Search

Automated concept generation is non-trivial task. The complexity of this problem is mainly due to lack of formal representation frameworks that lend themselves easily to a computational approach. Generative grammar has emerged as a potential solution to this problem and presents a number of different possibilities for conceptual design automation. This paper presents a novel search method that has been developed specifically for search trees defined by a special class of generative grammar in which rules of the grammar have parameters associated with them. A novel feature of the proposed search is ‘Human in the loop’ approach in which learning about the search space is achieved by querying the user. The user fatigue restricts the maximum number of comparisons of candidate solutions (30–50). From the data gathered from the comparisons, a stochastic decision making process proposed in this paper quickly converges to a region of design space which best meet the user’s preference. The method is implemented and applied to a grammar for shampoo bottle concept generation. It is shown through multiple user-guided and automated experiments that the method has ability to learn and adopt through human computer interaction process. The implications of the proposed search method for automated conceptual design are expounded on in the conclusions.Copyright © 2011 by ASME

This paper describes a design automation approach that combines various optimization research and artificial intelligence methods for synthesizing fluid networks. Unlike traditional software tools available today, this approach does not rely on having any predefined network topology to design and optimize its networks. PipeSynth generates its designs by using only desired port locations, and the desired fluid properties at each of those ports. An ideal network is found by optimizing the number and connectivity of pipes and pipe fittings, the size and length of each pipe, and the size and orientation of each fitting. A Uniform-Cost-Search is used for topology optimization along with a combination of non-gradient based optimization methods for parametric optimization. PipeSynth demonstrates how advances in automated design can enable engineers to manage much more complex fluid network problems. PipeSynth uses a unique representation of fluid networks that synthesizes and optimizes networks one pipe at a time, in three-dimensional space. PipeSynth has successfully solved several problems containing multiple interlaced networks concurrently with multiple inputs and outputs. PipeSynth shows the power of automated design and optimization in producing solutions more effectively and efficiently than traditional design approaches.Copyright © 2011 by ASME

This paper presents a new search method that has been developed specifically for search trees defined by a generative grammar. Generative grammars are useful in design as a way to encapsulate the design decisions that lead to candidate solutions. Since the candidate solutions are not confined to a single configuration or topology and thus useful in conceptual design, they may be difficult to computationally analyze. Analysis is achieved in this method by querying the user. The user interaction is kept to a maximum of thirty pair-wise comparisons of candidates. From the data gathered from the comparisons, a stochastic decision making process infers what candidate solutions best meet the user’s preference. The method is implemented and applied to a grammar for tying neckties. It is shown through 21 user experiments and 4000 automated experiments that the method consistently finds solutions within the 99.8 percentile. The implications of this method for conceptual design are expounded on in the conclusions.Copyright © 2009 by ASME

A method for automating the design of simple and compound gear trains using graph grammars is described. The resulting computational tool removes the tedium for engineering designers searching through the immense number of possible gear choices and combinations by hand. The variables that are automatically optimized by the computational tool include the gear dimensions as well as the location of the gears in space. The gear trains are optimized using a three-step process. The first step is a tree-search based on a language of gear rules which represent all possible gear train configurations. The second step optimizes the discrete values such as number of teeth through an exhaustive search of a gear catalog. The final step is a gradient-based algorithm which optimizes the non-discrete variables such as angles and lengths in the positioning of the gears. The advantage of this method is that the graph grammar allows all possible simple and compound gear trains to be included in the search space while the method of optimization ensures the optimal candidate for a given problem is chosen with the tuned parameters.Copyright © 2009 by ASME

This paper describes an approach to automate the design for sheet metal parts that are not only novel and manufacturable but also satisfies multiple objective functions such as material cost and manufacturability. Unlike commercial software tools such as Pro/SHEETMETAL which aids the user in finalizing and determining the sequence of manufacturing operations for a specified component, our approach starts with spatial constraints in order to create the component geometries and helps the designer design. While there is an enormous set of parts that can feasibly be generated with sheet metal, it is difficult to define this space systematically. To solve this problem, we currently have 88 design rules that have been developed for four basic sheet metal operations: slitting, notching, shearing, and bending. A recipe of the operations for a final optimal design is then presented to the manufacturing engineers thus saving them time and cost. The technique revealed in this paper represents candidate solutions as a graph of nodes and arcs where each node is a rectangular patch of sheet metal, and modifications are progressively made to the sheet to maintain the parts manufacturability. They are presented in the form of Standard Tessellation Language files (.stl) that can be transferred into available modeling software for further analysis. The overall purpose of this research is to provide creative designs to the designer granting him/her a new perspective and to check all the solutions for manufacturability in the early stage of design process. An example sheet metal design problem is shown in this paper with some of the preliminary designs that our approach created.Copyright © 2008 by ASME

