Parametric CAD Modeling to aid Simulation-Driven Design : An evaluation and improvement of methods used at Scania

(1)

PARAMETRIC CAD MODELING TO

AID SIMULATION-DRIVEN DESIGN

an evaluation and improvement of

methods used at Scania

Linköping University | Department of Management and Engineering Master’s Thesis, 30 hp | Master of Science - Mechanical Engineering and Machine Design Spring 2017 | LIU-IEI-TEK-A--17/02807—SE

(2)

(3)

Authors: Mattias Lokgård

M.Sc. Student in Mechanical Engineering Linköping University, Sweden

mattias.lokgard@gmail.com Andreas Grandicki

M.Sc. Student in Mechanical Engineering Linköping University, Sweden

andreas@grandicki.com Supervisors: Max Larsson

Design Engineer

Intake Components, NMGK Scania CV AB, Sweden max.larsson@scania.com Anton Wiberg

Ph.D. Student IEI Machine Design

Linköping University, Sweden anton.wiberg@liu.se

Examiner: Johan Persson Assistant Lecturer IEI Machine Design

Linköping University, Sweden johan.persson@liu.se

(4)

(5)

Abstract

This report documents a thesis conducted at Scania CV AB in Södertälje, Sweden. The main purpose of the thesis has been to examine and improve upon current prac-tices of parametric CAD-modeling at Scania, with the ultimate goal of increased design automation and simulation-driven design. The thesis was initiated with a literature study, mainly covering the fields of parametric CAD-modeling, design automation and knowledge-based engineering. Furthermore, a questionnaire and multiple interviews were conducted to assess the awareness and mind-set of the employees. Finally, a case-study was carried out to follow current methodologies, and address any deficiencies found. Some of the most important findings were that while parametric modeling has consider-able potential in enabling design automation, it is crucial, and most beneficial in terms of automation efficiency, to start with the fundamentals, namely achieving a uniform mod-eling practice. With these findings, a new proposed methodology has been introduced, as well as a recommended plan for a widespread implementation of parametric modeling at Scania. Such implementation would allow for shorter lead-times, faster adaptation to changing conditions, and reduced development expenditures.

(6)

Acknowledgements

This thesis concludes a Master’s programme in Mechanical Engineering, at the Insti-tute of Technology at Linköping University. The thesis was conducted at Scania CV AB, Södertälje in the spring of 2017 over the duration of 20 weeks, corresponding to 30 ECTS credits.

Firstly, we would like to extend our sincere gratitude and appreciation to Max Larsson, supervisor at Scania CV AB, for invaluable guidance, encouragement and support dur-ing the thesis work.

Secondly, we give our thanks to our supervisor, Anton Wiberg, and examiner, Johan Persson, at Linköping University, who have guided us through the thesis process. We would also like to thank our opponents, Munib Baber and Prajwal Shankar, for helpful discussion and feedback throughout the thesis work.

Finally, we would like to thank Kim Petersson and Hannan Razzaq, at Scania CV AB for arranging the thesis project. Furthermore, a sincere gratitude to the staff at the NMGK department of Scania CV AB for their cooperation and pleasant discussions.

Södertälje, June 2017

(7)

Acronyms & Nomenclature

CAD Computer Aided Design

CAE Computer Aided Engineering CAM Computer Aided Manufacturing CFD Computational Fluid Dynamics DA Design Automation

DOE Design of Experiments FEM Finite Element Method KBE Knowledge Based Engineering PDM Product Data Management PLM Product Life-cycle Management VBA Visual Basic for Applications

(8)

4.2.2. Knowledge-Based Engineering . . . 29 4.2.3. Information Exchange . . . 29 4.3. Methodology Development . . . 30 4.3.1. Information Collection . . . 30 4.3.2. Manufacturing Knowledge . . . 35 4.3.3. Parametric Modeling . . . 35 4.3.4. Model Quality . . . 39 4.4. Final Methodology . . . 41 4.4.1. Pre-CAD phase . . . 41

4.4.2. CAD Modeling and Robustness Evaluation . . . 42

4.4.3. Exploration and Hand-off . . . 43

4.4.4. CAD Model Structure and Best Practices . . . 44

5. Discussion 46 5.1. Discussion of Results . . . 46 5.1.1. Information Collection . . . 46 5.1.2. Manufacturing Knowledge . . . 47 5.1.3. Parametric Modeling . . . 47 5.1.4. Model Quality . . . 48

5.1.5. Environmental and Ethical Impact . . . 49

5.2. Discussion of Methodology . . . 50 5.2.1. Case Study . . . 50 5.2.2. Data Collection . . . 50 6. Conclusion 51 6.1. Research Question 1 . . . 51 6.2. Research Question 2 . . . 51 6.3. Research Question 3 . . . 52 7. Further Studies 53

Appendix A. Interview Guide (Design Engineers) 58

(10)

List of Figures

1. Interpretation of the different design spaces by Cederfeldt [9] . . . 6

2. Interpretation of Parametric Design Roadmap by Bodein et al. [5] . . . . 13

3. Proposed methodology of Luu [28] and Blomberg [3] . . . 19

4. Proposed methodology of Jansson and Wiberg [25] . . . 20

5. The methodology applied in this thesis . . . 22

6. Overview of the Excel sheets . . . 30

7. The Interface Matrix . . . 31

8. Updated Interface Matrix . . . 31

9. Revised Data Exchange Matrix . . . 32

10. Requirements and Constraints Matrix . . . 32

11. Updated Requirements and Constraints Matrix . . . 33

12. Parameter Definition Matrix . . . 34

13. Parameter Relations Matrix . . . 35

14. Base surfaces of the CAD-model . . . 36

15. Progression of the surface model . . . 37

16. Model tree structure . . . 38

17. A illustration of the iterative modeling process . . . 39

18. Solid models generated by the DOE sequence . . . 40

19. Final Methodology . . . 41

20. Updated Overview of the Excel sheets . . . 41

21. Design of Experiments with successful and failed designs . . . 43

(11)

1. Introduction

This Master’s thesis has been performed at the NMGK group of Scania CV AB (Scania), which is responsible for the development of engine intake components. Their goals are to create product documentation for manufacturing, ensure the products functionality and life span, in balance with the demands on product properties, production, service and purchasing strategies. This challenges the engineers to meet demands from multiple disciplines, which often tends to change during the course of a project.

Parametric CAD (Computer Aided Design) is a modeling method that allows the engi-neers to cope with quick and sudden changes to the design whilst gaining a more efficient design process. Parametric models can, at a later stage in the process, be used to verify and optimize the design, through e.g. CFD (Computational fluid dynamics) and FEM (Finite element method). The following thesis will strive to enable the incorporation of design requirements and knowledge from multiple disciplines early on in the design process by using parametric modeling techniques, to potentially increase productivity while ensuring manufacturability.

