Modelling theories, methodologies and approaches
In the last decades, many methodologies and frameworks are proposed for concurrent designing (Cutkosky & Tenenbaum, 1990; Domazet, 1992; Finger et al., 1992; Talukdar & Fenves, 1989) and to show how design fusion can save cost and time. Thurston & Locascio (1993) propose the following steps for IPPD to have a systematic methodology:
1) Define attributes from customer attributes
2) Replace ‘relative importance’ with multi-attribute utility analysis
3) Define design decision variables
4) Define constraint functions
5) Determine the attribute bounds
6) Structure the optimization problem: design to maximize utility
7) Initiate the computational approach
In the process of design, a series of decisions should be made by the designer/ employer/ client to have a robust result. Decision-Based Design (DBD) is one of design approaches that is based on a series of decisions that the designer must take during the process (Simon, 1960). The concept of DBD in concurrent designing is an important concept as discussed by (Westfechtel 1996; Hale et al. 1995). The expected result is an optimal personalized design solution with the cost of mass production. Thus, it is based on the fact that design is a process of decision-making in order to maximize the value of the design (Hazelrigg, 1998; Muster & Mistree, 1986; D. L Thurston, 1999; Wassenaar & Chen, 2001).
In each level of designing a product or process, designers face with several possible alternatives which they need to choose among. Every choice is directly or indirectly related to another. In addition, the decisions of product designers effects on process designers and vice versa. So, a proper approach to achieve an optimal design solution is vital. However, this is the case when all the variables are known and their performances are predicted but in most cases there is lack of knowledge in each level of design caused by many reasons. There is a growing recognition on decision-based design and many researches have been done on creating a framework to model and implement this method of design (Wassenaar & Chen, 2001; Wood, 2004). The research gap is to have a systematic framework that while it models the system, it can increase the knowledge of the designer in each level of design to help them in decision-making process.
In that case, DBD can be used for designing complex systems since it can break it down into smaller, more manageable sub-systems and the designer makes the design based on the choices. Nevertheless, the result is not guaranteed since it is depend on the choices that the designer makes.
The proposed approach of this thesis consist of a series of decisions that designer needs to make in the process. So, this concept is used in product-process relation, function-structure relation and the relations between models in different levels of decomposition.
In the next sub-section, various design theories, methodologies and approaches are presented to be used in order to manage complexity in IPPD. A decision-based design methodology which leads to a robust product and process is kept in mind while analysing them.
Design Theories and Methodologies (DTM)
Over the years many design theories and methodologies have been developed. Tomiyama et al. (2009) and Le Masson et al. (2013) gathered some of these methods. Tomiyama (1997) categorized them in two axes of “general vs. individual” and “abstract vs. concrete”. Among DTMs, the related methodologies in IPPD are presented shortly here. The objective is to compare different methodologies and choose the best option to use as a framework and a systematic decomposition approach with coherence in both product and process design aspects in early stage of design. So, related DTMs are compared according to the following criteria:
1. It can be used as a framework for design
2. It has a systematic approach
3. It can be used for product design
4. It can be used for process design
5. It creates a coherence between product and process
6. It reduces complexity in design
7. It has the ability to integrated mathematical equations
8. It is easy to understand and learn
9. It has a robust approach
10. It can integrate other methods
One of the methodologies is Adaptable design. It is one of the DTMs to modify an existed design in order to optimize the quality and lead time. It creates designs and products that can be easily adopted for different requirements (Gu, Hashemian, & Nee, 2004). The method works on the existed population of products rather than a complete new product. It designs the product adoptable for different environment so, it is suitable for robust design. Although the method creates a link between product and process, it considers only specific aspect of design and cannot be generalized and used as a framework of design.
Axiomatic Design is a widely used method created by Nam Suh (Suh, 1990, 2001). The method is based on two axioms; 1) Maintain the independence of the functional requirements (The independence axiom), and 2) Minimize the information content of the design (the information content). This method is very suitable for designing complex products and it also creates a good framework. It can be integrated with other methods and works well with mathematical equations. However, it doesn’t have an algorithmic and systematic approach that is easy to understand for everybody. Though, thanks to its systematic approach, it can be used for decomposition approach of the proposition of this thesis. Axiomatic design is discussed further in section 2.5.2.