Neural networks are increasingly becoming a useful and popular choice for process modeling. The success of neural networks in effectively modeling a certain problem depends on the topology of the neural network. Generating topologies manually relies on previous neural network experience and is tedious and difficult. Hence there is a rising need for a method that generates neural network topologies for different problems automatically. Current methods such as growing, pruning and using genetic algorithms for this task are very complicated and do not explore all the possible topologies. This paper presents a novel method of automatically generating neural networks using a graph grammar. The approach involves representing the neural network as a graph and defining graph transformation rules to generate the topologies. The approach is simple, efficient and has the ability to create topologies of varying complexity. Two example problems are presented to demonstrate the power of our approach.

The design of any MEMS component is subject to stringent manufacturing constraints. The knowledge about these constraints seems to be available to designers who have experienced the details of MEMS fabrication. In this paper, we put forth the idea of automatically generating a fabrication sequence for surface micro-machined MEMS components using the knowledge stored in grammar rules. As an analogy to CAD tools used in mechanical systems, we envision creating a tool which has the Pro-Engineer approach of determining fabrication sequences for a machine tool based on the final part shape. This tool could be an integral addition to the current MEMS design software so that the designers can freely draft devices and then allow an automated process to determine the fabrication sequence. In this paper we give a brief introduction about the graph grammars. Data from already designed MEMS components is extracted in the form of rules to create an expert system. We have also included an example of generation of the fabrication sequence for several MEMS components.Copyright © 2007 by ASME

Component selection in engineering design is a process in which an assembly of pre-defined component types is given and a choice of specific components is desired that satisfies a set of design requirements and constraints. Although algorithmic approaches to component selection have been researched for specific mechanical engineering problems such as bearing selection, a generalized technique has not been developed. This paper proposes a universal tool to automate the process of component selection by incorporating a tree search. Our technique evaluates the worth of candidate solutions in terms of the customer needs satisfaction and the compatibility between interconnected components. The tree search technique used in this research is not only quick and efficient but also guarantees an optimal solution.Copyright © 2007 by ASME

Sequential sampling refers to a set of design of experiment (DOE) methods where the next sample point is determined by information from previous experiments. This paper introduces a qualitative and quantitative sequential sampling (Q2S2) technique, in which optimization and user knowledge is used to guide the efficient choice of sample points. This method combines information from multiple fidelity sources including computer simulation models of the product, first principals involved in design, and designer’s qualitative intuitions about the design. Both quantitative and qualitative information from different sources are merged together to arrive at a new sampling strategy. This is accomplished by introducing the concept of a confidence function, C, which is represented as a field that is a function of the decision variables, x, and the performance parameter, f. We compare the sampling plans generated by Q2S2 to previously known sample plans on five test functions using various metrics. In each case, the performance of Q2S2 is highly encouraging.

The challenge of determining the best design in a multimodal design space with multiple local optimal solutions often challenges the best available optimization techniques. By casting the objective function of the optimization problem in the form of a Non-Uniform Rational B-spline (NURBs) metamodel, known as a HyPerModel, significant optimization advantages can be achieved, including the ability to efficiently find the global metamodel optimum solution with less computational expense than traditional approaches. This optimization strategy, defined by the HyPerOp algorithm, uses the underlying structure of a HyPerModel to intelligently select starting points for optimization runs and to identify regions of the design space that do not contain locations for the global metamodel optimum location. This paper describes the application of the HyPerOp algorithm to mixed integer programming problems and demonstrates its use with two example applications. The algorithm works with design spaces composed of continuous and integer design variables and provides a complementary approach for improved optimization capabilities.