1.1. Background

The current design process at Scania is largely iterative, where the design engineers se-quentially gather information from the different departments such as CFD, FEM and production. Experience has shown that the feedback from these departments often ar-rive late in the process, causing unnecessary expenses of time and resources due to the increased number of design iterations. By integrating intelligence and design require-ments into the models at an early stage, it is possible to satisfy the needs of multiple departments from the start, potentially reducing errors and costly changes later on in the development process. Previous theses at Scania has produced methodologies for cre-ating parametric CAD models, however, their results has not yet been implemented to the extent presented.

(12)

1.2. Purpose & Goals

The main purpose of this thesis is to bridge the gap between theory and user practice, by improving the usability of the developed methodologies, while determining how com-petence and knowledge can be incorporated into the parametric modeling process. Fur-thermore, to determine and identify the disciplines affecting the geometrical properties of a parametric CAD model. Finally, which information from those disciplines, and how it should be included in the geometry of a model to ensure and elevate manufacturability. The goal is to establish a functional method that can be used to determine what and

how information can be included in the geometry of the model and how it affects the

modeling process, regardless of discipline.

1.3. Research Questions

RQ1: What product related information and knowledge should be connected to a para-metric CAD model to ensure and improve manufacturability?

RQ2: What is the recommended practice on implementing a multidisciplinary, paramet-ric CAD modeling approach, to give good conditions for widespread usage? RQ3: How could a product development process benefit from implementing a parametric

CAD modeling approach?

1.4. Delimitations

The developed methodology will only be applied to one specific article in the form of a case study. The parametrization of the model will be limited to the disciplines which are directly affected by the CAD model, in this case production, CFD and FEM. The actual implementation of knowledge best suited in a PDM/PLM system will not be researched upon.

(13)

2. Theoretical Framework

The theoretical framework of the thesis covers previous work performed at Scania in addition to literature in the thesis field. The previous work is the foundation and starting point for this thesis. Additionally, areas that have been deemed relevant for design automation and more specifically parametric CAD modeling for multidisciplinary design has been covered, such as parametric modeling and knowledge-based engineering.

2.1. Parametric Modeling

Parametric CAD-models can be developed in modern CAD applications and allows en-gineers to efficiently alter and modify geometry without having to recreate the model. The parametrization of CAD-models can be performed on different levels [39, 1, 10], but they can all, in some sense, elevate the quality and the re-usability of a CAD-model. Furthermore, re-using CAD-models can decrease the lead time of the product develop-ment process if defined and used properly [7].

Parametric modeling refers to the creation of geometry that is controlled and defined by non-geometric features called parameters. Parameters can be controlled through con-straints based on geometry, dimensions or logic expressions [7]. Cederfeldt and Sunnersjö [10] suggests that parametric modeling could be used to ”achieve a flexible design at an

early design stage”. With a flexible design early in the design process, it is possible

for the designer to adapt quicker to changes that might occur downstream in the prod-uct development process. The level and complexity of how the parameters are controlled and what they control, depends on the design intent and design space of the CAD model. According to Tarkian [39] and Amadori [1], geometrical transformations of CAD mod-els can be divided into two categories, morphological and topological transformations. Morphological transformations are geometrical changes made within one instance of a CAD model, i.e. the geometry does not require more information than the information stored within the given instance to re-evaluate and regenerate. Topological transfor-mations occurs on a higher level than morphological transfortransfor-mations, and is a way to effectively alter a models topology, i.e. increase, replace or reuse geometrical objects and features. Topological transformations are therefore a way of controlling morphological transformations driven by the surrounding context.

(14)

2.1.1. Modeling Methods

Multiple general methods for creating parametric and robust CAD models has been examined in previous work conducted at Scania [28, 3], as summarized;

• PARAMASS, as explained by Salehi-Douzloo [34], is a methodology divided into three main phases; specification, where the parameters and relations to be included in the model are determined, to capture the information and knowledge desired. This is arguably the most important phase. The results of this phase is docu-mented in two matrices called Parameter Structure Matrix (PSM) and Associative

Structure Matrix (ASM). The second phase, structuring and creation, is where the

actual model is created. The phase involves decomposing the model to increase the re-usability and decrease the complexity. The final phase, modification, is where the model can be modified with help of parameters and relations in a robust way. • Explicit Reference Modeling Methodology, as introduced by Bodein et al. [6], is a methodology focusing on the modeling of parts. The aim of the methodology is to achieve a more uniform and consistent modeling structure, yielding benefits such as increased model understanding among designers, increased efficiency and reusability. This is achieved by decomposing the part into simpler solids (few topo-logical modifications per operation) and the use of explicit references to geometry created by the user, reducing the amount of parent-child relationships. The solids are then assembled using boolean operations, which increases the robustness of the model. Furthermore, Bodein et al. [6] proposes that all dress-up features (such as fillets and chamfers) are to be created as close to their defining solid as possible. • Resilient Modeling Strategy, as presented by Gebhard [15], is a ”best-practices”

guide to creating robust and reusable CAD-models. As Gebhard [15] explains, parent-child relationships may be a source of instability in the model, which might not become evident until the model is modified (often a long time after the creation, and/or by someone other than the initial designer). The methodology strives to yield editable, obvious and reusable models. This is achieved by structuring the feature tree in a standardized manner, by the purpose of the features. The proposed structure is;

1. Ref: reference entities

2. Construction: entities which aid the creation of complex solids 3. Core: features defining the core shape of the model

4. Detail: features adding small features, linking to Ref and Core 5. Modify: ”features that transform faces or replicate features”

6. Quarantine: ”isolates volatile features at the end of the tree” (e.g. dress-up features)

(15)

Furthermore, Gebhard [15] describes that the model structure should aim to be

obvious rather than intuitive, as the latter is dependent on the designer and its

experience. By using obvious naming of features, the design intent can be kept. Finally, the use of parametric models is advocated to allow generation of multiple configurations, to assure reusability. As Gebhard [15] explains, conformity to these guidelines can be assured by the use of pre-release checklists.

2.1.2. Design Space and Constraints

Determining how a parametric CAD model should be developed and what information and knowledge are to be included as design parameters, is largely dependent on how knowledgeable the design engineer is of the design process [9]. In addition, the knowl-edge on how a product is to be manufactured and what the performance requirements are, is imperative to set limitations to those parameters.

Limitations to the design parameters is what defines the actual design space. The

ac-tual design space is what describes physical constraints of the product, demands from

the customer, manufacturing, and other constraint spaces that affect the design of the product. As a design engineer it is important to evaluate and prioritize these constraints in order to define the designs constraint-boundaries.

Cederfeldt [9] discusses the importance of defining the constraint-boundaries and illus-trates the actual design space with surrounding design spaces (Figure 1). Furthermore, Cederfeldt [9] suggests that the different design spaces are limiting the number of rele-vant design proposals, i.e. it is irrelerele-vant to develop a design that meets all customer requirements but can’t be manufactured.

