Tutorial – Aero-Mapping Setup
The Benefits of Aero-Mapping
bramble’s ‘Aero-Mappings’ feature can be used to help visualise and predict vehicle performance at different test conditions by interpolating between the data points of an initial set of simulations.
The mapping process begins by running a set of initial simulations, such as a sweep of front and rear ride heights, and then using the outputs from this simulation (such as drag and downforce) to generate the mapping.
Once generated, it is possible to:
Predict design performance (such as downforce at a given ride height)
Visualise how performance varies with different parameters
Generate Pareto plots to show trade-offs between different performance measures (like drag and downforce).
Optimise for a maximum or minimum performance measure.
Identify new designs and add them to a Map for testing.
The first stage in running an aero-map is defining the initial test conditions, also know as the ‘Design of Experiments’ or DoE. The DoE can be defined a number of ways, however, the recommended approach is an Optimised Latin Hypercube.
An Optimised Latin Hypercube attempts to distribute tests equally throughout the design space. It can use fewer initial experiments to cover the design space than other approaches such as a Full Factorial which becomes expensive as the number of parameters increases.
Optimised Latin Hypercube maps are generated in bramble from the Maps view.
Begin by creating a new map as normal, but before adding any attitudes, press ‘optimised latin hypercube’ from the available options.
Fill in minimum and maximum values for the parameters that you wish to adjust in the map. Leave parameters that are not changing empty.
Enter an integer value for the Map Level Detail. This is used to control the total number of experiments, where total = MLD x Number of Parameters. 5 is a typical value, 10 will produce a dense DoE.
Finally, press ‘Generate Map’. After a few seconds the attitudes will appear. Generating larger aero-maps will take longer.
The attitudes can be edited as desired when they appear, although they should be fine as they are. Additional attitudes can still be added manually.
Blank columns can be left empty; these parameters will use the defaults set in the kinematic setup. Otherwise they can be filled in now, although they should be constant across the map.
Once the attitudes are set as desired, save the map and create a run using the map as you would normally.
Running the Simulations
From the ‘Runs’ view, create a new simulation by either selecting ‘New Run’ or ‘Create Child’ from a baseline. Ensure the appropriate Kinematic Setup is selected along with your newly created Map.
In the detailed view for this new run, you should find lifecycles for all the attitudes in the map. Before launching, ensure you have the correct CFD Setup configured.
Additionally, it is recommended that you set either ‘Modification’ or ‘New Attitude’ Check Templates in the Run Checks tab.
All the attitudes can be launched at once using the ‘Group Transitions’ ‘Launch’ button.
Generating the Aero-Map
It is possible to generate the mapping data once the DoE simulations have completed. This is done from the Mappings section. Note, before you do this, you need to have saved a force table in the Results view. The data contained in the force table will be used to generate the aero-map.
Begin by ensuring you have the correct Programme and Project selected, then click on ‘Mappings’ and then ‘New Mapping’
Give the mapping a name, select the run(s) that you have just completed.
Select the parameters that have been altered the map. Note: the list of parameters will be filtered to remove ‘constants’ (those that weren’t adjusted).
Select the Force Table settings that you wish to use for this aero-map, and finally the parameters you wish to generate mappings for.
Note, the more parameters you select, the longer it will take to generate all the data.
View the Aero-Map
Initially, the aero-map will appear with a ‘calculating’ notice whilst the aero-maps are calculated. The generation of the mappings will take a few minutes or longer depending on how many performances were selected and how busy bramble’s workers are.
The option to view or delete the data will appear after all the mappings have generated. Click on ‘view’ to load the outputs. You will be presented with a table showing the DoE results along with a number of options for interrogating the aero-map.
Predicting a Single Point
The simplest operation is to predict the performance of a single design. Select the ‘Predict One Point’ button, and then in the window that opens, fill in the design parameters you wish to know the results for. Press ‘Predict and Add To Table’.
The predicted performances will appear and the will be added to the Mapping Table. Press ‘Close’ to view this result.
Predicting a Linear Sweep
Similar to predicting a single point, it is also possible to predict the performance for sweep of designs. Press ‘Predict Linear Sweep’ and enter a range of min/max parameters that you wish to sweep between.
For example, for a rear ride height sweep, we would put values of 0.025 for front ride height (i.e. keeping constant) and then 0.05 to 0.1 in the rear ride height boxes.
