Fig 9 - uploaded by Matthew I. Campbell
Content may be subject to copyright.
Source publication
Function structures are used during conceptual engineering design to transform the customer requirements into specific functional tasks. Although they are usually constructed from a well-understood black-box description of an artifact, there is no clear approach or formal set of rules that guide the creation of function structures. To remedy the un...
Contexts in source publication
Context 1
... we see that in the propagation stage, active centers are recognized, rules are applied, and most active centers that are removed are replaced by new ones. In the termination stage shown in Figure 9, rules act on the open and dangling flows and complete the function structure by carefully finding func- tions that connect between the remaining active centers. The result is the completed function structure shown at the end of Figure 9. ...
Similar publications
AGILMAT is a web application designed to help students learn Mathematics, with focus on high-school algebra and calculus drills.
A modular and extensible architecture and a wizard-based configuration interface decoupled from the system core are major
design features of AGILMAT. The drill expressions are specified by grammars and constraints imposed...
Combining goal-oriented and use case modeling has been proven to be an effective method in requirements elicitation and elaboration. To ensure the quality of such modeled artifacts, a detailed model analysis needs to be performed. However, current requirements engineering approaches generally lack reliable support for automated analysis of consiste...
Citations
... The function structure representation has also been used to develop various formalisms for functional decomposition, which is the problem domain of this paper. A generative graph grammar was developed in prior efforts, where the function models within the Design Repository were used to mine rules for model growth that were later applied heuristically to grow function structures from seed models [31]. More recently, a formal representation of functions that is consistent with the laws of thermodynamics was proposed [41], which supported physics-based reasoning during modeling [42]. ...
... This means that the knowledge representation is produced by abstracting previously observed patterns of events, designs, strategies, or world knowledgecollectively called heuristics -which historically produced acceptable outcomes in similar cases. An example of such an approach for the case of function model synthesis is [31], where generative rules for growing function models were mined by observing patterns in the models stored in the design repository and then formalized into a grammar. Such heuristics-based approaches need not, and usually do not, capture the scientific basis or explanation behind the rules. ...
Functional decomposition is an important task in early systems engineering and design, where the overall function of the system is resolved into the functions of its components or subassemblies. Conventionally, this task is performed manually, because of the possibility of multiple solution paths and the need for understanding the physics phenomena that could realize the desired effects. To this end, this paper presents a formal method for functional decomposition using physics-based qualitative reasoning. The formal representation includes three parts: (1) a natural language lexicon that can be used to detect the changes of physical states of material and energy flows, (2) a set of causation tables that abstracts the knowledge of qualitative physics by capturing the causal relations between the various quantities involved in a physical phenomenon or process, and (3) a process-to-subgraph mapping that translate the physical processes to function structure constructs. The algorithms use the above three representations and some additional topological reasoning to synthesize and assemble function structure graphs that are decompositions of a given black box model. The paper presents the formal representations and reasoning algorithms, and illustrates this method using an example function model of an air-heating device. It also presents the software implementation of the representations and the algorithms and uses it to validate the methods ability to generate multiple decompositions from a black box function model.
... Various approaches have been proposed to retrieve knowledge from data that provoke designers for creative and cross-domain design concept generation [9][10][15][16]. Other works explored generative approaches of function synthesis based on design repositories [17][18][19][20][21][22][23][24] or deep generative models based on product images or geometry data [25][26][27][28][29][30][31]. To date, few studies have leveraged the creative reasoning and prior knowledge from data simultaneously and transformed them into understandable design concepts. ...
... Function structures can be generated based on the knowledge and relations stored in the repository. The resulted function structure can either be represented in a function model [20][21][22] or a function-means tree [23][24]. This approach has a significant advantage of leveraging the functional knowledge and structural relations from the design data. ...
... However, the reported design repository utilized in this kind of research usually contains no more than a few hundred domain-focused products [20,21,22]. This is probably due to the amount of work involved in constructing such a function-based design repository. ...
Generating novel and useful concepts is essential during the early design stage to explore a large variety of design opportunities, which usually requires advanced design thinking ability and a wide range of knowledge from designers. Growing works on computer-aided tools have explored the retrieval of knowledge and heuristics from design data. However, they only provide stimuli to inspire designers from limited aspects. This study explores the recent advance of the natural language generation (NLG) technique in the artificial intelligence (AI) field to automate the early-stage design concept generation. Specifically, a novel approach utilizing the generative pre-trained transformer (GPT) is proposed to leverage the knowledge and reasoning from textual data and transform them into new concepts in understandable language. Three concept generation tasks are defined to leverage different knowledge and reasoning: domain knowledge synthesis, problem-driven synthesis, and analogy-driven synthesis. The experiments with both human and data-driven evaluation show good performance in generating novel and useful concepts.
