Promoting Model-Based Definition to Establish a Complete Product Definition

Model-based definition (MBD) is a strategy for moving from two-dimensional (2D) paper-based drawings to three-dimensional (3D) computer-aided design (CAD) models where the model will contain all the information so that one day drawings may no longer be needed. However, in today's modeling environment, drawings are still used [1]. With advances such as better time-to-market, efficiency, and improved product quality, MBD has gained substantial popularity within the aerospace and defense industry [2]. However, a good majority of companies are not yet convinced on the idea of moving to an environment with no drawings [1].

While MBD has been gaining popularity, several questions remain regarding the full definition of MBD. Standards such as ASME Y14.41 [3] and ISO 16792 [4] exist to document how a model should be defined with annotations. These standards also help in understanding how to interpret the data within the model. However, the standards do not document the required amount of information that the model must contain [5]. It is important to understand what information needs to be communicated when considering moving from drawings to 3D-CAD models so the engineers can continue to do their jobs efficiently.

In today's industry, it is common that several disciplines and enterprises collaborate and share resources to complete various tasks. Elements that describe this type of scenario include entities and connections between the entities. The entities include applications, persons, and enterprises, whereas the connections between these entities include data exchange and collaborations. Product models are crucial in achieving this interoperability within the network of entities [6]. It is important to organize the information that is relevant to the user inspecting or working with the model so that they do not have to sift through layers of unnecessary data [7,8]. Designers from different disciplines usually work on the same models, which can distract them when they interact with design details that are unnecessary to them. Finding a common ground between different design disciplines can provide several benefits including protecting sensitive information, enabling collaborative supply chains, and facilitating multidisciplinary design [9].

This paper is focused on finding the information that is common among different aspects of the product's lifecycle. Design, manufacturing, and quality is the main focus of this paper. Maintenance, sustainment, and decommission will be addressed in future work. Ultimately, all phases of the product's lifecycle will be reviewed—leading to a common information model. Establishing an understanding for what all information needs to be in a 3D-CAD model so it represents and communicates the same level of information as a 2D drawing is key in formalizing the common information model, and the main reason why this paper focuses on the early phases of the product's lifecycle.

A review of the relevant academic literature has been composed to further investigate MBD and the information that needs to go into a 3D model to relay all the necessary information a drawing traditionally carries as well as how ontologies can be integrated to help product data. A review of frameworks and workflows has also been conducted.

MBD is the strategy of moving away from drawings and other means of product definition and moving to 3D-CAD models. This would establish the 3D-CAD model the only source for defining the product and its geometry. Adamski [10, p. 40] talks about the evolution of how MBD came to be

Traditional drawings have been used in industry to communicate design because they are easy to understand. The engineering drawing's main purpose is to carry and maintain product definition in a way that no assumptions or misinterpretations can be made. However, CAD software's development over the past decades has helped with the production of engineering drawings. Product development within CAD systems has become the standard and engineering drawings are no longer used as the primary product-definition source [1].

MBD is not widely utilized yet within the industry [10,11]; however, it is gaining popularity in engineering and manufacturing environments due to a wealth of benefits [2]. The benefits of MBD include reduction in manually reproduced data, reduced errors in design, better communication, quicker response times, fewer files to maintain, and reductions in cost [10,11].

Anderson and Vasilakis [12, p. 11] define an ontology as “a rigorous conceptual model of a specific domain.” These conceptual models have several contexts including “advanced information retrieval, knowledge sharing, web agents, natural language processing, and simulation and modeling.” Ontologies can either be domain specific or general. Domain specific ontologies model information used in a specific setting, while a general ontology serves several domain-specific ontologies [12].

Anderson and Vasilakis [12, p. 14] take their definition of an ontology further by stating:

The benefits to ontologies are they share a common understanding of information in knowledge domains, and they can improve interoperability within applications that use domain knowledge. Ontologies make assumptions explicit so applying changes is easier as assumptions evolve, and they enable reuse of domain knowledge, which means that the ontology can be used by multiple applications [12]. Ontologies help to bridge the gap of data interoperability between different software systems and assist the communication between software systems during a product's lifecycle. Ontologies can be used with standard file formats to allow various data types to be contained with a product, which can help to convey design intent. Using ontologies with standard file formats is also good for long-term archival [13].

