Design space exploration and design optimization will be prominent topics at the NAFEMS 2016 Americas Conference June 7-9 in Seattle. These presentation abstracts from the just-published preliminary conference agenda promise a rich and rewarding program for attendees:
Design Exploration and Design Optimization to Improve Brake NVH Comfort, Sergio Carvajal, Daniel Wallner – Dr. Ing. h.c. F., Porsche AG; Michael Klein, Reinhard Helfrich, INTES GmbH
Brake squealing is the most famous noise generated by brake systems. This phenomenon can be characterized by a number of parameters influencing the occurrence and frequency of the squeal noise like brake pressure, speed, friction, and stiffness of parts and connections. Usually, these parameters are not known in advance like friction between pad and disc, and the stiffness properties of the pad.
The squealing itself is successfully simulated by Complex Eigenvalue Analysis (CEA) and taking into account contact and rotation of the brake disc. In this way, squealing is considered as a friction-induced instability problem, and CEA is the right means to identify the frequencies, where such instabilities occur.
Due to the complexity of the phenomenon, it is of high practical importance to identify the relationship between varying or uncertain parameters and the predicted squeal frequencies. Design exploration (DE) methods are used to establish this relationship in an effective manner. One set of DE methods consists of sampling methods, where many different parameter sets of the brake system are used for CEA and a stability map is generated from the sampling results.
The huge amount of data generated by sampling allows the identification of brake parts, which have the highest influence on the occurrence of instabilities. Then, design optimization (DO) is the next logical step to reduce the tendency for squealing of the brake. Such relevant optimization methods are shape optimization methods, for example, which allow the optimization of the stiffness of brake parts without weight increase. The final optimization result is an updated stability map for the optimized brake system.
The presentation will show the use of CEA, DE, and DO methods integrated in the software system PERMAS. A PORSCHE brake system is used to demonstrate the simulation process.
Developing Quasi-Static and Fatigue Loads for Earthmoving Vehicles to Optimize System Performance, W. Tanner, Adaptive Corporation [Abstract not yet available.]
HPC CAE Simulation for Surf Bums, Arnaud Froidmont, Barrett Newill, Noesis Solutions
Of all the stereotypes surrounding surfing, nothing pervades society like the image of a “surf bum.” You know him, the blonde haired character on TV whose sole job it is to be unmotivated and worthless. Yet even as we mercilessly degrade people with a skill set that takes years of hard work, there’s something universally appealing about the surf bum.
These bums may not be successful in conventional terms, but surf bums have a lot more going for them than the average layman. They’re motivated in a quest of passion. So, next time you find yourself scooting out of the water in order to get your engineering drawings in on time, think about what it would be like to live the alternative: to consciously bum it up. To laugh in the face of financial instability and to avoid work at all costs. These guys need CAE on HPC in order to maintain their lifestyle because a) they can’t afford their own computers, b) they are too busy shredding to creating complex workflows and workloads and c) many don’t have advanced physics degrees.
We show how automating open source and commercial codes were used to enable someone that shapes surf boards with no CAD or CAE background to create a better surfing experience. By utilizing an automated workflow that interfaces with a CAD tool, mesh morpher, CFD code, and a post processor, the user can, through a Microsoft Excel based interface drive the entire process. With robust connectivity and a stable CFD mesh and solution setup reliable drag and vorticity results can be communicated back to the user.
In order to enable this seamless connectivity a process integration and design optimization tool with a fully open API was used. Optimus, by Noesis Solutions, was the selected tool. The Python API inside of Optimus allowed for linking of Excel to the workflow. The easy-to-use graphical interface and debugging tools enabled development of the workflow.
This presentation will show both the process to enable the study and the results of the optimization. It is sure to be interesting and engaging topic on the best surfing beaches in the world.
Innovation Design of a Curve Ribbed-Plate Subjected to Turbulent Boundary Layer Excitations, Dr. Nicholas A. Stowe, Eric Brown, Dr. Kuangcheng Wu, Ship Survivability, Newport News Shipbuilding, a Division of Huntington-Ingalls Industries
Flow-induced vibration and noise is a design focus for the aerospace and automotive industries, among others. Despite the increased computational power of today’s high performance computers, fully-coupled high-fidelity models of fluid, structure, and their interaction remain out of reach for large models. When the one-way coupling assumption is valid, however, simulation can be used to gain an understanding of the effects of design decisions on noise and vibration due to turbulent boundary layer (TBL) excitation.
