University at Buffalo: Mechanical and Aerospace Engineering

Mechanical and Aerospace Engineering Home

Apply

Research

Research Centers

Design and Optimization

Faculty

Affiliated Faculty

English, K.
Kasprzak, E.
Olewnik, A.

Laboratories

Research summaries

DEVELOPMENT OF MULTI-OBJECTIVE CONCURRENT SUBSPACE OPTIMIZATION AND VISUALIZATION METHODS FOR MULTIDISCIPLINARY DESIGN -- Most real- world design problems are complex and multidisciplinary, as there is always more than one objective (cost) function to be extremized in each problem. Hence, the primary goal of this research is to extend the capability of the Concurrent Subspace Optimization (CSSO) method to handle multi- objective optimization problems (MOPs) in Multidisciplinary Design Optimization (MDO). In this study, the framework of Multi-Objective Concurrent Subspace Optimization (MOCSSO) methods have been developed in which each discipline has substantial control over its own objective function during the design process. By means of MOCSSO, an aggregate objective function is not required before solving an MOP. Instead, each subspace optimizes its own objective function while obtaining a Pareto optimum. This is accomplished by incorporating the concept of Pareto optimality into the subspace optimizations. Another important consideration for large-scale multidisciplinary multi-objective problems pertains to the increasing large amounts of data being generated during the design process. As computing power has advanced, design problems that can be modeled and analyzed have grown in numerical size, dimensionality, and complexity. As a result, engineers often face large and complex datasets. Traditionally, design information has been passed as numerous pages of printed data or in spreadsheet form. It is not always an easy task or even a practical one to analyze such large amounts of data. Therefore, it has become more crucial to develop new methods and tools to understand and interpret complex datasets. Visualization has proven to be a valuable tool in providing the needed increased understanding. In this study, the Multi-Objective Pareto Front Visualization (MOPFV) interface, tailored for the Multi-Objective Concurrent Subspace Optimization methods, has been developed to enable such a need as mentioned above. -- C. L. BLOEBAUM, C.H. Huang.
DEVELOPMENT OF A WEB-BASED VISUALIZATION ENVIRONMENT FOR DECISION-MAKING IN MULTIDISCIPLINARY DESIGN OPTIMIZATION -- The analysis and optimization of most engineering and manufacturing systems such as automobiles, aircrafts etc. experience interactions amongst various components of the full system. These interactions are increasingly being handled using distributed design groups. Multidisciplinary Design Optimization (MDO) has evolved to assist designers from different disciplines in steering a product or process design towards its optimum. Visualization is increasingly being used in MDO to improve the product design by assisting the designers in their decision-making. A barrier to sharing visual models amongst designers from different disciplines is the use of different computing platforms to accomplish their tasks. Since the design groups are often separated geographically, a web-based design environment that allows collaboration amongst designers is preferred. This research focuses on the development of a platform independent visualization framework to assist designer in their decision-making while solving MDO problems that could potentially have a physical representation. A physical representation of the design artifact, along with the Dependency Structure Matrix (DSM) is provided. Using this interface, a designer can view components or sub-systems being impacted by changes in other components or sub-systems by means of visual cues. The development of an intuitive interface would allow a designer to trade-off accuracy and efficiency in an instinctive fashion. In addition, using the representation of the product being designed, the designer can better comprehend the impact that changes in a component will have on the overall physical system. A critical portion of this capability is the nonproprietary, platform-independent translation of various design related data into a visual representation. This transformation process provides a capability for collaboration amongst companies or design groups, each using their own internal design tools. Various data associated with the MDO system such as analysis data, geometric data, can be integrated into the representation and handled in an efficient way to be used for collaboration between d esigners. -- C. L. BLOEBAUM, G. Agrawal.
DEVELOPMENT OF AN INTUITIVE MULTI-VARIATE MULTI-DIMENSIONAL VISUALIZATION FRAMEWORK -- A visualization methodology is developed in which multidimensional data can be viewed in an intuitive and straightforward manner. Based on this visualization, it is possible to quickly identify ?￥good?| regions of the design space for an optimization application, regardless of space complexity. The method requires no dimension fixing. While the potential applications of such an approach are many, in this research we are presently focused on the application to multiobjective optimization, in order to demonstrate it?|s usefulness in visualizing n-f space (i.e. for more than three objective functions in a multiobjective optimization problem). The visualization is shown to aid in the final decision of what potential design should be chosen amongst all possible Pareto solutions. While the application here is focused on multiobjective optimization, the research has broader implications of the method, as well as broader potential applications. The method is equally useful and applicable for exploration (as is usually desired in design applications) as well as aiding in interpretation of relationships in existing databases (as in finding correlations amongst data in an economic database or bioinformatics database). -- C. L. BLOEBAUM, G. Agrawal.