A framework is created to help support a product throughout all the phases of the product's lifecycle. The framework is to help information flow and be obtained through the different phases of the lifecycle. Frameworks for PLM have been deployed to help integrate business and technical information systems. They also allow partners to collaborate effectively when creating products. According to Srinivasan [14, p. 464] these frameworks:

When used in a PLM system, a framework is “intended to capture product, design rationale, assembly, and tolerance information from the earliest conceptual design stage…to the full lifecycle” [15, p. 1399]. According to Sudarsan et al. [15, p. 1402], the National Institute of Standards and Technology (NIST) information modeling framework has the following attributes:

The NIST information modeling framework's implementation will provide a repository of all the product data and information from every stage of the design process. The framework will serve all the product description data to the PLM system using a single information exchange protocol, and “support direct interoperability among CAD, computer-aided engineering, computer-aided manufacturing (CAM), and other interrelated systems where high bandwidth, seamless information interchange is needed,” [15, p. 1399].

The NIST information modeling framework contains four components. These components are the core product model (CPM), the open assembly model (OAM), the design-analysis integration model (DAIM), and the product family evolution model (PFEM). The CPM establishes a base-level, generic product model. It is capable of capturing the entire context commonly shared in development. According to Sudarsan et al. [15, p. 1404–1407], the OAM establishes “a standard representation for exchange protocol for assembly and system-level tolerance information.” The DAIM is “a conceptual data architecture that provides the technical basis for tighter design-analysis integration than is possible with today's tools and information models.” Finally, the PFEM “represents the evolution of product families and the rationale of the changes involved.”

Understanding how information flows throughout a company and through different processes is crucial knowledge. Workflows are an important technology. There are a vast amount of tools that support workflow design. Having a good workflow can help to share data efficiently. Good workflows can also help workers to find where data were created, and understand how the “original source of data was used” [16, p. 537].

A primitive science of workflow designs contains workflow orchestration, workflows, and workflow instances. According to Deelman et al. [16, p. 528], “workflow orchestration refers to the activity of defining the sequence of tasks needed to manage a business or computational science or engineering process.” A workflow is a template for the workflow orchestration and a workflow instance refers to the specific workflow of a problem, which includes the definition of input data. In a science and engineering environment, these terms have a broader meaning and can be spread out into four areas. These four broad areas are composition, mapping, execution, and provenance. Composition, representation, and data model refer to the composition of the workflow using means such as text, graphics, etc. Mapping is defined as “mapping from the workflow to underlying resources.” Execution is the “enactment of the mapped workflow on the underlying resources” [16, p. 529]. Metadata and provenance refers to “the recording of metadata and provenance information during the various stages of the workflow lifecycle” [16, p. 529].

A common information model represents details that are relevant in different versions of models including design, manufacturing, and quality models. Within these models used in different workflows are domain specific elements. The common information model will contain the information that is common amongst these different domain specific elements. To reach a common information model, several sets of information will need to be understood. In an MBD environment, the model is the main knowledge artifact for product definition—what information a MBD needs to provide must be known. Also, in certain circumstances, different disciplines in industry will use the same model, but require different perspectives or contexts of the model. Breaking the data up across different platforms can be a challenge, but beneficial to the users. Bouikni et al. [7, p. 71] state “generating an appropriate view makes it possible to provide a favorable environment to the actors, where information is targeted in quantity and in contents to be adapted to the requirements of the task.” To understand what information is common among different versions of models such as design, manufacturing, and quality, the information that goes into an MBD environment must be understood.

Before attempting to establish a common information model, it is important to understand what information needs to be in the 3D-CAD model to be able to communicate the same amount of information as a 2D drawing. Quintana et al. [1, p. 506] point out “significant time and effort is required to properly assess the drawings' replacement,” meaning it will not be easy to determine what information needs to be contained within the 3D-CAD model.

For models to convey all the information contained in drawings, they will need to contain a wide variety of data. MBD should consist of one central knowledge artifact containing 3D geometry with GD&T and functional tolerances and annotations (FT&A). GD&T and FT&A refer to the products dimensions, tolerances, and any other annotations that the model must contain to be correctly interpreted [10].

Product lifecycle management (PLM) is imperative and its core aspects should be consistent for the designer to keep that the designer focused on the information that is relevant for a particular phase. According to Bronsvoort and Noort [8, p. 929]:

Companies within industry have certain standards while working with CAD/CAM systems. These standards include layers arrangement, new projects naming and numbering rules, rules for creating drawings, rules for creating 3D-CAD models, rules of creating models of parts machined on computer numerical control (CNC) machines, notes, comments, tolerances, etc. MBD files must contain basic characteristics of the product. These characteristics that must be contained within the model are notes, base-coordinate systems, dimensions, tolerances, flag notes, technical comments regarding material, surface smoothness, weight, and other general notes [10].