In Figure 1, Cederfeldt [9] defines six different design spaces. The design spaces are described on a generic level, and may therefore look different depending on product and industry.

• ”Infinite” Design Space: Limited by the laws of nature.

• Physical Design Space: Limits the infinite design space even further by declaring whats possible to create and produce.

• Customer Space: Representing the demands and wishes from customers. • Product Design Space: Limits the physical design space which is dependent on

how a company have defined their product design configuration.

• Company Design Space: Limits the physical design space based on resource and manufacturing limitations.

(16)

Figure 1.: Interpretation of the different design spaces by Cederfeldt [9]

2.2. Knowledge-Based Engineering

Knowledge-Based Engineering, or KBE for short, is a broad term describing methods to streamline and automate the engineering design process. While its definition varies depending on author and perspective, many of its aspects has been captured and sum-marized by La Rocca [26];

”Knowledge based engineering (KBE) is a technology based on the use of dedicated software tools called KBE systems, which are able to capture and systematically reuse product and process engineering knowledge, with the fi-nal goal of reducing time and costs of product development by means of the following:

– Automation of repetitive and non-creative design tasks

– Support of multidisciplinary design optimization in all the phases of the design process” [26]

While KBE is often described as having great potential for improving the engineering design process, its implementation and amount of available literature has so far been lacking, partly due to its applications being limited to the most complex engineering fields [26, 41]. However, as competition increases in all fields, KBE is becoming an in-creasingly interesting topic to study [41].

(17)

Main elements of KBE includes identifying relevant knowledge, embedding it into a KBE application, and ease the reuse of such knowledge [26]. La Rocca [26] also describes how KBE is commonly implemented in modern CAD applications programmatically as rules, utilizing logic reasoning combined with provided knowledge and facts. Thus, a specific set of inputs will always yield the same output.

As Verhagen et al. [41] explains, it currently does not exist a universal method of deter-mining whether KBE is beneficial to implement for a specific article or situation. How-ever, La Rocca [26] argues that situations which are associated with multidisciplinary aspects, repetitive work through geometry modification or reconfiguration, both would favor the use of KBE. Furthermore, La Rocca [26] describes how KBE could enable the use of MDO (Multidisciplinary Design Optimization), as multiple configurations and material for CAE (Computer Aided Engineering) could be automatically generated. In the traditional engineering design process, the manual handling required for MDO would most likely be unfeasible.

As described by La Rocca [26], some of the main difficulties in implementing KBE may be;

• Licencing costs, which are often required to allow the implementation of KBE in the CAD application.

• Lack of metric, to quantify the benefit and profit of implementation.

• Lack of know-how, as the implementation is associated with a certain learning-curve.

Furthermore, the extensive use of programming, while being a powerful method of im-plementing KBE, may be intimidating to the traditional CAD user, which highlights the importance of proper learning material [26].

2.2.1. KBE implementation in CATIA V5

As in many modern CAD programs, CATIA V5 has specific workbenches, i.e. collections of tools, allowing the user to develop KBE applications directly within the modeling en-vironment. The most common and accessible tools consists of formulas, rules, reactions and checks, available in the knowledge advisor (KWA) workbench. The use of program-ming through CAA (API1 _{of CATIA [33]), allows for even more complex operations}

[4].

2.2.2. KBE methodologies

(18)

MOKA, or Methodology and tools Oriented to knowledge-based engineering Applications, can be summarized in the following steps [12, 41];

“1. Identify: Determine needs and technical feasibility 2. Justify: Validate scope and assess risks

3. Capture: Collect and structure the raw knowledge 4. Formalise: Develop Product and Process models 5. Package: Develop application

6. Activate: Introduce, use and maintain” [12, 41]

As explained by Curran et al. [12] and Verhagen et al. [41], MOKA distinguishes between

informal and formal models, where the informal models are utilized to capture and store

knowledge in a format suitable for humans, while formal models adapt the knowledge to a format suitable for computer processing [36]. The informal models are defined us-ing ICARE forms (Illustrations, Constraints, Activities, Rules and Entities), while the formal models are defined using the so called Moka Modelling Language (MML) [12]. KNOMAD, which stands for Knowledge Nurture for Optimal Multidisciplinary Anal-ysis and Design, is as the name implies a KBE methodology for implementation of

multidisciplinary knowledge [12]. The methodology is aimed to improve earlier method-ologies, such as MOKA, to better incorporate aspects related to multidisciplinary design optimization and the implementation into the design process [12]. A brief overview of the steps in KNOMAD [12];

• Knowledge capture: information and knowledge of interest is collected either from previously documented sources or existing expertise, and is documented. • Normalisation: the quality of the collected knowledge is assessed to determine

whether it should be included or not, and the knowledge is normalised.

• Organisation: the knowledge is organised to allow for easy and automatic re-trieval among the stakeholders. Here, a specific ontology can be introduced, i.e. a classification and definition of the domain and its content. The ontology aid in formalizing the knowledge.

• Modeling: the product is modelled, and by use of parameters and knowledge defined using the ontology introduced in previous steps, each discipline involved is provided with a specific model. The models are documented in a report.

• Analysis: each of the disciplines involved analyze their respective model and assess the performance, in regards to the specific discipline. This step may involve optimization.

• Delivery: the feasibility of an optimized solution is ensured, after which it is delivered to the stakeholders.

(19)

2.2.3. Manufacturing Knowledge

As explained by Gembarski et al. [16], guidelines for manufacturing methods in the prod-uct development process has had widespread use for a long time, however, they often consist of implicit and heuristic knowledge, thus relying on the skill and experience of the engineer to translate these into explicit knowledge. Gembarski et al. [16] introduces a framework for the integration of manufacturing knowledge into CAD-models, by the

operationalization of manufacturing restrictions, i.e. translating ratings such as ”good”

or ”bad” when referring to the manufacturability of products into parameters and rules which can be implemented into a CAD model using KBE.

Gembarski et al. [16] utilizes the theory of Characteristics-Properties Modelling/Property

Driven Development (CPM/PDD), which features a differentiation between the following

three elements influencing or influenced by the product;

• Characteristics: parameters which can be directly altered by the engineer, such as geometric dimensions and choice of material [16]. Characteristics are often added and altered inside a CAD-environment.

• Properties: parameters which can not be directly altered by the engineer, but are rather resulting from the characteristics, such as component weight or stress distribution [16]. Properties are often examined using CAE-tools.

• External Conditions: conditions which restrict the design space, e.g. available manufacturing methods and connecting interfaces [16].

Furthermore, synthesis is described as the process of altering a products characteristics to yield a desired property, and analysis is described as the process of verifying that the product reaches the requirements imposed upon it [16]. Thus, several iterations of synthesis and analysis may be required until all requirements are met.