In this paper, one-way coupling is used to investigate the noise and vibration generated by TBL flow over an elastic structure. The first step is to characterize the boundary layer, which is achieved using panel methods in conjunction with boundary layer models developed by Thwaites, Michel, and Head (Thwaites, B, Incompressible Aerodynamics, 1960; Michel, R ONERA Technical Report, 1951; Head, Aeronautical Research Council, England 1960). A panel method is used in order to capture the influence of plate curvature on the TBL fluctuating pressure. The TBL excitation is calculated using methods described by Goody (AIAA 2004) once the critical boundary layer parameters have been determined. This excitation is combined with structural acoustic transfer functions generated by Energy Finite Element Analysis (EFEA) to predict the structural and acoustic responses. EFEA is a computationally efficient means of calculating high-frequency acoustic and vibration effects. The combined techniques provide a quick turn-around model for TBL-induced noise and vibration.
More interestingly, the integrated model has been exercised by optimization algorithms to explore the relationship between design parameters and design performance. The analysis of TBL-based excitation effects has been completely automated to achieve this aim. Using optimization to manipulate the model enables engineers to perform simulation-driven design, where decisions are directly informed by the physics in a manner that frequently reduces risk and increases performance. In this case, stiffened plate geometry, thickness, damping, and curvature are modified by the algorithm to search the design space and identify the best design for acoustic performance within the defined parameter bounds. Optimization is conducted by gradient-based and genetic optimization algorithms and the results and optimal designs are compared and discussed.
pSeven: A First Full-Cloud Design Space Exploration Platform, Sergey Morozov, DATADVANCE
Nowadays a new generation of full-cloud design and simulation software tools is emerging. Among of the pioneers are Onshape, SimScale and Sim4Design. Although these tools offer great design and simulation capabilities, one component—design space exploration—is missing. But this is an essential component for implementing simulation-driven design methodology.
In this paper we present a full-cloud design space exploration system, pSeven Cloud. It is based on the industry-proven efficient data analysis and optimization algorithms supplemented with SmartSelection metaheuristics allowing to select the most efficient algorithm for a given problem. pSeven Cloud features easy-to-use graphical user interface allowing simulation experts 1) to easily automate existing design processes by creating simulation workflows, 2) to effortlessly implement multidisciplinary design space exploration strategies, 3) to share these workflows as a ready tools, a.k.a. as vertical applications, with non-experts within the company, and 4) to execute studies using existing HPC infrastructure.
We demonstrate solution of real-life industrial problems using pSeven Cloud integrated with full-cloud CAD and CAE solutions like Onshape, SimScale and Sim4Design.
Random Response and Fatigue Optimization in the Frequency Domain, Fatma Koçer, Sridhar Ravikoti, Altair Engineering, Inc.; Neil Bishop, CAEFatigue
Frequency-based methods for random response and fatigue are becoming more widely used in the automotive industry. The use of PSD’s (Power Spectral Densities) coupled with system properties in the frequency domain (transfer functions) offer significant benefits over time based approaches in terms of analysis time, model sizes that can be handled, and advanced diagnostic information available to improve the designs. Most importantly, such methods facilitate the incorporation of optimization techniques in the product development cycle to find the optimal values for parameters such as part thicknesses, material properties for reduced weight and improved design life.
However, even with analysis methods that provide affordable solution times for large system models, engineers and designers do not yet explore the full potential of optimization. Process automation that is required to conduct optimization studies can be daunting for several reasons. The design simulation process is usually not comprised of only one solver execution but includes several solvers. In addition to solvers, the engineer needs to automate parametrization and response extraction. While dealing with even faster-to-design cycles than before, engineers don’t always have the time to develop connections for all the tools required and incorporate it in their design process. The bandwidth to learn the technology can also be another challenge. In order to facilitate the transition from trial and error process to design exploration and optimization, the tools that make up the analysis process needs to communicate with each other seamlessly and offer powerful optimization methods that are also highly usable by engineers.