Information assurance is critical within each step of a models process through PLM, and there are several information assurance issues in the context of collaborative design. Information assurance creates new problems that need to be addressed accordingly so there can be development of collaborative CAD systems. These issues include protecting sensitive information; enabling collaborative supply chains; facilitating multidisciplinary design, role-based viewing, and security framework for collaborative CAD and role-based-view generation [9].

Each process of PLM security is extremely important for any company. Certain technologies exist managing digital rights. Organizations such as NIST's Information Technology Laboratory and the World Intellectual Property Organization (WIPO) are creating standards within this area [9].

Standardizing product meta-data is crucial for company collaboration and efficiency in production. Product meta-data includes information such as part number, bill-of-material, product-assembly structure, author, approver, supplies, version, and change history. Having this information standardized throughout engineering systems reaches out to other information systems. These systems include Enterprise Resource Planning (ERP), Manufacturing Execution Systems (MES), Customer Relationship Management (CRM), and Enterprise Asset Management (EAM), which leads to an increasing demand for standardization. Srinivasan [14, p. 465] clarifies on this increase:

A critical part within each process of a common information model is keeping it a singular data file for downstream consumers, in which case can be easily distributed within other areas of other departments such as design, manufacturing, and inspection. Briggs et al. [14, p. 11] state:

Aside from what information needs to go into the common information model, another issue that must be addressed is if the model needs to be used in a different software package or if the model will ever need to be translated using a neutral file format. If this is the case, it is important to know what information needs to come out of the model after being translated as opposed to what information actually does come out in the resulting file. It is also important to know and understand what information gets lost in this translation.

To help investigate the common information model, a survey was conducted with the industry professionals. This survey was sent out to a large number of industry professionals from multiple companies and locations around the world. This diverse group of industry members helped to give a good look into how the models are used throughout different industries. The survey helped to understand how the models are used in different industries, and where industry members are when it comes to using models in the place of drawings.

The survey was comprised of a demographics section, which gave background information on from where the results are coming. Questions about how information is received, as well as in what format were asked, and also where models are used in processes. If the respondents to the survey did not use the models, the survey ended. If the respondents did use the models in their processes, more questions were asked to get a better understanding of how and where. An understanding of where the respondents' level of capability was with using 3D-CAD models in their processes is crucial information for this study. After this, we asked what types of inspections the respondents do in-house, as well as what tools they use. Along with this, respondents were asked what types of manufacturing processes they use. The respondents were asked to give impacts of different issues typically faced within a manufacturing environment. The last set of questions for respondents was on why they have not moved to an MBD environment and the risks involved.

The survey information was collected and observed using charts and graphs. The following section is a summary of the survey results. Conclusions about the survey have been made, as well as recommendations, and will be given after the survey summary.

To get an understanding of how the models are used within the companies, the promoting model-based definition survey was given to industry professionals and returned 37 responses. To give an understanding of the sample being used, some questions were asked regarding the size of their company and where they were located. The largest amount of respondents (38%) worked at a company with more than 500 employees. Most of the responders (86%) are located within the United States, with the majority (75%) being located in the Midwest. The primary role of the respondents within their companies varied greatly as seen in Fig. 1. These answers were fairly diverse and ranged from sales, engineering/design, manufacturing/production, quality/inspection, management, as well as others, with the majority coming from engineering/design and management. The respondents who answered “other” possessed roles such as CEO, system analyst, owner, training, and consulting. This range of roles can help to provide a diverse look into the questions within the survey.

The respondents were asked how they receive customer order information and were given the following options: drawings only, primarily drawings (with supplemental models), primary 3D-CAD models (with supplemental drawings), and 3D-CAD models only. There were 27 responses for this, and Fig. 2 is a breakdown of the responses. Primary drawings with supplemental models were the highest at 44%. 3D-CAD models only received just over a quarter of the responses at 26%. And drawings only and primary 3D-CAD models with supplemental drawings received 15% of the responses each. This shows that the drawings are still playing a crucial role in the transfer of data with 74% of the responses using a drawing somehow. While 85% of the responses use 3D-CAD models in some fashion for carrying data, only 41% of the responses use the 3D-CAD model as the only or primary source of information.

The next question asked to the respondents was whether or not they would be able to produce a part according to specification if given only a 3D-CAD model and no drawing, which received 25 responses. Figure 3 gives a breakdown of the responses, with only 4% of the respondents giving a definite “no.” A solid 36% responded they could produce the part with no other conditions. The other 60% responded they could produce the part to specification; however, they would need to interrogate the model manually for dimensional information, with 40% of the overall respondents needing to consult with the customer to gather manufacturing and inspection details.