2.2.4. CAD-CAE handoff

As discussed by Corallo et al. [11], the hand-off between CAD and CAE departments often require the creation of so called context models, i.e. simplified models intended for various CAE analyses, as it might be disadvantageous to send the complete CAD model. Common tasks in creating a context model is simplification of geometry and conversion to a neutral file format. The preparation of these models is often a tedious and time-consuming process, requiring the design engineer to possess specific knowledge in CAE. Corallo et al. [11] investigated the benefits of developing KBE applications to facilitate and automate this model preparation and hand-off (in the aerospace industry), and found major efficiency improvements.

(20)

2.3. Implementation of Design Automation

Design Automation (DA) is a collection name of methodologies and theories that allows businesses to ”cut lead times, workloads, and ultimately costs in order to become more

competitive” [9]. The implementation of a DA-system could become a tedious task, and

just like most implementation projects, requires proper planning where it is essential to identify and define a need and potential of the system to be implemented. [9] The idea is to automate repetitive and monotone design tasks, which would allow designers to focus on more problem-solving, creative and value-adding work [39, 1, 9]. Design automation covers tools and methodologies related to improving the design process, e.g. parametric modeling, PDM, PLM and KBE.

2.3.1. Implementation Preparation

Ferreira et al. [14] and Siddiqui et al. [35], claims that the problem today is no longer the technology, as technology and knowledge is something that is available and exists. The issue is how to best streamline the process of how to capture, use and re-use the knowl-edge through the technology. Businesses have different processes and ways of working, which results in a need to uniquely tailor the implementation of a design automation system to given business.

An implementation project of a DA-system could result in a significant change in the design process and overall business process. Salehi-Douzloo [34] means that the planning of a methodology implementation can be compared with the planning of a product de-velopment process, which could result in a complex, long and critical process. Luu [28] mentioned in his work, ”Changes are stressful whether they are positive or negative and

can therefore be met with resistance”, referring to the implementation of such a

method-ology in an organization. Design automation is in itself a type of methodmethod-ology and should therefore follow the same principles as any design implementation methodology. Cederfeldt [9] developed a structured method and approach for implementing design automation systems on a larger scale. He discusses the importance to incorporate the users into the planning phase to make the system usable for the intended users.

Siddiqui et al. [35], describes the process of implementing a PDM-system for the first time in an organization. A PDM-system and its functionality is the core foundation for a successful design automation implementation. The system acts like a communication tool between the different organizational disciplines and their software [35, 9]. Siddiqui et al. [35] further describes the reasons for a successful implementation which, if followed, could result in a better used and user-friendly system;

• good preparation and implementation; • good education and training of employees;

• clear strategy and specification of business processes; • good evaluation and selection of suitable systems.

(21)

Siddiqui et al. [35] performed a pilot study aimed to identify the difficulties companies encountered during the implementation stage of a PDM system. It had a focus on the issues surrounding a successful implementation and the obstacles experienced by the participating companies. Based on the responses, they identified lack of management

support and implementation issues as the two largest obstacles when implementing a

business changing system, such as a PDM-system. The implementation issues were due to not knowing the structured approach of the implementation, which made it difficult to know where to start. Approximately 65 % of the respondents indicated that lack of

management support was an obstacle.

The findings made by Siddiqui et al. [35], stresses the importance that information is shared with all potential users, owners, and management. The implementation needs to be supported from senior management, i.e. ‘Ensure that you have ‘‘true support’’ from

senior management i.e. they do not just agree but actively promote its usage’ [35].

With-out the support from senior management, it is a risk that the developed methodology will lack the number of resources and enough allocated time for it to be implemented. In order to give the users the opportunity to work with parametric modeling, it is impor-tant that the management team allow the designers to take the time to learn the tools and follow developed methodologies [35].

2.3.2. Implementation Strategy

The main goal with implementing parametric CAD and KBE is to reduce the devel-opment time while improving quality of a product. It is a challenging process and it requires extensive planning and a large-scale commitment.

As described in section 2.3, the implementation of a design automation system could result in a significant change in the design process and overall business process. [35] Furthermore, Bodein et al. [5] suggests that there is a lack of structured and effective approaches for deploying and implementing CAD methodologies to utilize the advan-tages of parametric CAD, causing the implementation of a design automation system to become even more complicated.

There are several studies that gives suggestions to methods and methodologies that demonstrates high potential for parametric CAD and highlighting techniques to achieve parametric models. Bodein et al. [5] claims that these suggestions doesn’t offer a clear vision on how to implement the methods and methodologies, or how to prioritize be-tween them in order to be more efficient in the design phase.

(22)

Bodein et al. [5] presents a roadmap for the future efficient use of parametric CAD mod-eling in the automotive industry, reviewing current research and trends. The roadmap consists of five phases;

1. Standardization 2. Methodology 3. Generic Modeling 4. Expert rules 5. Automation

The phases and their contents are illustrated in Figure 2. As seen, the main benefits in efficiency are achieved with low effort, through the initial phase of standardization. While the efficiency is further increasing with the following phases, the return in relation to the investment is diminishing. The standardization phase is divided into four sub-areas;

• CAD Infrastructure

Here, issues related to the management of software and licensing is addressed. • CAD Standards

Company-established standards addressing how the model is expected to be defined and structured in context of surrounding functions, e.g. the PLM system and other models, to ensure an efficient exchange of data.

• CAD Rules

Company-established rules which addresses how the model itself should be struc-tured and modeled, e.g. the choice of tools and functions inside a part. This step could be seen as a ”best practice” instruction to creating a model.

• CAD Data Quality

Ensuring that the model is of sufficient quality, by compliance to e.g. CAD stan-dards, rules and established naming conventions. As suggested by Bodein et al. [5], dedicated tools such as Q-Checker could be utilized to verify these.

(23)

Figure 2.: Interpretation of Parametric Design Roadmap by Bodein et al. [5]

2.4. Multidisciplinary Design Optimization

Multidisciplinary Design Optimization (MDO), as described by Martins and Lambe [29], concerns methods of applying numerical optimization to coupled systems involving sev-eral disciplines, i.e. systems where the performance is not merely determined by the in-dividual disciplines, but their interactions as a whole. Possible benefits of MDO includes reduced development time and thus cost [29], however, the method is computationally expensive. As Sobieszczanski-Sobieski and Haftka [37] explains, the computational cost does not simply increase as a sum of the individual optimization problems of each dis-cipline, but at an even higher rate. The computational cost can however be reduced at the expense of accuracy, by the use of various approximation techniques [37], such as meta-models [8].