This paper introduces research leading to a computer-aided design tool in which engineering designers can test various concepts in an environment equipped to automatically model the dynamics and then optimize the specified components to best meet the specifications of the design problem. The input to the system is a graph of components where the components' design variables are to be determined by a subsequent optimization process. Design goals are often as clearly defined as possible prior to any design effort, which suggests that the objective function can be automatically generated if a system dynamics model is available and the transformation from the system model to objective function can be automated. In this research, automated objective function derivation is demonstrated through automated bond graph modeling and model transformation according to the goals defined by designers. A component (or sub-system) repository is being developed to store not only the component dynamics models, but also various information including typical component design constraints and physical constitutive laws. The paper discusses a systematic approach to automatically prepare a mechatronic design problem for optimization, which can decode and encode proper genotypes of intended design variables, which account for the existing of design constraints and physical constitutive laws. An example of a weighing machine design is used to showcase the approach.

This paper introduces a method for sequentially determining experiments in a “design of experiments” where optimization and user knowledge are used to guide the efficient choice of sample points. Typical approaches to the design of experiments involves determining the sample points all at once prior to any experimentation, or sequentially based on the results of previous sample points. This method combines information from multiple fidelity sources including actual physical experiment, computer simulation models of the product, first principals involved in design and designer’s qualitative intuitions about the design. Both quantitative and qualitative information from different sources are merged together to arrive at new sampling strategy. This is accomplished by introducing the concept of confidence, C, which is represented as a field that is a function of the decision variables, x, and the performance parameter, f. The advantages of the approach are demonstrated using different example cases.Copyright © 2006 by ASME

Over the last eight years, the Machine Design courses at the United States Air Force Academy and at the University of Texas, Austin have evolved through the development, implementation and assessment of extensive active learning methods. In particular, the courses have evolved to include extensive hands-on projects that are integrated throughout the course as well as a significant multimedia component. The hands-on educational innovations, which promote experiential investigation using devices such as remote controlled cars, Lego RoboLab, and reverse engineering of consumer products, have received very positive assessment. The multimedia content, which includes extensive foundational content on Mechanics of Materials as well as a separate multimedia experience for learning about planetary gear systems, has also been assessed and received very affirmative feedback. The assessment of these active learning educational innovations has been multifaceted. Quantitative components of the assessment have included student end-of-course critiques, homework, specific exam questions and survey data. Qualitative assessment has been achieved through focus groups as well as both written and verbal feedback from students and professors using the active learning aids. Although the majority of the assessment has been positive, we have also received important constructive criticism during the development of these educational enhancements. The “iterative” development of these active learning techniques has involved responding to these criticisms and reassessing the program’s effectiveness. In this paper, we first provide an overview of the previous work done in this area, then move on to show new developments and related assessment. In particular, new assessment, which is correlated with Myers Briggs personality types, is reported, showing results of the current integrated use of active learning techniques, including hands-on and multimedia experiences. In this light, the current paper should work as a roadmap for others who desire to integrate active learning into their courses, whether they are courses in Machine Design or not.Copyright © 2005 by ASME

This paper presents an assembly model process that fully characterizes the structural and flow interactions between artifacts in a product. Reverse engineering techniques were employed during the analysis of thirty-three existing consumer products to arrive at a concise standardization of the modeling process. During the product investigation, four different types of structural interactions were identified. These structural interactions, couple, secure, position and guide, were defined using a standardized vocabulary of functional terms. These four structural interactions are rigorously described in this paper in an effort to outline an assembly model method that is accurate and repeatable. Additionally, flow interactions between components are also characterized within the presented modeling technique. A rough representation of the artifact configuration of a product can also be achieved through placement of the component structures in the model. Analysis of the consumer product set also revealed that a new design tool can be generated using the structural interaction information contained in the described assembly models.Copyright © 2005 by ASME