The respondents had a diverse use for models in their processes. The respondents were to select all the processes for which they use the models. There were 26 responses, and most of the options presented to the respondents were selected with high quantities, almost evenly, with CMM/Inspection programs receiving the most selections. Only one respondent selected that they do not use the models in their production inspection or processes. Figure 4 gives the distribution of the answers. The two votes for “other” were finite-element analysis and design.

After seeing where the respondents were using the models, the respondents were asked in what formats they receive information for making parts and to select all the formats that apply. There were 18 responses. Figure 5 shows the responses, with native 3D-CAD model (14) and STEP (11) receiving the most responses.

Knowing how the respondents received information, they were asked what format of information to make parts best suits for their process/needs. Figure 6 gives the distribution of these answers, which came from 18 of the respondents. The options given were native 3D-CAD model, 3D PDF (portable document format), JT (Jupiter Tessellation), STEP (standard for the exchange of product model data), IGES (initial graphics exchange specification), 2D PDF, DXF (data exchange format), and other. Native 3D-CAD model received the highest selection at 56%. The next highest was STEP with 22%. 3D PDF, IGES, 2D PDF, and DXF all received 6%, and there were no selections for JT.

The next question in the survey was regarding what types of inspections were done in house. Again, there were 18 responses. The options were first article inspection (FAI), receiving, in-process, and final. All the options received several selections, with FAI, in-process, and receiving getting the most, as seen in Fig. 7.

Knowing what types of inspections the respondents do in-house, they were asked what inspection equipment they currently use. All 18 responded, however, none of the respondents selected that their inspections were outsourced. The highest selected options, in order, were visual, non-CMM gauges, and CMMs with 3D-CAD models only. The lowest two options receiving votes were CMMs with drawings only and scanning, as seen in Fig. 8.

The next question in the survey asked the respondent to rate the level of impact of issues on their business from 1 to 4, 1 being not an issue and 4 being a serious issue. In between were minor issue (2) and moderate issue (3). Figure 9 shows the mean frequency of the impact of the issues. Below are the issues given to the respondent (1–19) to rate. In Fig. 9, these issues are represented by the number associated with them. There were 18 responses to this question:

1. (1)

   Performing inspection is a bottleneck.

2. (2)

   Performing off-line programming for inspection is time consuming.

3. (3)

   Receiving multiple files and/or media formats for as single product.

4. (4)

   3D-CAD models and associated drawings do not agree.

5. (5)

   3D-CAD model derivations/translations are problematic.

6. (6)

   Verifying CMM programs is time consuming.

7. (7)

   3D-CAD model is not available from customer.

8. (8)

   Communication with the customer is difficult and/or not timely.

9. (9)

   New designs have producibility issues.

10. (10)

    Time/volume of report requirements is overwhelming.

11. (11)

    There are limited design feedback opportunities from supplier to OEM.

12. (12)

    There is too much variation in production scheduling from OEMs.

13. (13)

    Data such as 3D-CAD models, drawings, and specifications from customer are not always up to date.

14. (14)

    Unable to change manufacturing processes due to certification regulations or customer policies.

15. (15)

    Certification process is sometimes difficult.

16. (16)

    Obtaining capital is challenging.

17. (17)

    Ability to hire and retain qualified/skilled workers is problematic.

18. (18)

    It is expensive to implement model-based manufacturing.

19. (19)

    Help from the local, the state, and the federal government is either nonexistent or hard to identify.

According to the chart in Fig. 9, the issues that impacted companies the greatest (mean above 2.7) were the ability to hire and retain qualified/skilled workers (3.17), performing inspection is a bottleneck (2.89), 3D-CAD models and associated drawings do not agree (2.89), and new designs have producibility issues (2.72). Several issues still had a mean over 2.6 including obtaining capital is a challenge (2.67), it is expensive to implement model-based manufacturing (2.67), verifying CMM programs is time-consuming (2.61), data such as 3D-CAD models, drawings, and specifications from customer are not always up to date (2.61). The issues with the lowest impact based on mean were 3D-CAD model derivations/translations are problematic (2.33) and unable to change the manufacturing processes due to certification regulations or customer policies (2.33).

Respondents were then asked their current level of capability with using 3D-CAD models as input to their CAM and CMM processes and given three options. The answers they had to choose from were:

• highly proficient; only minor difficulties

• somewhat proficient; internal deficiencies still exist

• currently using drawings and manual input, but have no desire to move to model-based manufacturing

Only one of the respondents claimed they used drawings and had no desire to move to model-based manufacturing. Eleven of the 18 respondents selected somewhat proficient, and six selected highly proficient.