(24)

2.5. Design of Experiments

Antony [2] defines a process as a ”...transformation of inputs into outputs”. Furthermore, Design of Experiments (or DOE) is described as a method of efficiently gaining knowl-edge about the input/output relations of a process. As described by Cavazzuti [8], DOE is ”...a way of choosing samples in the design space in order to get the maximum amount

of information using the minimum amount of resources...”. When DOE is performed,

de-liberate changes are made to the input variables, and the subsequent changes in outputs are observed. By use of statistical methods, models can be created from the observed data [2], which can be used to predict the process response. One such approximate method is response surfaces, also called meta-models [8]. Several methods of performing DOE exist, i.e. how to sample the design space efficiently. Common methods include

(25)

2.6. Manufacturing Methods

Choosing the best technical and economical casting method for manufacturing can be a difficult process. High cost in the casting process does not necessarily imply a more ex-pensive end product. It is important to look at the total cost, e.g. if the manufacturing process requires after-treatment of the component or not.

The component production quantity is a critical parameter when it comes to selecting the casting method. Different casting methods sets different limitations on the castings, the equipment and its tools. The castings geometry, i.e. the model to be manufactured, is typically affected by the casting method, which could result in a costly process if not considered in the earlier stages of product development.

When the product is being developed in a CAD environment the designer engineer needs to follow certain requirements in order to meet the demands from the manufacturer. The following are typical component requirements that needs to be considered when deciding on a manufacturing method; • Material • Quantity • Dimensional Accuracy • Surface Roughness • Cargo Weight • Geometrical shape • Cost

Depending on the component requirements, a decision needs to be made on what casting method is the most suitable for the component. Depending on the chosen casting method certain materials might be more or less suited for casting. Once the casting method have been chosen and the material is selected, some alterations and modifications might have to be performed to the component. It is at this stage possible to optimize and verify the design based on the casting method and feedback from the manufacturer.

(26)

2.6.1. Casting Methods

These are some of the most commonly used casting methods at Scania (NMGK); • Die Casting (High Pressure Die Casting) is a method where pressurized, molten

metal is injected into a permanent mold consisting of two or more dies. After the metal is solidified, the molds are separated and the part is ejected. The method provides good dimensional accuracy, surface finish, repeatability and is able to produce complex geometries with thin features [17, 38]. However, large production volumes are often required as the tooling cost is relatively high [38].

• Chill Casting2 _{(Gravity Die Casting) utilizes gravity to fill a permanent mold}

with molten metal. As the method permits the use of disposable sand cores, com-plex geometries with undercut internal features can be produced, which would be difficult or impossible using metal cores [38]. The tooling is cheaper than for die casting, making smaller production volumes feasible.

• Sand Casting is a method using disposable molds created from sand. The molten metal is injected under low or atmospheric pressure. The tooling is cheap, making smaller production volumes feasible [38]. As the thermal conductivity of the sand is lower than that for metal (of which the molds in die casting and chill casting are made), the cooling rate is slower which leads to worse mechanical properties [17].

2.6.2. Design Guidelines for Casting

The casting process can present some complications, requiring planning during the design stage to avoid manufacturing defects. Special care needs to be taken to compensate for the material shrinkage during solidification [17]. The following guidelines are common for all casting methods mentioned above;

• All walls/geometry which are perpendicular to the parting plane (called parting line in 2D) needs to have a draft angle, to facilitate withdrawal from the mold halves [38].

• To reduce the cost and increase the dimensional accuracy, it is desirable to have as few cores as possible, or ideally none at all.

• Uniform thickness is to be preferred to reduce the risk of porosity and shrinkage cracks during solidification.

• Sharp edges are to be avoided to reduce stress concentrations, hot spots and cracks.

(27)

2.7. Design Process at Scania NMGK

A development project at Scania NMGK, from start to finish, is an agile and iterative process. The process has three main phases; Concept Development, Product Develop-ment and Verification.

Concept Development

The concept development at Scania NMGK is divided into two phases; Concept

Devel-opment and Configuration Phase.

The concept development phase is typically initiated by a pre-development group. The group investigates the possibility to meet new performance requirements with existing articles. If they can’t meet the requirements the task is given to the development team, NMGK. A performance group typically sets the requirements based on information and demands from the marketing department. The NMGK group then designs the compo-nent in CATIA and evaluates the design with simulations, using tools such as CFD and FEM. Parallel to this work, NM, the engine development division, is planning the over-all engine layout together with its sub-divisions and groups. Representatives from each affected group present their limitations and requirements which leads to an agreement on the location and the design space for each group.

Depending on the given design space, the geometry might be affected which in turn could affect the performance of the component. NMGK then explore different design options to meet both performance requirements and limitations in the design space. Once a design is found and evaluated, a concept design is selected and the concept is taken into the configuration phase.

The configuration phase is an extension of the concept development and it allows more it-erations and evaluations of the design, but within a narrower design space and with more precise requirements. During the development of the conceptual design, some groups and disciplines might have to ”borrow” the design space from each other. During this phase a focus group can be created where affected parties attend. The goal of this focus group is to come to an agreement about the design space. It is imperative that each group can motivate their changes either by cost reductions, ease of production or performance enhancement. Once the attending parties comes to an agreement, final modifications are made and the process continues to the next phase, the product development phase. Product Development

Unlike concept development, the product development phase is driven towards improving and increasing the life-span of the products. The product development is a phase with

(28)

Each generation starts with model development in CAD, but unlike the concept

develop-ment phase, this CAD developdevelop-ment is more driven to make the model manufacturable

whilst optimizing the performance. Approximately halfway into the phase, a design freeze halts the modeling process which allows the groups to order their products for the upcoming assembling of the prototype engine. Once the engine is assembled, the engine is tested for longer periods of time to simulate the life-expectancy of the engine components.

Once the testing is done, the groups reviews the result and suggests improvements to the next design iteration. The suggested improvements are implemented in the design and tested in the next prototype engine. If all requirements are met and the products withstand the tests and simulations, they are further improved and prepared for mass production.

Verification

The Verification phase is the last step in the development process. In the last phase the last small changes can be made for the final prototype engine. Once the engine passes the tests and meets the given requirements, they can start sending the products to manufacturing. The development process concludes with a lessons-learned which gets documented to ensure implementation in future projects.

Some of the design related information is collected in an Excel spreadsheet, a design guideline, where the information is documented with a risk assessment and contingency plan. The design guidelines (unique for each group), aids in ensuring improvement of product quality and manufacturability.

(29)

2.8. Previous Work at Scania

Previous theses at Scania has had similar purpose and objectives as this thesis [28, 3, 25, 31], i.e. developing methodologies for creating parametric CAD models. Preceding these, a thesis work examined the possibilities of implementing KBE in CATIA V5 [27]. The following section will provide a short review of previous work.