... The grammar rules tackled this issue by supporting the creation of formal function structures through rules for combing blocks of functions. Still, their extensiveness is hard to comprehend, resulting in errors, such as the wrong definition of functions and flows on the function block level [13]. ...
... Thus, currently, many methods are being developed inside the DfAM, including ones based on function models of an AM product. However, the current function modelling approaches, while claiming to be general and domain-independent (e.g., [8,13]), are mostly built through analysis of consumer and electro-mechanical products. Consequently, they emphasise functions associated with flows of electrical energy and signals typical for electronic control. ...
... However, just following the rule does not guarantee the creation of a formal or consistent function model. Sridharan & Campbell [13] proposed a set of design grammar rules for the semi-automated creation of function structures. The rules are formed around action centres that suggest the rules for a given context. ...
Functional modelling is an essential part of systematic design approaches and is often prescribed in engineering design textbooks. However, function models created with current function modelling techniques often lack formal and repeatable representation, limiting their use in computational reasoning. Therefore, this paper presents a new functional modelling method to support function models' creation with formal and repeatable representation. The key element of the proposed method is a Function Class – a function-modelling element that categorises defined functions on a function block level by specifying operating flow, input and output flows, and integrates primary rules for functional modelling such as conservation law. The formalisation on a function block level reduces the number of morphological errors and provides a theoretical framework for future computational processing of function models. This paper proposes a protocol for developing Function Classes and defines a theoretical function modelling framework through Function Class Method. The development and use of the Function Class Method are demonstrated through the development of Function Classes for the Design for Additive Manufacturing domain as the first step toward a universal function modelling approach.
... Currently, data for nearly 200 products and 7,000 components are stored in this database, which are used to support various research efforts. Examples of such research include tools that use this large amount of legacy design data to extract patterns for reasoning such as conceptual design automation [26], concept generation [25], bio-inspired design [29,30], function model decomposition [31], and tools to model, predict, and mitigate failures and their propagation in complex systems [32][33][34][35]. Function structure models have also been used to measure the information content of design concepts [36,37], to study human designers' cognitive and behavior patterns during function modeling [38], and to study the mental metadata behind analogy formation [39]. ...
... The function structure representation has also been used to develop various formalisms for functional decomposition, which is the problem domain of this paper. A generative graph grammar was developed in prior efforts, where the function models within the Design Repository were used to mine rules for model growth that were later applied heuristically to grow function structures from seed models [31]. More recently, a formal representation of functions that is consistent with the laws of thermodynamics was proposed [41], which supported physics-based reasoning during modeling [42]. ...
Functional decomposition is an important task in early systems engineering and design, where the overall function of the system is resolved into the functions of its components or subassemblies. Conventionally, this task is performed manually, because of multiple possible solution paths and the need for understanding the physics phenomena that could realize the desired effects. This paper presents an approach of developing a formal method for functional decomposition using physics-based qualitative reasoning. The representation includes three parts: (1) a natural language reasoner that detects the changes of physical states of material and energy flows, (2) a set of causation tables that abstract the knowledge of qualitative physics by capturing the causal relations between the various quantities involved in a physical phenomenon or process, and (3) a process-to-subgraph mapping that translate the physical processes into function structure constructs. The algorithm uses the above three representations and some topological reasoning to assemble function models that represent the decomposition of a given black box model. The paper illustrates the potential of this method for functional decomposition using an example of an air-heating device. The paper also discusses the limitations and challenges in maturing this approach into an end-usable design tool.
... Various approaches have been proposed to retrieve knowledge from data that provoke designers for creative and cross-domain design concept generation [9][10][15][16]. Other works explored generative approaches of function synthesis based on design repositories [17][18][19][20][21][22][23][24] or deep generative models based on product images or geometry data [25][26][27][28][29][30][31]. To date, few studies have leveraged the creative reasoning and prior knowledge from data simultaneously and transformed them into understandable design concepts. ...
... Function structures can be generated based on the knowledge and relations stored in the repository. The resulted function structure can either be represented in a function model [20][21][22] or a function-means tree [23][24]. This approach has a significant advantage of leveraging the functional knowledge and structural relations from the design data. ...