DECISION SUPPORT TOOL FOR COLLABORATIVE MULTIDISCIPLINARY DESIGN AND OPTIMIZATION USING MULTIPLE SYSTEM DECOMPOSITION DOMAIN -- Design of complex engineering systems often involves decomposition of the complete system into smaller and manageable subsystems or elements. The system can be decomposed either based on physical components, or based on disciplinary analysis or based on the tasks involved in the design. The decomposed subsystems or elements (tasks, analysis & components) are still interdependent amongst themselves. The decomposition and interdependencies within each domain is represented by Design Structure Matrix (DSM). This research effort is intended to understand and analyze the product development process effectively by capturing the correlation between the three decomposition domains, namely the physical domain, analysis domain and task domain. This research will help the designers and managers to capture the effects of changes in one of the domain, onto the other domains of the system. Based on this research, a decision support tool will be built for multidisciplinary design and optimization (MDO) problems. The decision support tool could be used to monitor the design process, and also to estimate the time and cost involved with a change in the design process. The decision support tool will be a web-based platform independent software tool; primarily using XML data transfer standards. This research effort has also focused on development of an XML-based generic data architecture that could be applied to MDO problems by combining optimization data and analysis data in a platform independent format. This XML data format can be used as a foundation for the development of distributed MDO tools by governing the data exchange between disciplinary analysis tools, optimizers and user interfaces. The Golinski?|s speed reducer optimization problem from NASA MDO Test Suite is modified as a tightly coupled MDO problem to be used as a test problem for this research. The methodology and decision support tool will also be demonstrated on other MDO test problems. -- C. L. BLOEBAUM,S. S. Parashar.
DEVELOPMENT OF A PARETO FRONTIER USING MOPCSSO AND APPLYING UTILITY THEORY TO CAPTURE PREFERENCES -- Designing complex systems requires information from many disparate disciplines. While these disciplines all must achieve an overall goal of designing a product, they each have their own objectives (cost functions) that they wish to satisfy. The research area of Multidisciplinary Design Optimization (MDO) deals with the conflicting nature of the different disciplines. Yet, with advances and changes arising in the design process, new methodologies must be implemented to arrive at the optimum design for all subsystems. It is known that while generating a Pareto frontier yields all optimum solutions, it is then up to the designer(s) to chose the final solution. This choosing relies on designer preferences, which is ambiguous. The first goal of this research is to verify that it is in fact possible to generate a well distributed Pareto frontier by implementing concepts of the recently developed Multi-Objective Pareto Concurrent Subspace Optimization (MOPCSSO) method in conjunction with the ?? - constraint method. Generating a smooth Pareto frontier will assist in satisfying the overall goal of this research. The overall goal is to capture designer preferences on the Pareto frontier implementing concepts from utility theory. -- C. L. BLOEBAUM, J. Galuski.
MULTI-HAZARD DESIGN AND DECISION SUPPORT -- Robust computational approaches are under development for multi-hazard design and decision support based upon evolutionary methodologies. These new approaches explicitly account for environmental uncertainty and incorporate both engineering and sociotechnical aspects of the problem. One research thread of the current work focuses on the evolutionary seismic design and retrofit of structures using passive energy dissipation systems. A second innovative research thrust is directed toward the development of a general computational methodology for organizational modeling and complex decision processes. The initial application of this approach addresses seismic retrofit decisions in hospitals and other critical healthcare organizations. Collaborators include D.J. Alesch, W.J. Petak, M.L. Green, Y. Hu, O. Lavan and S. Dogruel. Funding has been provided by MCEER, NSF and NYS. -- G.F. DARGUSH
PHYSICALLY-BASED MODELING THROUGH A DYNAMIC ATOMIC UNIT APPROACH FOR HAPTIC RENDERING: TOWARDS NON-LINEAR, ISCOELASTIC, ANISOTROPIC BEHAVIOR -- The intention of this research is to design a methodology which allows 3D models to be enabled with haptic rendering capability. To provide this capability, the 3D model needs to be voxelized and the mechanical behavior of the body to be simulated, specified. Using the dynamic Atomic Unit Approach, the model can then be hapitcally rendered. The user can build complex hapitcally enabled objects such as a hard body embedded within a soft one, by performing boolean operations of separate hapitcally enabled objects. -- Amrita Chanda, T.KESAVADAS.
UB VIRTUAL SITE MUSEUM -- The purpose of this project is to document and publish a uniquely comprehensive report on ancient assyrian palace, by bringing together, for the first time, all new, updated and extant materials and research using leading-edge computer graphics technologies, and to create a multimedia interactive educational and scholarly research tool with applications for colleges, universities, independent scholars, and museums. The main advantages are: Preservation of Cultural Heritages Reconstruction of Precise and Authoritative Ancient Site Building Efficient Education/Research Tool. -- T.KESAVADAS, S.Paley, Y.Kim