C-K Theory developed by (Hatchuel & Weil, 2003, 2009) aims at creating a unified formal framework for design. According to this theory the design can be modelled between two spaces of Concepts (C), and knowledge (K). This is a good theory to have a global overview of the design and creating a framework. However, it does not include mathematical relations and coherence between product and process, because it is rather a conceptual framework.
Characteristics-Properties Modelling (CPM) (Weber, 2005a, 2007, 2008) is a modelling framework which is based on the distinction between Characteristics and Properties of a system. It uses an approach called Properties-Driven Development to gradually create the model of product. It includes mathematical relations between characteristics and properties. It can integrate some other methods. Because of its advantages, CPM is used as the framework of the proposed approach and it is explained in section 2.5.4. Nonetheless, it is not enough since identifying the relations is not easy and it needs some improvements to be used for complex systems. An improved version of CPM is presented in the next chapter. Moreover, CPM is introduced for product development and does not include process design considerations. So, this is another extension of CPM in this thesis.
Contact and Channel model of Albers (C&CM) allows the designer to establish an integrated model where technical functions, shape as well as the environment in which the system should perform are represented coherently (Albers & Alink, 2007). In decomposition of complex systems, designers might lead to different description of the system. It cannot be used as a framework for complex systems but it comes more useful as a tool in integrated design.
Function-Behaviour-Structure (FBS) identifies the elements of the system as function, behaviour or structure. By using knowledge representation diagram, in a systematic approach, the designer starts from the function of the product and therefore the behaviour of the system to find the proper structure and therefore the design of the product (Gero & Kannengiesser, 2004). This is a suitable framework for designing complex systems. So, this model is discussed further in 2.4.3.
Hansen method, created by Friedrich Hansen, is a structured approach with three main stages of task clarification, reasoning on functions and working principles, and layout and detail design (Hansen, 1966). In this method, Hansen defines a system as three sections of Structure (S), Function (F) and environment (U). As (Hansen, 1974) stated, there is a meaningful relationship between these three sections of any system. The method creates a language close to FBS approach. Considering the analysis/synthesis approach in this method and using mathematical equations, it might be the closest approach to the proposed approach of this paper. This method is a good base for more developed methods introduced after.
Hubka and Eder is another basic framework among DTMs. This method consists of 1) Considerations on the objects being designed and their properties 2) Statements and recommendations about the process of and useful operations in designing 3) A concept of how to structure of design-related knowledge (Hubka & Eder, 1996). This method can be perfectly integrated in CPM method. Hubka and Eder divide the system into two types of properties; ‘internal properties’ (which is called ‘characteristics’ in CPM) and ‘external properties’ (which is called ‘properties’ in CPM).
Koller is a physically and algorithmically oriented design method (Koller, 1998). Although this method can be a good base for developed methods, it does not use mathematical equation to simplify the simulation phase.
Systematic design of Pahl and Beitz is based on an elaborate analysis of the fundamentals of technical systems, the fundamentals of systematic approach and general problems solving processes (Pahl et al., 2007). It considers the whole life cycle of the product and could be proper framework for integrated design. It includes four phases of planning, conceptual design, embodiment design and detail design. It is useful especially in the early stages of the design. This method although gives a systematic approach of design but it might not be proper for complex systems though can be used as a complementary tool.
Taguchi method is a widely used method that aims at improving the product and process quality by eliminating losses (Taguchi, 1987). According to Taguchi, two types of variables (factors) are defined in robust design: Control factors (easy-to-control) and Noise factors (hard-to-control). Noise factors might have different sources such as external noise, internal noise, and unit-to-unit noise. The aim of robust design is to determine the control factors to achieve the best performance that is insensitive to the variability of noise factors. The recommendation of Taguchi is to perform specific experiments to determine and set the control and noise factors using orthogonal arrays. This method provides a simple and systematic framework for identifying critical characteristics in systems to achieve best quality characteristics while minimizing the variation and cost. Mathematical equations can also be integrated in the model. Although it looks like the method satisfies all the criteria, but it cannot be used as a framework to explain the product and process model. Taguchi method focuses on the wastes. So, it is not used in the approach of this thesis but its loss elimination approach toward robust design is used in chapter 6 to assure the robustness of the approach achieving a robust product as a result.