In this paper, application of frequency based methods for random response and fatigue optimization is demonstrated on an automotive full body system. The virtual durability process followed uses MotionSolve for multi-body simulation to generate body loads and OptiStruct for finite element analysis to compute stress states as input to frequency-based fatigue analysis using CAE fatigue. The process is automated and executed for optimization in HyperStudy and the results are then post processed using HyperView. This process is set up and executed within a work week due to the existing integrations between the products and usability of optimization methods in HyperStudy, specifically Global Response Surface Method. Using the process outlined, it is feasible to achieve considerable weight reduction while maintaining desired fatigue life within a typical project timeline.
Weight Optimization of a Transmission Housing, Nils Wagner, Reinhard Helfrich, INTES GmbH
The shape of a transmission housing is derived from function and space of the contained transmission. It can be best characterized by a free geometry stiffened by ribs. From an environmental point of view, the housing provides an oil-tight hull of the transmission. From a structural point of view the housing is a stiff connection between the bearings of the transmission. Due to assembly reasons, housings usually consist of several parts, which are bolted together. Then, after final design, the question is, can we reduce the weight of the housing by reducing the wall thickness of the frequently cast parts of the housing? Additional conditions are a stress limit due to durability reasons and a displacement limit at the bearing positions to keep the overall stiffness of the housing.
The design of a transmission essentially depends on the contact behaviour between gears, in the bearings, and between the bolted parts of the housing. Consequently, the shape optimization of the housing has to use contact analysis as basic analysis method. Due to the free surface geometry of transmission housings, only a nonparametric free form optimization can be used, which has to be combined with parametric constraints like displacement limits.
The presentation will show the freeform optimization of an industrial example to reduce weight by wall thickness changes under stress and displacement constraints. The optimization integrates contact analysis in one single software (PERMAS). This drastically reduces the run time for the optimization.
Design Space Exploration: Adoption Drivers, Adoption Constraints, Potential Adoption Accelerators, Bruce Jenkins, Ora Research
Design space exploration today is enjoying ever-increasing levels of recognition, adoption, and successful application to some the world’s most challenging engineering problems. Nonetheless, there remain significant impediments to the even broader deployment and usage that champions of these tools and methods believe will be possible in the not-too-distant future.
Design space exploration, in the sense used here, encompasses a family of applications and work processes that include design of experiments (DOE), multidisciplinary optimization (MDO), multi-objective (Pareto) optimization, stochastic (robustness and reliability) optimization, and the broad array of structural optimization methods – shape, size, topology, topometry, topography, more.
These are some of the latest developments that are driving and accelerating the pace of adoption and impact today:
- Increasing levels of built-in intelligence to let design exploration software choose the best search algorithms and solution methods autonomously, based on the user’s description of the problem in engineering terms. The presentation will cite specific examples of feedback from end users who report that increased levels of built-in intelligence and autonomous operation are making this class of software more usable, beneficial and impactful than in the past.
- Appification of simulation—the embedment of design exploration and optimization technologies inside easy-to-use, product- and customer-specific simulation apps.
- Full-cloud solutions that expand accessibility, affordability and usability of design space exploration.
- Continued vigorous marketing and sales activity by large CAE vendors that own premier design exploration and optimization technology.
- Compounding pressures on engineering organizations to find new ways to do more with fixed resources, for example meeting automotive CAFE and emissions mandates.
On the other hand, these are some of the legacy conditions responsible for constraining and retarding adoption and impact:
- Design exploration and optimization are still not part of the standard work process at enough engineering organizations.
- The technology remains too often implemented at only the workgroup or department level, instead of as an enterprise competency.
- Many small software developers offering highly capable technologies continue to struggle under marketing/sales resource constraints.
- Some PLM vendors have yet to embrace design space exploration, fearing it a troublesome complication in an already complex CAE sales process.
These are some “wild-card” issues whose near- and intermediate-term impact will bear watching closely:
- Market-development ripple effects from the growing democratization of topology optimization?
- More acquisitions of design exploration/optimization software developers by major CAE and/or PLM vendors?
- New startups fielding breakthrough technologies?
The presentation will use these and related issues as a springboard to explore what steps can be taken by technology developers, engineering practitioners, and engineering management to further unlock the full potential of design space exploration.
All information © NAFEMS.