This paper introduces research leading to a computer-aided design tool in which engineering designers can test various concepts in an environment equipped to automatically model the dynamics and then optimize the chosen components to best meet the specifications of the design problem. The input to the system is a graph of components where the components either include defined parameter values or are left as variables for the subsequent optimization. In this paper, “conceptual dynamics” is introduced with the goal of revealing the qualitative information of the system states, energy flow patterns, possible operation modes, degrees of freedom, etc. The dynamics of a system are captured in a multi-domain representation referred to as a conceptual dynamics graph (CD-Graph), which is composed of components and virtual couplers. The components feature both detailed multi-complexity models (bond graphs) and qualitative dynamics information (Conceptual Dynamics Matrices), and the virtual couplers include information on the interconnections of the eight domains (three translational, three rotational, electrical, and hydraulic). The Conceptual Dynamics Graph provides a platform for various design automation approaches and further dynamics performance optimization. This paper discusses an automatic approach that extracts the partial configuration of a complete design structure that contributes to the system dynamics. This allows designers to better understand the system and focus on the functional aspects that meet certain dynamic specifications. Finally an automatic model generation approach of CD-Graph is briefly illustrated.Copyright © 2005 by ASME

Few computational tools exist to assist designers during the conceptual phase of design, and design success is often heavily weighted on personal experience and innate ability. Many well-known methods (e.g. brainstorming, intrinsic and extrinsic searches, and morphological analysis) are designed to stimulate a designer’s creativity, but ultimately still rely heavily on individual bias and experience. Under the premise that quality designs comes from experienced designers, experience in the form of design knowledge is extracted from existing products and stored for reuse in a web-based repository. This paper presents an automated concept generation tool that utilizes the repository of existing design knowledge to generate and evaluate conceptual design variants. This tool is intended to augment traditional conceptual design phase activities and produce numerous feasible concepts early in the design process.Copyright © 2005 by ASME

In an ideal design process, designers envision a configuration of components prior to determining dimensions or sizes of these components. Given the breadth of suppliers and production methods that exist today, most engineered artifacts are a mix of both custom-made parts and OEM (original equipment manufacturer) parts. The design of any future artifact must be carefully planned to take advantage of the diverse set of possibilities. We conjecture that computational design tools could be developed to help designers navigate the design space in creating configurations from detailed specifications of function. In this research, a methodology is developed that extracts design knowledge from an expanding online library of components in the form of grammar rules. From an initial implementation of forty-five rules compiled from 15 components extracted from three products, we demonstrate a computational process that builds a new design configuration by borrowing concepts from how common functions are solved in related designs.

Multistable equilibrium (MSE) systems are a type of adaptable system that can have multiple mechanical configurations requiring no power to maintain the stable configurations. Thus, power is only needed to move among the stable states, and each stable configuration represents a level of adaptability. Since stable equilibrium configurations can be defined by potential energy minima, we base the design of MSE systems on shaping the potential energy curve at desired equilibrium configurations. This view allows one to construct a performance space defined by how well candidate systems meet a desired potential energy curve. By using a Monte Carlo mapping to link the performance space to the design space in tandem with stochastic optimization methods, the designer determines whether or not a certain system topology can be designed as a MSE system. Qualitative and quantitative mapping procedures enable the designer to decide whether or not the desired design lies near the center or periphery of a performance space. This dictates how the optimization is to be executed which in turn informs the designer as to whether or not a feasible limit in the system performance has indeed been reached.Copyright © 2004 by ASME

Collecting data to characterize an unknown space presents a series of challenges. Where in the space should data be collected? What regions are more valuable than others to sample? When have sufficient samples been acquired to characterize the space with some level of confidence? Sequential sampling techniques offer an approach to answering these questions by intelligently sampling an unknown space. Sampling decisions are made with criteria intended to preferentially search the space for desirable features. However, N-dimensional applications need efficient and effective criteria. This paper discusses the evolution of several such criteria based on an understanding of the behaviors of existing criteria, and desired criteria properties. The resulting criteria are evaluated with a variety of planar functions, and preliminary results for higher dimensional applications are also presented. In addition, a set of convergence criteria, intended to evaluate the effectiveness of further sampling are implemented. Using these sampling criteria, an effective metamodel representation of the unknown space can be generated at reasonable sampling costs. Furthermore, the use of convergence criteria allows conclusions to be drawn about the level of confidence in the metamodel, and forms the basis for evaluating the adequacy of the original sampling budget.Copyright © 2004 by ASME

