, MA.
51.
Smith, B.
, 1995, “
Formal Ontology, Common Sense and Cognitive Science
,”
Int. J. Man-Mach. Stud.
,
43
(5–6), pp. 641–667.
52.
Smith
,
B.
,
Ceusters
,
W.
, and
Klagges
,
B.
,
2005
, “
Relations in Biomedical Ontologies
,”
Genome Biol.
,
6
(
5
), p.
R46
.
53.
Smith
,
B.
, and
Ceusters
,
W.
,
2015
, “
Aboutness: Towards Foundations for the Information Artifact Ontology
,”
Sixth International Conference on Biomedical Ontology
(
ICBO
), Lisbon, Portugal, July 27–30, pp.
1
5
.http://ceur-ws.org/Vol-1515/regular10.pdf
54.
Wong
,
J. Y.
, and
Pfahnl
,
A. C.
,
2014
, “
3D Printing of Surgical Instruments for Long-Duration Space Missions
,”
Aviat., Space, Environ. Med.
,
85
(
7
), pp.
758
763
.
55.
Musen
,
M. A.
,
Gennari
,
J. H.
, and
Eriksson
,
H.
,
1995
, “
PROTEGE-II: Computer Support for Development of Intelligent Systems From Libraries of Components
,”
Medinfo
,
8
(
Pt. 1
), pp.
766
770
.https://www.researchgate.net/publication/14609873_PROTEGE-II_Computer_support_for_development_of_intelligent_systems_from_libraries_of_components
56.
O'connor
,
M. J.
,
Halaschek-Wiener
,
C.
, and
Musen
,
M. A.
,
2010
, “
Mapping Master: A Flexible Approach for Mapping Spreadsheets to OWL
,”
International Semantic Web Conference
, Shanghai, China, Nov. 7–11, pp.
194
208
.http://iswc2010.semanticweb.org/pdf/414.pdf
57.
Parsia
,
B.
, and
Sirin
,
E.
,
2004
, “
Pellet: An Owl Dl Reasoner
,”
Description Logic Workshop
(
DL
2004), Whistler, BC, Canada, June 6–8, pp.
212
213
.http://iswc2004.semanticweb.org/posters/PID-ZWSCSLQK-1090286232.pdf
58.
Edmondson
,
B. J.
,
Bowen
,
L. A.
, and
Grames
,
C. L.
,
2013
, “
Oriceps: Origami-Inspired Forceps
,”
ASME
Paper No. SMASIS2013-3299.
59.
Zhou
,
Y.
,
Huang
,
W. M.
, and
Kang
,
S. F.
,
2015
, “
From 3D to 4D Printing: Approaches and Typical Applications
,”
J. Mech. Sci. Technol.
,
29
(
10
), pp.
4281
4288
.