Correlation studies and design of experiments (DOE) algorithms automatically sample the design space and provide insight into correlation between inputs and outputs, design sensitivities and more. You can filter parameters to focus on the most relevant design inputs and easily create charts and reports to provide design insight. You can also use the resulting design point information to determine the Min and Max points of the design and to build a response surface.
Understanding how the outputs (performance) are related to the inputs (design and operating conditions) is the best path towards innovation and significant return on investment for your simulation dollars.
ANSYS DesignXplorer includes industry-leading algorithms, as well as many industry standard algorithms, that analyze the table of design points to produce a response surface. You can use the response surface to instantly predict the performance of the design without needing to run the actual simulation. This reduced order meta-model can then be used for sensitivity studies, optimization and six sigma analysis.
DesignXplorer uses statistical methods to evaluate the fitness of the response surface to ensure it can accurately predict the performance of the simulation model. If necessary, auto-refinement is available to improve the resolution of the response surface.
DesignXplorer offers a variety of algorithms, including a number of industry standards. Our most advanced algorithm is known as “GARS”, Genetic Aggregation Response Surface. This industry-leading algorithm automatically finds the best response surface formulation and includes built-in cross-validation and auto-refinement capabilities.
Once created, response surfaces can be used in DesignXplorer or can be exported as reduced order models (ROMS) for use in other tools, such as Simplorer, Excel or even other Workbench Projects.
Industry-leading design optimization algorithms drive Workbench to find improved designs automatically. These algorithms can search the design space with direct solves or use the response surface. You can take multiple objectives, constraints and parameter relationships into account easily define them in the user interface. Design Xplorer tracks the progress of the optimization and produces a variety of charts and tables that make it possible to evaluate tradeoffs and choose from the best design candidate(s). You can easily transfer the winning design parameters back to the CAD software via our bi-directional CAD interfaces.
Using automated design optimization tools leverages the simulation software for rapid gains in design performance. The algorithms can find improvements that may have eluded the engineer's intuition or experience.
Six Sigma Analysis
A single simulation gives an indication of how a design performs under ideal circumstances, but a six sigma analysis investigates the range of performance as the design and operating parameters vary. In many cases, this can reveal significant design sensitivities or flaws that may have been missed by the simulation.
DesignXplorer enables you to enter the statistical variation of the inputs and study the statistical distribution of the product performance. DesignXplorer automatically produces charts for six sigma distribution or design sensitivity. You can then use this information to tighten parameters as needed to improve quality and robustness, or loosen some tolerances to reduce costs.
Export Response Surfaces
DesignXplorer can export its response surfaces for re-use elsewhere. Thee surfaces can be exported as industry standard *.FMU files and imported into a wide range of tools including Simplorer, Matlab, Excel and many others. They can also be imported back into Workbench for rapid evaluations.
Response surfaces can be used to instantly predict design performance for any given set of design inputs or operating conditions.
As a native Workbench application, DesignXplorer capabilities can easily be extended with ACT. Many are already available in the ACT store, or you can even create your own.
Extensions can be used to just about any sort of custom functionality to DX. You can use them to add algorithms for , such as for DOE or Optimization. They can also be used to auto create workflows and pass data between systems, such as DOE from Correlation or Direct optimization from RSO.
The ROM builder uses advanced mathematical methods to combine 3D (or 2D) solver result snapshots from a set of design points into a standalone digital asset. The 3D ROM file can be exported in standard formats (including FMU) that can be imported into hundreds of software tools, including ANSYS Twin Builder (for predictive maintenance). It can also be used with 3D ROM viewers so you can visualize design ideas almost instantly.
Once you have the ROM, you can show it to others at a board meeting, pitch it on an elevator, post it for customers on a website, build it into a vertical application, or use it as a digital twin out in the field. Wherever you take it, you can illustrate, in colorful 3D, how your design will perform.