The popularity of sheet metal in modern engineering artifacts is due to the fact that it is both inexpensive as a raw material and inexpensive to form into components. In comparison to forging or machining components, sheet metal can produce lightweight and inexpensive design solutions. The main shortcomings of sheet metal are that resulting components have a limited rigidity and the feasibility of the parts is constrained by the inherent two-dimensionality of the initial sheet. Furthermore, design and manufacturing engineers are challenged by finding a shape that satisfies all spatial constraints and by deciding the optimal sequence of operations for making this product that minimizes both time and associated manufacturing costs. In the past two years, we have been working towards an automated tool that creates candidate sheet metal topologies and optimizes them for spatial constraints as well as time and cost objectives. While we have yet to complete this goal, we have to date developed a representation capable of creating a wide variety of sheet metal topologies (see DETC2002/DAC-34087) and have recently created a thorough evaluation method which is presented in this paper along with some preliminary results.Copyright © 2004 by ASME

Metamodels approximate complex multivariate data sets from simulations and experiments. These data sets often are not based on an explicitly defined function. The resulting metamodel represents a complex system’s behavior for subsequent analysis or optimization. Often an exhaustive data search to obtain the data for the metamodel is impossible, so an intelligent sampling strategy is necessary. While multiple approaches have been advocated, the majority of these approaches were developed in support of a particular class of metamodel, known as a Kriging. A more generic, commonsense approach to this problem allows sequential sampling techniques to be applied to other types of metamodels. This research compares recent search techniques for Kriging metamodels with a generic, multi-criteria approach combined with a new type of B-spline metamodel. This Bspline metamodel is competitive with prior results obtained with a Kriging metamodel. Furthermore, the results of this research highlight several important features necessary for these techniques to be extended to more complex domains.

This work is a part of a larger research project to understand the challenges in creating MultiStable Equilibrium (MSE) devices. MSE devices are those that have more than one stable position or configuration which can be maintained with no power input. The study of potential energy minima in magnetic systems can be used to create novel and efficient MSE mechanisms. This research focuses on using the magnetic energy in space as the main criteria for analysis of MSE devices. Permanent magnets are used to create a 2D magnetostatic field. The magnetic energy density in air is then plotted along a path through the field. For a piece of iron following a specified path, the stable equilibrium positions correspond to the locations of maximum air energy density. Furthermore, the stiffness at each of the equilibrium positions can be correlated to a pseudo-stiffness of the energy density. This information can help one design systems for multiple stable positions with minimal computational time and effort without simplifying the problem geometry.Copyright © 2003 by ASME

Product architecture is the transformation of function to layout. Like much of conceptual design, it is a highly dynamic process whereby engineers must consider a deluge of information in terms of both function and form. One shortcoming of current engineering practice is the absence of representations or abstractions used to aid in developing, refining, and exploiting alternative layout solutions. The purpose of this paper is to present a representation for product architecture that sufficiently captures the design factors relevant to product architecture design which are not taken into account in current practices. An example is given to illustrate the technique, and results of a validation experiment are shown.

This paper involves the development, implementation and testing of a geometric representation scheme for building feasible sheet metal components. The approach taken here is based on prior shape grammar methods for engineering design. A series of seventeen grammar rules have been developed to represent a variety of cutting and bending operations that can be applied to sheet metal to construct feasible shapes. The implemented system has benefits as both a user interaction tool or as the basis for a computational design synthesis approach for designing sheet metal components. An example of a real component is shown as well as the method for invoking the sheet metal grammar to create this component.Copyright © 2002 by ASME

This paper presents some key extensions to a representation for electromechanical components that is sufficiently detailed enough to provide for the automated construction of new design configurations. The extensions, shown here, build upon a representation that is both formal and implemented, and can handle the interactions between components even if such interactions represent only partial configurations. The results of these new additions elucidate the power and flexibility of this functional representation for electromechanical components. The method can be used as an interactive tool to allow an engineering designer to explore new design concepts or can be combined with additional software tools to achieve a computational design synthesis approach to topological design problems. The inclusion of function sharing, component layout, and qualitative dynamics achieve a closer approach to how the human mind represents components.Copyright © 2002 by ASME