Total design of Pugh is a design framework for a structured design process model for application of design methodology in design practice by industrial practitioners (Pugh, 1991). It is also a DBD method. The core of this method is product design specifications. So, mathematical equations are integrated already. It is an easy to learn methodology and can be integrated with tools like QFD. However, it cannot be used for process design and create a coherence between product and process design.
Universal design theory (UDT) combines findings about product design from various scientific disciplines in a consistent, coherent and compact form (Grabowski & Lossack, 2000; Grabowski, Rude, Grein, Meis, & El-Mejbir, 1998). This theory focuses on two problem of universality which it refers to the mutual understanding of engineers in different disciplines, and the problem of practical applicability which refers to explicitly and complicity of the process. The theory creates must-have characteristics for the used method. Yet, it cannot be used as a framework, or alone to design a complex product.
Mechanical design process of Ullman is an approach focuses on mechanical design process (Ullman, 2002). It has a step-by-step approach, it is clear to understand and it is practical. It also creates a language for the designer. However it cannot widely be used for complex products and designing processes.
Ulrich and Eppinger method focuses on complex product development. The approach simply includes four phases; 1) Make a schematic representation of the product, 2) Cluster components within this scheme, 3) Make a hard geometric layout, and 4) Identify fundamental and incidental interactions (Ulrich & Eppinger, 2007). The method has a general approach and can be integrated with DfX approaches. Nonetheless, finding the interaction of the elements based on the functioning of the product is a challenge.
Function-Behaviour-Structure (FBS) which first introduced by Gero (1990) is a method to model a system based on its function, behaviour and structure. Function is defined as an intermediate between the needs and the behaviour of a system. Behaviour describes the attributes that are derived (or expected to be derived) from the structure. Structure is the components of the system and their relationships. There are eight processes in FBS framework as shown in Figure 2.12.
First, the design requirements (expressed as function (F)) are transformed into the expected behaviour of the system (Be). Then, the expected behaviour is transforms into the desired behaviour which is the solution structure (S). The actual behaviour (Bs) is derived from the structure. The forth process is evaluation by comparing the expected behaviour with actual behaviour. Comparing Bs with Be is actually the difference between the objective design and the resulting design. So, the goal of the designer is to minimise this difference. Step five is the documentation which produces the design description (D). After that, there are three types of reformulation in terms of structure variable, behaviour variable and function variable if the actual behaviour is evaluated to be unsatisfactory (Gero & Kannengiesser, 2004).
In another publication from Gero (Gero & Kannengiesser, 2004), the situated FBS framework is introduced. It includes three design environments which are called expected world, interpreted world and external world. In this framework the number of processes increased to 20 so it can be able to deal with the agent’s interaction processes with the external world and within itself, interpretation, focussing and action.
In FBS, by using knowledge representation diagram, in a systematic approach, the designer starts from the function of the product and therefore the behaviour of the system to find the proper structure and therefore the design of the product (Gero & Kannengiesser, 2004; Gero & Neill, 1998).
FBS has a systematic approach for modelling products. It divides a system into different domains based on its function, structure and behaviour. So, it can be integrated in axiomatic design to create the desired approach. With some modifications, it can also be used for modelling processes. It is suitable for designing systems with complexities. However, the modelling would be conceptual. Because of the nature of relationships in FBS, comprising mathematical formulas is not possible. In managing complexity in product and process, the relationship between qualitative model and quantitative model is important. Due to the conceptual relationships between function and structure through behaviour, the links are not quantitative. In addition, including mathematical formulas is important in evaluation, simulation and calculating “value”. If complexity cannot be measured, it cannot be reduced.
About ten years ago Weber developed an approach called Characteristic-Property Modelling (CPM) for product/system modelling and respectively Property-Driven Development (PDD) to explain the process of developing and designing the products (Weber et al., 2003; Weber, 2009).
The concept of CPM/PDD is mainly based on the distinction between “characteristics” and “properties” of a product or proces. (Tomiyama et al., 2009; Weber et al., 2003).