The survey then asked the respondents to select all their manufacturing processes, with only 17 respondents opting to answer. Figure 10 gives a distribution of the selections. Traditional material removal such as cutting, turning, milling, and drilling received every vote, with assembly being the second highest selection.

To wrap up the survey, the respondents were asked what they perceived was the biggest risk for adoption of the model-based manufacturing approach as manufacturing and inspection technologies increasingly rely on 3D-digital data. Eighteen respondents were given seven options including other, and capital investment is too large was biggest risk at 28%. Figure 11 gives a breakdown of the responses. The responses for other were interoperability.

This breakdown helps give insight into why some companies are not interested yet in moving to MBD. Legacy designs (22%) is almost always an issue because drawings have been used as the main source of information and moving all that data to models can be time consuming and costly. Of the respondents, 22% said there was a lack of business pull, which appears to be that the companies do not necessarily see the potential benefits of MBD just yet.

The survey helped give insight to current standing in industry. A fairly wide range of affiliations were represented as well as job positions. A common information model cannot yet be fully defined from these surveys, but critical information has been identified. This information will be used to develop plans for replacement of drawings with the 3D-CAD models. These surveys developed the capability of industry's readiness to use models as the master definition and the potential inhibitors of their use.

This paper has supported the need to establish a common information model. A common information model contains the information that is the same from domain specific elements among different aspects of the product's lifecycle. A review of the literature was conducted and a survey was analyzed to help give a greater understanding of what information needs to be addressed in the common information model, and where industry stands in terms of implementing MBD. The following are the key results upon which we drew our conclusions:

• a majority of the survey respondents are potentially accepting of the idea of MBD

• most of the survey respondents already use 3D-CAD models as a source of product data

• most of the survey respondents still utilize 2D drawings (along with their 3D-CAD models)

• the survey respondents have skepticism and concern about eliminating 2D drawings

• the survey respondents identified several risks when moving from drawings to 3D-CAD models

From our observations of the survey results, we conclude (1) the common information model would need to be workflow specific and (2) more information is needed to establish a common information model for the early phases of the product's lifecycle.

The conclusions from the survey seem to contradict each other; however, they are consistent with what was concluded from the literature review. Industry may be accepting of the idea of MBD, and most already utilize 3D-CAD models for product data, although most still use 2D drawings along with their 3D-CAD models. From these results, it can be concluded industry only accepts the idea of MBD as long as 2D drawings are still used because skepticism remains in completely getting rid of 2D drawings.

While research, such as Hedberg et al. [17], shows MBD can be a major benefit to companies, the survey shows that many industry members have legitimate concerns for only using 3D-CAD models. For example, there are times when using 2D drawings would be easier or make more sense to a company, such as on a shop floor where the company does not have the infrastructure to support 3D-CAD technology. Many respondents felt there was too big of a risk in moving solely to 3D-CAD models from 2D drawings.

While the survey provides evidence that the industry is potentially accepting of the idea of MBD and may support the fact that 3D-CAD models can be used as the main source of product data in a production environment, it cannot yet be concluded what information needs to go in to the common information model. The survey helped lay a foundation of knowledge, but more research needs to be done to help understand what specific information goes into the models in the different aspects of the lifecycle.

As of right now, it is difficult to conclude what information is common amongst different models. Based on the results of the surveys, a proposed common information model would need to be workflow specific because of the varying degrees of information in the different workflows. A general common information model would lack enough information to be beneficial to a company's processes.

To establish a common information model, more specific information regarding the workflows is needed. Also, a clearer definition of “common” and “domain specific” will have to be established. A proposed solution would be to have a follow-up survey that lists the different elements from this survey and has the respondents “rank” each of them from 1 to 10, 1 being common and 10 being domain specific. This could help to shed light on how the members of industry see the different elements from the lifecycle, which would help to further establish the common information model.

This research has been carried out through Purdue University's Product Lifecycle Management Center of Excellence, a part of the Purdue Polytechnic Institute, and in collaboration with the National Institute of Standards and Technology (NIST).

The Engineering Laboratory of the National Institute of Standards and Technology (NIST) supported research reported in this paper under the following grants and/or cooperative agreements:

Grant No. 70NANB15H311, Purdue University, “Extending and Evaluating the Model-Based product definition.”

The contents of this report do not necessarily reflect the views of NIST. NIST and the authors do not make any warranty, expressed or implied, nor assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, product, or process included in the reports.