Once you create the initial model and define parameters, your next step is to create a response surface. After you insert a Response Surface system in the Project Schematic, you define the design space by specifying the minimum and maximum values to consider for each input variable. Based on the information that you enter in the Design of Experiments (DOE) cell, the Response Surface system creates the design space sampling. The sampling depends on the selected DOE type and that the default CCD design type provides good accuracy for the final approximation, which is typically the case. Next, the DOE must be computed.
When you update the DOE, DesignXplorer creates a response surface for each output parameter. A response surface is an approximation of the response of the system. Its accuracy depends on several factors, including complexity of the variations of the output parameters, number of points in the original DOE, and choice of the response surface type.
DesignXplorer provides a variety of response surface types. The default type, Genetic Aggregation, automates the process of selecting, configuring, and generating the response surface best suited to each output parameter in your problem. Several other response surface types are available for selection. For instance, Standard Response Surface - Full 2nd Order Polynomials, which is based on a modified quadratic formulation, provides satisfying results when the variations of the output parameters are slight. However, Kriging is more effective for problems with a broad range of variation.
After response surfaces are created, you can thoroughly investigate the design using a variety of graphical and numerical tools. Additionally, you can use optimization techniques to identify valid design points.
Usually, the investigation starts with viewing sensitivity graphs because they graphically display the relative influence of the input parameters. These bar or pie charts indicate how much output parameters are locally influenced by the input parameters around a given response point. Varying the location of the response point can provide a totally different graph. Explanations of these graphs typically use a hill and valley analogy. If the point is at the top of a steep hill, the influence of the parameters is large. If the response point is in a flat valley, the influence of the input parameters is small.
The response surfaces provide curves or surfaces that show the variation of one output parameter with respect to one or two input parameters at a time. These curves or surfaces also are dependent on the response point.
Both sensitivity charts and response surfaces are key tools for answering such what-if questions as "What parameter should we change if we want to reduce cost?".
DesignXplorer provides additional tools for identifying design candidates. In addition to thoroughly investigating response surface curves to determine design candidates, you can use optimization techniques to find design candidates from a Response Surface cell or any cell containing design points. DesignXplorer provides two types of goal-driven optimization (GDO) systems: Response Surface Optimization and Direct Optimization.
You can drag a GDO system from the Toolbox and drop it in the schematic. If you drop a Response Surface Optimization system on an existing Response Surface system, these two systems share this portion of the data. For a Direct Optimization system, you can create data transfer links between the Optimization cell and any other cell containing design points. You can insert several GDO systems in the project, which is useful if you want to analyze several hypotheses.
Once a GDO system is inserted, you must define the optimization study. This includes choosing the optimization method, setting the objectives and constraints, and specifying the domain. You then solve the optimization problem. In many cases, there is not a unique solution, and several design candidates are identified.
The results of the optimization are also very likely to provide design candidates that cannot be manufactured. For example, a radius of 3.14523 mm is likely difficult to achieve. However, because all information about the variability of the output parameters is provided by the source of design point data, whether a Response Surface cell or another DesignXplorer cell, you can easily find an acceptable design candidate close to the one indicated by the optimization.
You should check the accuracy of the response surface for the design candidates. To verify a candidate, you update the design point, which checks the validity of the output parameters.