How to generate a Design Of Experiments (DOE) in ODYSSEE CAE?

You can generate a Design of Experiments (DOE) within the DOE generator tool. To access it, either click on "DOE" button in the main ribbon, or click on (top-menu) > Tools > DOE generator

A new window appears

Focusing on the Generate DOE section on the left, you can Add one or several parameter(s) depending on the number of parameters of your system.

For each parameter, you can specify its name and min-max values.

Finally, you can click on "Generate DOE" button to create the parameters values combinations.

Going further, you can change the number of points to generate, as well as the DOE generation algorithm to use.

You can also visualize the points of the DOE thanks to the scatter plots, as well as in PCA (Principal Component Analysis) scatter plots.

When working on the DOE generation, there are some important points to keep in mind:

• The number of parameters and their ranges directly drive the size of the domain. If it is too small, the model will not allow much variation. If it is too large, it will be difficult to predict, a lot of sample points will be required to capture the system behaviors.
• The number of points drive how dense the domain is covered. If it is too small, the model may not be able to predict easily with a good level of accuracy. If it is too large, the model will take more time to predict.
• Depending on the system studied, some zones within the DOE may require more points to focus on a region which is more interesting for final processing, or where prediction might be more difficult due to versatile behaviors.
• DOE scatter plots and PCA scatter  plots are useful to help to visualize the distribution of the points of the DOE generated.

You can generate a Design of Experiments (DOE) within the DOE generator tool. To access it, either click on "DOE" button in the main ribbon, or click on (top-menu) > Tools > DOE generator

A new window appears

Focusing on the Generate DOE section on the left, you can Add one or several parameter(s) depending on the number of parameters of your system.

For each parameter, you can specify its name and min-max values.

Finally, you can click on "Generate DOE" button to create the parameters values combinations.

Going further, you can change the number of points to generate, as well as the DOE generation algorithm to use.

You can also visualize the points of the DOE thanks to the scatter plots, as well as in PCA (Principal Component Analysis) scatter plots.

When working on the DOE generation, there are some important points to keep in mind:

• The number of parameters and their ranges directly drive the size of the domain. If it is too small, the model will not allow much variation. If it is too large, it will be difficult to predict, a lot of sample points will be required to capture the system behaviors.
• The number of points drive how dense the domain is covered. If it is too small, the model may not be able to predict easily with a good level of accuracy. If it is too large, the model will take more time to predict.
• Depending on the system studied, some zones within the DOE may require more points to focus on a region which is more interesting for final processing, or where prediction might be more difficult due to versatile behaviors.
• DOE scatter plots and PCA scatter  plots are useful to help to visualize the distribution of the points of the DOE generated.