2.8.1. KBE capabilities in CATIA by Lundin and Sköldebrand [27]

The capabilities of CATIA V5 in regards to KBE has been thoroughly examined in pre-vious theses. From the work conducted by Lundin and Sköldebrand [27] (2008), it was concluded that a specific group of experts working with KBE would be beneficial for Scania, aiding with the implementation and maintenance of parametric models. This was reinforced by Blomberg [3], however, at the time of writing this has yet to be imple-mented. Moreover, as discussed by [27], simpler knowledge in the form of tools such as

rules, reactions and checks in CATIA V5, while lacking the potential of pure

program-ming tools, could more easily be implemented without the support of a specific expert group.

Furthermore, interviews conducted by Lundin and Sköldebrand [27] revealed a strong opinion of the intended users that the knowledge embedded in the models should not inhibit the work of the designer, but rather advise and alert. This raised the possibility and desire of documenting the recommendations and advice typically ignored by the user, to gain a better understanding of the model history.

2.8.2. Methodology development by Luu [28] and Blomberg [3]

Two theses [28, 3] were conducted simultaneously during 2015, with the same purpose and goals, i.e. developing a methodology for parametric CAD modeling in CATIA V5, with special focus on CFD. Case studies on a turbine volute and on intake ports were used in the two studies. The final methodology was created by merging the results from the two theses.

(30)

The proposed methodology of Luu [28] and Blomberg [3] which is divided into three main phases, can be seen in Figure 3. The first phase, pre-CAD, takes place before the usage of any CAD tools, as the name implies. Here, any information necessary to create a parametric model is established and collected, such as interfaces, required parameters (and their relationships), and desired outputs to various departments. This is done by meeting with various stakeholders, and documenting the progress in an established Ex-cel spreadsheet which is unique for the article. The spreadsheet contains a number of matrices based on PARAMASS [34, 28] (see subsection 2.1.1).

The second phase, CAD, is where the actual modeling takes place. The phase includes moments such as decomposition into simpler sub-geometries and parameter creation. Emphasis is put on organizing the model structure, renaming features logically and making sure the model is stable. The phase is partly based on Explicit Reference

Mod-eling Methodology [6] and Resilient ModMod-eling Strategy [15] (see subsection 2.1.1).

The third and final phase, evaluation, is used to ensure the robustness of the model and compliance with the steps in the methodology. The robustness is checked by performing DOE (see section 2.5), which is integrated in the previously mentioned Excel-spreadsheet. By using checklists, compliance with the methodology is ensured (partly based on

Re-silient Modeling Strategy [15] see subsection 2.1.1).

2.8.3. Methodology development by Jansson and Wiberg [25]

As a continuation of the methodology created by [28, 3], further studies were conducted in 2016 by Jansson and Wiberg [25]. Case studies of a compressor wheel and a compressor volute were made, to further develop and expand the methodologies of [28, 3]. The new methodology had, contrary to earlier work, a larger focus on multi-disciplines, i.e. to incorporate requirements and knowledge not only from CFD by also FEM and to some extent manufacturing.

(31)

The methodology created by [28, 3] was used as a starting point for the development, which becomes evident when comparing Figure 3 and Figure 4, where the initial phases are similar. There are however some differences, especially in the evaluation phase. As described by [25], an initial FE-analysis has been added (using GPS/GAS3), to give the design engineer an indication of the performance and reduce the number of iterations between the design- and CAE-engineers. Furthermore, as described by [25], this can allow for basic optimization of the component, possibly in PEO4_{. As discovered by}

Jansson and Wiberg [25], the developed methodologies by Luu [28], Blomberg [3], and Jansson and Wiberg [25] has apparent similarities with the KNOMAD methodology, see subsection 2.2.2.

(32)

3. Thesis Methodology

The methodology used in this thesis has followed a structured and systematic approach which has been implemented and evaluated in a realistic environment. Initially a data collection was performed through literature study, analysing prior work at Scania, in-terviews and questionnaires. The gained knowledge and data was then implemented in a suggested methodology by following NMGK’s design process. Through iterations the methodology was documented and improved which eventually resulted in a final suggested methodology.

Figure 5.: The methodology applied in this thesis

3.1. Data Collection

As the main purpose of this thesis was to improve upon already developed methodolo-gies, existing work was utilized to the largest extent possible, and complemented with additional relevant information.

(33)

3.1.1. Literature Study

The literature study was conducted to understand the underlying theoretical concepts on which the results presented in previous theses are based on, allowing for further development. Additionally, great consideration was taken to the findings and discussions of previous theses regarding the implementation process, with its potential obstacles and incentives.

3.1.2. Interviews & Questionnaire

To gain knowledge regarding internal processes and procedures at Scania, as well as to get an assessment of the mind-set and awareness of parametric modeling among the employees, the literature study was complemented with interviews and a questionnaire. An area of special interest was how previous work had been presented, and why the findings had not been implemented to the extent possible. Some of the information of interest to receive from design engineers was;

• Is there an awareness of the possible benefits of using parametric CAD models? • Is there an awareness of the available tools for creating parametric CAD models? • To which extent is parametric CAD modeling used in practice?

• What factors are prohibiting the more widespread use of parametric CAD models? • Has there been any attempts (from management or within the group) to implement

the results of previous theses, or parametric CAD modeling in general?

• When in the development process is manufacturing aspects considered, and are they likely to change during the course of the project?

Interviews

Due to the open nature of the questions, semi-structured interviews were deemed the most suitable method to obtain qualitative data, as it allows the interviewees to elaborate and explain their reasoning and perspective on the topic in a more flexible way than in a structured interview [13]. This is beneficial as it opens the possibility of adjusting the direction of the interview, and even including topics previously not thought of, depending on the response. There were however a set of questions or topics to start the discussion from, in the form of an interview guide (see Appendix A). The design engineers to be interviewed were mainly sampled from the NMGK department, as the thesis was conducted there.

(34)

Questionnaire

To complement the interviews and to collect quantitative data from a larger target demographic, a questionnaire was conducted. The questionnaire was sent by e-mail to users with access to a CATIA license, at the department of engine development (NM). To ensure that the questionnaire was only answered by the intended target audience, a control question asking whether the respondent were using CATIA on a daily basis was included. The questionnaire allowed for easier comparison with data from previous theses, and to assess any changes made since.

3.2. Methodology Development

After the data collection, the methodology was developed by following the second phase in Figure 5, Methodology Development. The methodology development phase was a iter-ative process, divided into five different steps. By using the case study as the foundation for the development, the methodology could be tested and evaluated in a environment similar to the reality at Scania.

3.2.1. Formulation of modeling methodology

As previously mentioned, the goal of this thesis has been to improve and further develop methodologies from previous theses [3, 28, 25, 31]. Available results were used as a start-ing point, and by applystart-ing new literature and findstart-ings, a first methodology hypothesis was developed. The methodology was developed with a multi-disciplinary mindset, fo-cusing on manufacturing aspects. The information from the data collection was used as an identifier to guide the research through applicable areas to investigate. Through the case study new knowledge was gained and implemented in the iterative process until the research questions could be answered.