... However, the reported design repository utilized in this kind of research usually contains no more than a few hundred domain-focused products [20,21,22]. This is probably due to the amount of work involved in constructing such a function-based design repository. ...
Generating novel and useful concepts is essential during the early design stage to explore a large variety of design opportunities, which usually requires advanced design thinking ability and a wide range of knowledge from designers. Growing works on computer-aided tools have explored the retrieval of knowledge and heuristics from design data. However, they only provide stimuli to inspire designers from limited aspects. This study explores the recent advance of the natural language generation (NLG) technique in the artificial intelligence (AI) field to automate the early-stage design concept generation. Specifically, a novel approach utilizing the generative pre-trained transformer (GPT) is proposed to leverage the knowledge and reasoning from textual data and transform them into new concepts in understandable language. Three concept generation tasks are defined to leverage different knowledge and reasoning: domain knowledge synthesis, problem-driven synthesis, and analogy-driven synthesis. The experiments with both human and data-driven evaluation show good performance in generating novel and useful concepts.
... Furthermore, it is hard to define the design space and to measure the value of each point of this area. After the design space has been defined, it can be searched and measured thanks to tools such as Quality Function Deployment, Axiomatic Design and Functional Analysis (Sridharan and Campbell, 2005). ...
Engineering design (ED) is the process of solving technical problems within requirements and constraints to create new artifacts. Data science (DS) is the inter-disciplinary field that uses computational systems to extract knowledge from structured and unstructured data. The synergies between these two fields have a long story and throughout the past decades, ED has increasingly benefited from an integration with DS.
We present a literature review at the intersection between ED and DS, identifying the tools, algorithms and data sources that show the most potential in contributing to ED, and identifying a set of challenges that future data scientists and designers should tackle, to maximize the potential of DS in supporting effective and efficient designs. A rigorous scoping review approach has been supported by Natural Language Processing techniques, in order to offer a review of research across two fuzzy-confining disciplines.
The paper identifies challenges related to the two fields of research and to their interfaces. The main gaps in the literature revolve around the adaptation of computational techniques to be applied in the peculiar context of design, the identification of data sources to boost design research and a proper featurization of this data. The challenges have been classified considering their impacts on ED phases and applicability of DS methods, giving a map for future research across the fields. The scoping review shows that to fully take advantage of DS tools there must be an increase in the collaboration between design practitioners and researchers in order to open new data driven opportunities.
... The function-modeling methodology established by design texts [1,2] and function researchers [35][36][37][38] have produced methods to decompose a functional black-box model into detailed function models. These models were unable of producing repeatable function models of a particular product, and the variability within function models produced by different modelers posed a question on the accuracy and consistency of the modeling language and representation [39][40][41]. To address this gap, efforts were made into formalizing the language for function modeling. ...
... Progressions on more AI-based approaches such as problem decomposition [14,55], analogical reasoning [19], qualitative and semantic reasoning [56], and causal analysis [35] will be possible as AI algorithms will be more compatible with a more expressive language [57]. Efforts were also made in developing a graph grammar for function modeling which was followed by a human study to demonstrate consistency in function modeling when the grammar was applied [40,41]. ...
Graph-based function models used in early-stage systems design usually represent only one operational mode of the system. Currently there is a need but no rigorous formalism to model multiple possible modes and states of a device in the same model and to perform model-based reasoning with that information such as predicting state transitions or causal propagations. This paper presents a formal representation of operational modes and states of technical devices based on automata theory for both discrete and continuous state transitions. It then presents formal definitions of three signal-processing verbs that actuate or regulate energy flows: Actuate_E, Regulate_E_Discrete, and Regulate_E_Continuous. The graphical templates, definitions, grammar rules, and application of each verb in modeling is illustrated. Finally, the representation is validated by implementing it on a graphical function modeling tool and using it to illustrate the verbs' modeling and reasoning ability for predicting mode and state transitions in response to control signals and cause-and-effect propagation throughout system-level models.
... The function-modeling methodology established by design texts [1,2] and function researchers [35][36][37][38] have produced methods to decompose a functional black-box model into detailed function models. These models were unable of producing repeatable function models of a particular product, and the variability within function models produced by different modelers posed a question on the accuracy and consistency of the modeling language and representation [39][40][41]. To address this gap, efforts were made into formalizing the language for function modeling. ...