Enter then number of points you would like to predict and finally press ‘Calculate’. The window will close and after a short wait, the data will appear in the Mapping Table.
Optimise a Performance
The ‘Optimise a Performance’ can be used to determine a minimum or maximum value of a specific performance and the design parameters that give this results.
Begin by selecting which performance you wish to optimise and whether you want a minimum or maximum value.
Then select the bounds of the parameters you wish to search with. This will default to the whole design space, but you can adjust these settings to search within a localised area.
Finally, press ‘Calculate’. After a short moment, the optimised variable will appear in the Mapping Table.
Plot Interpolated Points
‘Plot Interpolated Points’ is used to generate a 2D contour plot showing how a performance varies when two parameters are altered.
Select the parameters you wish to have on the x-axis and y-axis of the plot. Then select the performance measure you wish to view and then set bounds.
From the ‘set bounds’ window, specify the range over which you would like view the results. These will default to the whole design space, but you adjust these to view a localised area.
The ‘Number of Points’ options can be used to control the refinement of the contour plot. The greater the number of points selected, the longer it will take to generate the contour. 50 x 50 is a good value.
For aero-maps of more than two parameters, you will also need to specify constant values for the remaining parameters.
Finally, press ‘Plot’ to load the contour plot.
The contour plot will take a short moment to load, taking longer if you selected a larger number of grid points.
The contour plot will have a the x-axis and y-axis labelled, and a scalar bar for the performance metric will be displayed to the right.
Hovering your mouse over the contour will show the parameters and performance at that particular location.
Selecting the ‘Show Normalised Uncertainties’ will alter the contour plot to show the uncertainty values for this data. Essentially, this is measure of confidence in the predicted result.
Uncertainty can be used to determine where best to refine the DoE with additional simulations.
Extending the Aero-Map
Clicking on a design in the contour map will add that point to the mapping table. This can be used to keep a record of the interesting results.
It can also be used to pick out the designs that have a high uncertainty. Once picked it is possible to add the designs to the original Map and run the simulation. This will help improve the quality of the Aero-Map as a whole.
To add the design to your Map, select the ‘Choose Map’ option. This can be done for any of the predict/optimised designs.
In the window that appears select your original Map and then press ‘Add’.
Once added, if you go into the Maps view and load your Map, you will see that the interpolated points will have been added.
The easiest way to run these additional attitudes is to go into the run where the map was initially completed. Press the run edit button in the top-left of the view, and then press ‘Save’ on the window that opens (no other changes needed).
This action will cause bramble to compare the attitudes currently in the run with those in the map. Any missing ones in run (i.e. the newly added designs) will be added and their lifecycles will appear.
These attitudes can be launched as normal.
Viewing a Pareto Plot
More often than not, it is desirable to understand how one performance varies with another. For example, how drag changes and drag is increased. This is done using a Pareto plot. To do this, select the Generate/Choose Pareto Plot button.
If you have not viewed the Pareto plots before you need to first press the ‘Generate Pareto Plot’ button. The button will go into a pending state whilst the data is generated. This may take a few minutes depending on the number of performances in the Aero-Map. Once completed, a ‘Choose Pareto Plot’ button appears. Select this and in the window that opens choose the performance measures you wish to look at.
The resultant graph will show a scatter of points showing the range of two performances that can be achieved by altering the design parameters.
When generating the Pareto, the limits of the design parameters will be extended beyond the original DoE. Design points within the original limits appear as blue, and those outside appear as orange.
Hovering over a point will show the design parameters and click on it will add the design to the Mapping Table. From here, the design can be added to the original Map as previously.
Viewing a Cloud Plot
A cloud plot is very similar to the Pareto plot and will need generating the first time it is clicked on. The key differences are that the plot is more densely populated with points and there are ‘sliders’ for selecting ranges for the design parameters.
This plot is used to understand where in the performance space certain designs appear. For example, if we wanted to see where high rake ride heights appear, we would select a range of low front rides and high rears.
Press ‘Filter’ to update the plot once you have selected your ranges.
The points that fall within the selected range appear as blue dots and those that are outside appear as grey.
As with the Pareto plot, hover your mouse over the design to view the performances and design parameters. Click on a point to add it to the Mapping Table.
The Mapping table can be downloaded at any point as a comma separated variable (CSV) file. This file can be loaded into Microsoft Excel or similar software.
The downloaded CSV will include the original DoE results and any interpolated points that have been added to the Mapping Table.