3.2.2. Case Study Component

One of the many components used in the continuous struggle to meet the stringent re-quirements on efficiency and emissions for trucks is turbochargers, which works on the basis of utilizing energy otherwise lost through the exhaust gases of internal combustion engines. A turbocharger consists of a turbine and a compressor, paired through a shaft. The turbine is driven by the exhaust gases of the engine, driving the compressor which forces air into the combustion chamber. The use of turbochargers allows for downsizing of the engine, potentially reducing the fuel consumption and emissions while preserving, or even increasing the power output [32].

(35)

To ensure high efficiency of the turbocharger, the inlet air to the compressor needs to be as uniform and free from disturbances as possible [40, 30]. However, the lack of space inside the engine compartment is a complicating factor, as the sharp bends often required to route the inlet pipes causes adverse conditions for the airflow entering the compressor. The issue could be remedied by using a radial inlet, occupying a smaller axial space. Therefore, by Scanias request, a radial inlet was chosen as the case study component.

(36)

4. Results

The results of the thesis is presented in four different sections. In the first two sections the results from the interviews and the questionnaire is presented and summarized. The third section covers the methodology development, which describes the deficiencies found in the methodology that the thesis aims to improve. Finally the developed methodology, i.e. the final methodology, is presented.

4.1. Interview Results

All interviews were held with design engineers and development engineers, as their input was deemed most important to the findings of this thesis, seeing as they are the intended users of the methodology to be developed. In total, 7 interviews were conducted, and the following section will summarize the key findings.

4.1.1. Information Collection

Some respondents mentioned a lack of usage of tools and developed design documenta-tion, such as design guidelines (see section 2.7). The designers are aware of the docu-mentation but consider themselves knowledgeable enough of its content, which makes the documentation redundant to continuously use. Furthermore, an expressed opinion is the lack of integration of the documentation in the design process, the format is cumber-some and the storage location is not a natural part of the modeling process. Instead, the process of gathering and exchanging information related to the geometry of the model is to a large extent made with personal meetings between design engineers, without any explicit documentation [18, 19, 20, 21, 24]. The pitfalls to such approach is that the relation to the person in question can influence the ability to compromise regarding the design space [19]. Furthermore, [19] elaborates that it would be better to take rational and fact-based decisions based on profitability. [21, 24] stresses the importance of find-ing the right person to talk to, which has proven to be difficult at times.

There exists a system, GEO, in where all engine related models are published, which aids in finding available design space and components in the close vicinity [18, 21]. Some respondents expressed a difficulty in finding among internal documents, due to counter-intuitive software. Steps has however been taken to replace such software, improving the ease of usage [19, 20].

(37)

4.1.2. Manufacturing Aspects

As described by most respondents, contact with the supplier often occurs late in the project, when it is too late to make any major changes to the design. However, aspects such as the manufacturing method and its limitations has already been considered, with knowledge originating from experience. After contacting the supplier, minor modifica-tions due to supplier-specific demands are conducted [18, 21, 22, 24]. While it would be beneficial to have the supplier contact earlier to reduce modifications [22, 21], it could also be disadvantageous as the design becomes more tailored to a specific supplier, reducing competitive advantage [21, 20].

4.1.3. Parametric Modeling

The interviews revealed a varied knowledge of parametric modeling and the capabilities of CATIA among the interviewees. Most are using explicit parameters to some extent [21, 20, 19, 18, 23], however, the usage of formulas or more advanced tools is rare. Some respondents were aware of the knowledgeware functionality of CATIA, but there was a difference of opinion to the usage of it. While some had not found any meaningful areas of usage [18], others saw great potential in implementing logic into their models [19]. The main benefit of using parametrized models, as seen by the majority of interviewees is the possibility of optimizing for performance, by auto-generating designs [19, 20, 21, 22, 24, 23]. Another benefit was expressed, as having a structured, robust and well thought-through model is a pre-requisite for parametrization, raising the quality of the models [19]. Furthermore, [18] sees the main benefit in allowing quick changes to the geometry, saving time and as a result cost. While all respondents were more or less pos-itive to parametric modeling, some commonly expressed drawbacks were that it might introduce unnecessary and time-consuming work [20], and that the model with its intri-cate relations can become complex [21, 22, 20].

When asked if there had been any initiative to use parametric models from manage-ment or others, it became apparent that no major effort had been made [18, 19, 20, 21]. However, results of previous theses (especially the work of [28], with associated tutorial) had sparked the interest of the users (i.e. design engineers) [18, 19]. Furthermore, the performance groups had shown a great interest in parametrized models to allow for op-timization [18, 19, 23]. While no initiative had been taken from management, there had neither been any resistance when asking for permission to spend more time to look into the matter [18, 24, 23]. However, the perception of several respondents [19, 20] was that the designer itself still must take time from other task to be able to do so, as no addi-tional time is given. Furthermore, [20] expresses a low understanding from management as to why the benefits of parametrization can motivate a larger initial time investment,

(38)

When asked about implementation procedures, some respondents were in favor of estab-lishing a requirement on using parametric models, but at a level suitable for all regardless of skill or interest [19].

Model Structure

A commonly expressed opinion of the design engineers was that having a clean and struc-tured model is important, but most design engineers had a different idea as to how the entities should be organized. While some believed that all entities should be renamed and grouped based on intent [18, 24], others were of the opinion that grouping them was most important [21].

When asked whether a convention for model structure existed at the company, most interviewees referred to the start-up templates, containing a set of default geometry sorted based on type. However, as apparent from the answers, the users has the free-dom to change the ordering, naming and grouping of entities themselves (and does so frequently). All were however positive to the idea of using a standardized method of structuring the model tree, as long as it doesn’t pose an obstacle to the user by being to controlling and restrictive [20]. Attempts to do so had already been made by one of the interviewees [18]. Furthermore, many of the respondents mentioned a lack of quality control for the 3D models [20], which in contrast to 2D-drawings has few methods in place to control their quality. While it exists a tool, Q-checker, intended to ensure the quality of the models by compliance to a set of company defined standards (as described by [22]), the tool merely asks the user to correct any discrepancies detected, it does not verify that such corrective action has been taken.

(39)

4.2. Questionnaire Results

The questionnaire was sent to 238 persons, of which 35 had either stopped working at Scania, or auto-responded with an out of office message, yielding a total of 203 possible answers. A reminder was sent 8 days after the initial e-mail. The questionnaire had 109 respondents, yielding a response rate of 54%, which is very close to that of previous theses with a similar approach and target [28]. In total, 73 responses were received from the target group, i.e. daily CATIA users. The questionnaire response in its entirety can be seen in Appendix B, and a short summary of the questionnaire results will follow;

4.2.1. Design Intent & Model Structure

A vast majority of the respondents states that modifying geometries of their own models is easy and takes little time, however, the opposite is true for models of other design engineers. Furthermore, the bulk of respondents states that they both rename and group features and entities based on intent, while believing that the model structure and design intent of models created by other design engineers are hard to understand. The majority of respondents state that they follow a specific methodology when creating CAD models.