... Progressions on more AI-based approaches such as problem decomposition [14,55], analogical reasoning [19], qualitative and semantic reasoning [56], and causal analysis [35] will be possible as AI algorithms will be more compatible with a more expressive language [57]. Efforts were also made in developing a graph grammar for function modeling which was followed by a human study to demonstrate consistency in function modeling when the grammar was applied [40,41]. ...
Modern design problems often require multi-modal, reconfigurable solutions. Function modeling is a common tool used to explore solutions in early mechanical design. Currently, function modeling formalisms minimally support the modeling of multi-modal systems in a formal manner. There is a need in function modeling to capture multi-modal system and analyze the effects of control signals and status signals on their operating modes. This paper presents the concept of functional conjugacy, where two function verbs or functional subgraphs are topological opposites of each other. The paper presents a formal representation of these conjugate verbs that formally captures the transition from one mode of operation to its topological opposite based on the existence of, or the value of, signal flows. Additionally, this paper extends functional conjugacy to functional features, which supports conjugacy-based reasoning at a higher level of abstraction. Through the example of a system-level function model of a geothermal heat pump operating in its heating and cooling modes, this paper demonstrates the ability to support modal reasoning on function models using functional conjugacy and illustrates the modeling efficacy of the extended representation.
... This work focuses on the function and not the form or geometry of a design. This work does not, as yet, include in its scope approaches for the automatic synthesis of designs (e.g. using a search for design prototypes as in Gero(1990), Umeda, et al. (1990) or using grammar based approaches such as those described by Stiny and Gips (1971), Schmidt and Cagan (1995), Sridharan and Campbell (2005), etc.). The above limitations could be the subject matter for future work. ...
A symbolic logical framework (L) consisting of first order logic augmented with a causal calculus has been provided to formalize, axiomatize and integrate theories of design. L is used to represent designs in the Function-Behavior-Structure (FBS) ontology in a single, widely applicable language that enables the following: seamless integration of representations of function, behavior and structure; and generality in the formalization of theories of design. FRs, constraints, structure and behavior are represented as sentences in L. FRs are represented (as abstractions of behavior) in the form of existentially quantified sentences, the instantiation of whose individual variables yields the representation of behavior. This enables the logical implication of FRs by behavior, without recourse to apriori criteria for satisfaction of FRs by behavior. Functional decomposition is represented to enable lower level FRs to logically imply the satisfaction of higher level FRs. The theory of whether and how structure and behavior satisfy FRs and constraints is represented as a formal proof in L. Important general attributes of designs such as solution-neutrality of FRs, probability of satisfaction of requirements and constraints (calculated in a Bayesian framework using Monte Carlo simulation), extent and nature of coupling, etc. have been defined in terms of the representation of a design in L. The entropy of a design is defined in terms of the above attributes of a design, based on which a general theory of what constitutes a good design has been formalized to include the desirability of solution-neutrality of (especially higher level) FRs, high probability of satisfaction of requirements and constraints, wide specifications, low variability and bias, use of fewer attributes to specify the design, less coupling (especially circular coupling at higher levels of FRs), parametrization, standardization, etc..
... In order to accomplish this, we first automate the process of finding the functions and flows that a product's constituent components usually perform. After finding the most common functions and flows, we can order them based on simple guidelines known as grammar rules, which dictate the order of functions and flows within a component [8] [9] [10]. If our algorithm found the most common functions and flows of an electric cord component to be export electrical energy, transfer electrical energy, and import electrical energy, an example of grammar rules would be that import must come before export and transfer must come before export for the same flow. ...
... In this work, we create an algorithm that mines data from a de-sign repository that stores component-centric design information about products. This multi-decade project is housed at Oregon State University and it is simply called The Design Repository 1 [10] [12] [13] [14] [15] . ...
The purpose of this research is to find the optimum values for threshold variables used in a data mining and prediction algorithm. We also minimize and stratify a training set to find the optimum size based on how well it represents the whole dataset. Our specific focus is automating functional models, but the method can be applied to any dataset with a similar structure. We iterate through different values for two of the threshold variables in this process and cross-validate to calculate the average accuracy and find the optimum values for each variable. We optimize the training set by reducing the size by 78% and stratifying the data, whereby we achieve an accuracy that is 96% as good as the whole training set and takes 50% less time. These optimum values can be used to better predict the functions and flows of any future product based on its constituent components, which can be used to generate a complete functional model.