Characteristics (Ci) are the parameters that can be directly influenced or determined by the designer. For instance: shape, structure, dimension, Bill of material (BOM), material and surface of the product.
Properties (Pj) are the product behaviour, which means, the parameters that the designer cannot change directly but they can be changed indirectly by means of the characteristics. For instance: function, weight, aesthetic properties, safety and reliability, cost, manufacturability.
Required Properties (RPj) are the parameters that the designer/customer is desired to achieve.
Characteristics are also called “internal properties” or “design parameters” in other methods, and properties are also called “external properties” or “functional requirements” (Hubka and Eder, 1996; Suh, 1990).
Dependencies (Dij) are between the characteristics which could be geometrical, spatial, or material.
Relations (Rj) represent the interrelation between characteristics and properties.
The relations correspond with two main activities; analysis and synthesis.
Analysis; Based on known/given characteristics of a product its properties are determined or, if the product does not yet exist in reality, predicted. Analyses can, in principle, be performed experimentally or ‘virtually’ (e.g. using digital simulation tools).
Synthesis; Based on given/required properties, the product’s characteristics are to be determined. The development/design process begins with a list of required properties. The designer’s task is to find appropriate solution patterns and determine/assign their respective characteristics in such a way that the required properties are met to the customer’s satisfaction (Weber et al., 2003).
Table of contents :
CHAPTER 1: Introduction
1.2. Research questions
1.3. Research objectives
1.5. Thesis outline
CHAPTER 2: Literature Review
2.3. Concurrent designing
2.3.1. Integrated Product and Process Design (IPPD)
2.3.2. Feature-based design (FBD)
2.3.3. Design for X (DfX)
2.3.4. Concurrent engineering (CE)
2.4.1. Definition of Complexity
2.4.2. Classification of Complexity
2.4.3. Proposed Solutions for Complexity
2.4.4. Complexity – Our Position
2.5. Modelling theories, methodologies and approaches
2.5.1. Design Theories and Methodologies (DTM)
2.5.2. Axiomatic Design (AD)
2.5.3. Function-Behaviour-Structure (FBS)
2.5.4. Characteristics-Properties Modelling
CHAPTER 3: A Proposition for Product Modelling
3.2. An extended version of CPM
3.3. Energy flow modelling
3.3.1. Product flows
3.3.2. CTOC: An energy flow model
3.4. A proposed approach for product modelling
3.4.1. Level 1
3.4.2. Level 2
3.4.3. Level n
3.5. Case study: Hair dryer
3.5.1. Level 1
3.5.2. Level 2
3.5.3. Level 3
CHAPTER 4: Case Study: Product Modelling Approach
4.1. Oil pump: a state of the art
4.2. Product modelling
4.2.1. Level 1- System analysis
4.2.2. Level 2- System decomposition
4.2.3. Level 3 – Identifying the characteristics
4.3. Complementary tools
4.3.1. Functional analysis
4.3.2. Primary flow analysis
CHAPTER 5: Uncertainty & complexity management in product design
5.1. Uncertainty management
5.1.1. Uncertainty taxonomy
5.1.2. Robust design
5.1.3. Epistemic uncertainty mitigation
5.1.4. Uncertainty elicitation by the proposed approach
5.1.5. Case study: uncertainty management
5.2. Sensitivity analysis
5.2.1. Local Sensitivity Analysis (LSA)
5.2.2. Global Sensitivity Analysis
5.3. Tolerance analysis
5.3.2. Using the proposed approach in tolerancing
CHAPTER 6: A Modelling Proposition for Integrated Product/Process Design
6.2. Flows in product and process
6.3. CPM in Concurrent designing
6.4. The proposed approach for concurrent modelling
6.4.1. Determination of process model
6.4.2. Mapping: OTCS – CPM
6.4.3. Decision making in IPPM
6.5. Case study
6.5.1. Determination of process model of the oil pump
6.5.2. Detail process model of the oil pump
6.5.3. Risk analysis of the process of the oil pump
6.5.4. Finalising and decision making
6.6. Application: Tolerance optimisation
6.6.1. Problem formulation
Conclusion, limitations and perspective