4.2.2. Knowledge-Based Engineering

A majority of the respondents did not find the CAD modeling process to be repetitive and monotone. Most respondents considered themselves aware of KBE and associated tools in CATIA. When asked which tools they used, most answered that they were using explicit parameters and formulas occasionally. A majority had however not used any advanced tools involving programming (such as rules, reactions and VBA programming) at all. Those who had never used any KBE-tools in CATIA, stated reasons such as a lack of knowledge of how and why to use them.

A small majority of the respondents stated that no initiative to use KBE tools had been taken. However, a relatively large amount (48%) of respondents stated that initiative had originated from either themselves or colleagues. Only 7% stated that such initiative had originated from management.

4.2.3. Information Exchange

When asked regarding the ease of information exchange between departments, the ma-jority of respondents answered that the process was neither easy nor difficult, with some outliers on either end of the spectrum.

(40)

4.3. Methodology Development

The component assigned for the case study was a radial inlet which is currently being re-searched and developed at Scania. The component is in a conceptual phase which limits the amount of detailed knowledge of the component. The previous work (section 2.8) at Scania has been conducted with, but not limited to, already existing components. Choos-ing a component in an early conceptual stage, with the intention to identify potential issues and possible drawbacks with the current methodology, made it possible to reveal the potential reasons to why the methodology has yet to be implemented as intended. By following this approach, it was also possible to address the issues mentioned by the design engineers during the interviews.

The developed guideline to the current methodology, which is the only documentation of parametric CAD modeling available to the design engineers, was thoroughly followed throughout the case study. This approach was considered a suitable way of evaluating and suggesting changes to said methodology.

4.3.1. Information Collection

As described in the section 2.8 the pre-CAD phase is the first step in the methodology. The main purpose is to gather information and requirements about the component, then define its interfaces as well as possible parameters to control. This was done through a planning meeting with involved parties, before using any CAD tools. The gathered information was collected in a pre-defined document. The pre-defined document, an Excel workbook, is divided into eight different sheets, where six of the sheets are intended to be filled out before starting the CAD modeling phase (Figure 6).

Figure 6.: Overview of the Excel sheets

In the upcoming sections, the issues discovered in each of these sheets are being presented and proposed with a solution. The first sheet is the ”information” sheet, which contains the overall component information. The sheet gives the user a good overview of who is responsible for a certain discipline, what project the component is intended for, the article number and the status of the component.

(41)

Interfaces

Establishing what interfaces the component had, was fairly straightforward. However, determining the design space and its limitations, and how those affects the component proved to be more problematic. To differentiate the ”second-level interfaces” (surround-ing components, who are in no direct contact) from the ”first-level interfaces” (connect-ing components, who are in contact), with the use of the correlation matrix (right section in Figure 7), was difficult and did not describe the correlations in an simple and intuitive way. Furthermore, if a component is listed in the interface matrix, it naturally affects the base component in some way, and a correlation matrix would not give the answer to how it affects the base component.

Figure 7.: The Interface Matrix

In the updated interface matrix, Figure 8 the correlation matrix has been removed and replaced with a dropdown-list that defines the type of interface;

• Base Component: The component to be created, i.e. dependent on the interfaces listed.

• Connecting Component: The components that have direct contact with the Base Component, e.g. connecting inlets.

• Surrounding Component: The components that have no direct contact with the Base component, but can limit the design space.

Additionally, a description column has been added to further describe the interfaces and their impact on the base component. During the pre-CAD phase there is a possibility that it could be hard to determine how the known interfaces affect the base component. The description field could therefore be used to describe the interface further, once that knowledge has been obtained.

(42)

Data Exchange

The main purpose of the CAD model in the conceptual phase was to establish a concept with feasible performance, i.e. flow characteristics determined by means of CFD. Thus, the main stakeholder to be documented in this phase was the calculation department, who required solid geometry of the flow volume, in the format of STP-files. In this step, a need of examining and documenting the objective of each stakeholder was discovered, to get a better understanding of the design choices to be made during the development process. Thus, the data exchange matrix was extended with a field intended for docu-mentation of objectives (see Figure 9). The main objectives and performance indicators of the CFD department was flow uniformity and pressure drop. Additionally, for sake of demonstration, FEM calculations were also added as a required data exchange. The ob-jective with the FEM calculations was to ensure that the limitations to natural frequency was met.

Figure 9.: Revised Data Exchange Matrix

Requirements and Constraints

Determining requirements and constraints proved to be quite problematic, as the compo-nent was still in the early conceptual phase at the time of the case study. The main geo-metrical requirements imposed on the component by the manufacturing method (gravity die casting) was an outer draft angle (to ensure ejection from the mold), a minimum material thickness, and an even thickness distribution throughout the component.

(43)

The requirements were documented in the Requirements and Constraints matrix (see Figure 10). While the matrix makes a distinction between requirements and constraints, it was unclear as to what the difference between the two is. To avoid any confusion and the possibility of getting caught in technicalities, it is recommended to remove this distinction and instead write a comment regarding the requirement, to clarify its purpose. Furthermore, logical operators (e.g. =, <) was added to the value column, to indicate acceptable values. Finally, the column previously called parameter was renamed to implementation, to allow planning and documentation of not only explicit parameters, but knowledge-based features as well (see Figure 11).

Figure 11.: Updated Requirements and Constraints Matrix

PDM - Parameter Definition Matrix

Defining the parameters and their relations are done by meeting stakeholders and collect-ing the necessary information. Defincollect-ing the parameters at this early stage was noticeably complicated, even with the assistance of stakeholders, and only a handful of parameters could be established. The main reason for this was the lack of detailed knowledge of the component due to its conceptual status. Furthermore, PDM is conventionally used as an abbreviation for Product Data Management and should therefor be avoided, referring

PDM to anything else, is prone to cause confusion. However, the matrix is considered

a valuable tool in the Pre-CAD phase but could be problematic to populate prior to developing a CAD model. The process of developing a parametric model have proven to be quite iterative and certain information, namely Min.Value, Ref.Value, Max.Value and

Step Size (see Figure 12), could therefore become restrictive rather than value-adding

for the designer if they are required to be filled out prior to the modeling process. The recommendation is therefore to populate the matrix as far as possible before starting with the CAD process, but can be updated later once more knowledge of the model has been learned.

Parametric CAD Modeling to aid Simulation-Driven Design : An evaluation and improvement of methods used at Scania