POST DISASTER MONITORING -- The goal of this project are: - Monitoring Earthquake or Post-disaster Situation with GIS/GPS - Having Multiple Choice of Optimal Path to Nearest Hospitals - Implementation of Large Amount of Data and its Fusion The purpose of this project is to simulate monitoring of disaster and its relief situation for a control center. When disaster occurs, there normally is huge amount of data flowing into the control center and thus data fusion process is needed for monitoring and relief action, such as data visualization, status awareness and decision making. Main technique includes geo-referencing GIS data, real-time optimal path simulation and multiple optimal path direction in immersive virtual environment. -- James Llinas, T.KESAVADAS, Y.Kim , et. al
HAPTIC BASED MULTI-PURPOSE GRINDING SIMULATOR -- Material removal operations such as polishing and grinding appear in a variety of fields. This projects aims at creating a haptic based simulator for these operations. Polishing and grinding operations are used in a variety of fields ranging from medical applications like dentistry and orthopedics, to machining operations on metals. These operations are labor-intensive and usually require high level of skill. It is not always viable or possible to train new operators directly on a real specimen for a variety of reasons. In the case of dentists for example, training would depend on the availability of a volunteer or on physical prototypes that are replicated using various artificial materials. Artificially created specimens that are most commonly used to train students are rarely able to simulate the feel of a real specimen that is often critical to the application. The recent onset of sophisticated haptic devices, capable of delivering precise and controllable forces has opened new avenues for the use of computer simulations to replicate these complicated tasks. In our research we aim to combine these new haptic technologies and ever growing computational power to produce training tools that are accurate, realistic and most importantly affordable. A terrain modeling method called Heightmaps is used to replicate the surface roughness. In this method a digital image of the work piece is converted into a 3-D heightmap. Analyzing each pixel of the image relative to its surroundings and assigning it with a height leads to the creation of the heightmap that is imported into the haptic model. The surface roughness can now be felt using the PHANToM haptic device. Some screenshots of the process are shown below. A real-time analysis of the input forces, validated by the force model is used to modify the texture and enhance realism. This gives the user a sense of material removal. The rate of removal is governed by a variety of factors like input force and tool attributes like speed, depth of cut, etc. -- T.KESAVADAS, A.Balijepalli.
VIRTUAL FACTORY PROJECT -- The UB VR-Fact provides an manufacturing environment for:
- Modeling before building
- Simulating before you producing
- Anticipating and solving problems before they occur
VR-Fact! is an interactive virtual factory development software developed as a part of a research to explore the applications of Virtual Environments (VE) in the area of manufacturing automation. VR-Fact creates digital mock-up of real factory shopfloor. Designers can drag and place modular machines in the factory to study issues such as plant layout, cluster formation and part flow analysis. For the design of this shop floor the user can interactively choose the machines which are required to manufacture a set of parts. Furthermore, the VR walk through environment provides a unique tool for studying physical aspects of machine placements. Mathematical algorithms are used for generating independent manufacturing cells. This approach simultaneously identifies part families and machine groups and is particularly useful for large and sparse matrices. The biggest advantage of using VR-Fact! for the factory design is that it supports the user in planning space or logistical issues by interactively moving and relocating the machines after the simulation has been carried out. With integration to all levels of product development and production cycle, the virtual factory provides unique environment for achieving the above goals. The VR-Fact! software was developed on a Silicon Graphics ONYX 2 computer and can be visualized using stereo head mounted display and Crystal Eye stereo glasses. This model also has supporting software for viewing on a Fakespace Boom3C. The software can also be implemented on Pentium based PCs. -- T.KESAVADAS, M.Ernzer, L.Lefort, S.Narayansamy, A.Sudhir, C.Singh, Y.Menezes, A.Shedge
UNCERTAINTY VISUALIZATION AND HAPTICS IN THE SCOPE OF MINEFIELD DETECTION -- To develop an immersive environment for Uncertainty Visualization for minefield detection using visual, audio and haptic techniques. Visualized data often have dubious origins and quality. Different forms of uncertainty and errors are also introduced as the data are derived, transformed and interpolated, and finally rendered. In the absence of integrated presentation of data and uncertainty, the analysis of visualization is incomplete at best and often leads to inaccurate or incorrect conclusions. Whenever we have a large amount of data, it becomes quite difficult to interpret the data using visual senses alone. In such a situation, it is essential to make use of the sense of touch and hearing. This project deals with recreating a Virtual Minefield Environment using Uncertainty Visualization Principles. The Graphical User Interface has been developed using Open GL / C++ and GHOST SDK. and enhanced by incorporating Graphics, Sound and Haptics. -- T.KESAVADAS, A.Bisantz, A.Kamerkar, P.Bhide, S.Basapur.
DECISION SUPPORT AND VISUALIZATION TOOL DEVELOPMENT -- The primary research questions being addressed pertain to the development of visualization techniques that can be effectively used in collaborative design environments. These environments are to be used to make multiobjective decisions under uncertainty and to enable rapid trade-off decisions by geographically distributed participants in the context of Rapid Virtual Prototyping. Design decision-making is largely multiobjective in nature and characterized by various forms of uncertainty. Developing decision support tools, to support the tradeoffs inherent in these types of design environments is the focus of this work. -- K. LEWIS, C.L. BLOEBAUM.
RECONFIGURABLE SYSTEMS --The design of reconfigurable systems involves designing systems that are able to adapt to their surroundings based on new requirements or a change in operating conditions. These flexible systems allow for one system to perform multiple functions in an optimal sense. Designing flexible systems requires measures of flexible, integrating profit benefit and demand theory with engineering decisions, and developing evolutionary computing techniques to embody the systems in an optimization-driven design process. -- K. LEWIS
CYBERINFRASTRUCTURE IN ENGINEERING DESIGN -– We are using cyberinfrastructure to build more ubiquitous, comprehensive digital environments that become interactive and functionally complete for people, data, information, and tools. Our technology base for this work consists of design repositories, reverse engineering, ontology development, product platforms, functional modeling, and distributed computing. The current projects include Cyber-Infrastructure-Based Engineering Repositories for Undergraduates (CIBER-U) and developing a National Engineering Dissection Cyber-Collaboratory. -- K. LEWIS, K. ENGLISH, A. OLEWNIK.
COMPLEX SYSTEMS MODELING AND REALIZATION -- The design and realization of complex systems requires continuous interaction of multiple designers, design teams, companies, and their associated computer-based decision support tools. In this work, different game theoretical constructs are developed to model these interactions based on the level of cooperation among the decision-making entities and inherent organizational structure. The effect of various concurrent and sequential relationships on final product structure are studied and compared, and are used to re-engineer existing computational and organizational structures in industries that design complex systems.-- K. LEWIS.
AGENT BASED DISTRIBUTED DESIGN -- XLarge design companies typically use a number of suppliers to support the design and manufacture of large systems. However, when the suppliers are distributed and have limited information, they act as design agents making decisions with incomplete information based on their own incentives. Modeling this scenario using game theory and concepts from decision theory (group and individual rationality) to supply design decision support both descriptively and normatively is the focus of this research. -- K. LEWIS.
MULTIATTRIBUTE DECISION MAKING IN CENTRALIZED AND DECENTRALIZED DESIGN -- In this project, the focus is on multiattribute decision making, where the decision is being made by a single centralized decision maker or a decentralized set of decision makers each responsible for a single attribute. We are extending an approach to multiattribute decision making under certainty to conditions of uncertainty and robustness by ensuring consistency and transitivity within the stated preferences of the decision maker. We are also developing provable conditions for convergence and divergence of decentralized design processes. Lastly, we are using these conditions, to study the interaction and balance between individual and group rationality in decentralized design processes. -- K. LEWIS, T. SINGH.
OPTIMIZATION METHODS and APPLICATIONS -- this is a continuing effort in the study of optimization methods and their application to mechanical engineering problems. Past research in optimization methodology has considered a wide variety of optimization methods and applications. Recent efforts have been devoted to optimization in the PC Windows environment and have included: a library of optimization methods supported by graphics, visualization methods for exploring optimization problems, and an optimization interface to CAD/CAE packages -- R.W. Mayne, N. Khedkar, U. Bharadwaj, S. Erande.
PROGRAMMING FOR CAD/CAE PACKAGES -- recent projects in this area have considered the programmability of AutoCad, ProEngineer and Unigraphics. We have also developed an application using Visual C++ to perform automated optimal design and interact with CAD/CAE packages through spreadsheet formatting. The purpose of these efforts is to use existing CAD/CAE packages to study the complete automation of the design process from engineering performance requirements through optimization, analysis, design modeling and drawing -- R.W. Mayne, H.P. Qi, N. Sane, S. Sirsikar, S. Erande.
THERMOMECHANICAL MODELING OF SLIDING CONTACTS -- the modeling of engineering surfaces, with roughness, near-surface gradients in mechanical properties and complex chemical layers, represents, in many respects, the last frontier of mechanical design. We are using combined computational mechanics approaches and experimental studies to advance our understanding of such contact problems and to develop advanced computational tools for engineering analysis and design. Both friction and wear are being studies. The thermal aspects of the problem, coupled with elasto-plastic deformations significantly push the boundaries of current computational capabilities. Fast algorithms have been developed in order to make the solution of large problems more practical. -- A. SOOM, G.F. DARGUSH, Q. Wang, H. S. Ng, T. S. Hung.

Mechanical and Aerospace Engineering 318 Jarvis Hall • Buffalo, NY 14260-4400 | (716) 645-2593