Wednesday, February 27, 2013

Simulating a lap

In F1 world anything that can help you make the car better is welcome. This time we will talk about software simulators and in particular, showcase the one from Ansible Design. It's called AeroLap and is able to simulate a whole lap given certain car parameters and even take environmental factors into account. This entire article would have not been possible without the kind cooperation from the people at Ansible Design, Woking, UK.
This software has been used successfully for years throughout all kinds of motorsport divisions, including existing F1 customers.


How it works
Using a complex, multi-layered model with many non-linear components, and performing hundreds of thousands of calculations for a run AeroLap can provide more realistic and accurate results than other methods.
The underlying calculation method is to discretise a given 3D path into small segments. For each segment the maximum thrust is applied to the car, according to the authority of the engine or braking system and limited by the grip available, driver behaviour and other forces on the car e.g. aero or gravity. A pseudo steady state solution is found for the sprung mass position and the solver focus moves to another segment. Segments are solved in the most efficient order, which is often not sequentially. When all segments have been solved the results can be presented as a continuous time history.
How the teams are using it
  • Decide race engineering strategy before going to the track.
  • Prepare your setups in advance for changes in track layout, or for going to a new track.
  • Make key design decisions before committing to building a prototype.
  • Choose gear ratios based on their effect on laptime, automatically taking account of many factors that are not considered when using ratio charts or spreadsheet modelling such as engine performance, track gradients, ambient wind speed.
  • Compare different engine torque/power curves, both in terms of predicted lap time and by plotting the engine torque or power actually being used throughout the lap.
  • Run a variety of aerodynamic trims and look at lift/drag trade-offs.
  • Predict ride heights and alter the suspension setup so that you are getting the best out of the car's aerodynamics.
  • Examine the effect of different fuel loads on performance for pit strategy.
  • Setup your suspension to get the best from the tires.
  • Gain a better understanding of what affects your car's performance and why.

They say picture is worth a thousand words, so instead of describing all parameters that can be modified I will feature them in a series of screenshots. The sample data we have available is for Le Mans car and track layout, but assume that track data can be modified to match precisely each of the F1 tracks, just as well as car parameters. Let's dive in and don't be scared by the amount of numbers - these are all vital parts of car setup. The article itself could always be updated to include even more screens.

Define and setup a model of car - you can choose parameters for:

  1.  Aerodynamics - reference frontal area, front and rear downforce, wind tunnel offsets, etc. This data can be also generated from AeroContour software, which is going to create the aero maps. Example:

    This is the one displaying downforce related to front and rear ride height, here's also one showing the drag:
  2. Chassis
    Defining the physical dimensions of the car, masses - front, rear and even mass on the grid (assuming driver plus fuel)
  3. Suspension

  4. Tires
    This includes data for all four tires
  5. Gearbox
  6. Environment factors

  7. Engine data

  8. Track data
Finally, as soon as we have all the input parameters set, we can hit the big Play button, which activates the solver and interactively the little yellow circle (the car) moves around the track. On a moderate normal desktop PC the simulation took around 8 seconds and the result looks like this: 

Other charts and graphs that are useful to data engineers are various parameters which progress as the lap advances, such as Speed vs. Lap or Torque / Engine RPM vs. Lap.
Here's an example:

On this final tab where you see the results of the simulation, there are multiple channels to select from, i.e. you can overlay the Speed vs. Aero drag, in our case labeled as Cd (coefficient of drag):

Quite obviously, the drag increases as the speed goes up. In the same manner multiple parameters can be displayed, derived from the base suspension, chassis, tires, aerodynamics, etc. These results are also available in table format as data values, so that you can export them and pass the output to another chart-drawing software. 

Finally, we would like to have a look at our results and revisit the strategy for the race. Back to the suspension module gives us the option to change the ride height from 65 mm to another value, I will select 80 and then re-run the simulation. The results are clear from this simple change: 
We lost about 0.7 seconds for the whole lap, and this is a clear indicator that increased ride height is perhaps not the most optimal strategy. In the same fashion you can tune and play with all available parameters, check the results and decide what to do. Again, this gives precious insight to race engineers, designers and aero people to make early predictions on what direction to follow when building / modifying a car, as well as create and adjust racing strategies. 
Aerolap, as a mature software, has a lot more to offer, for example integration with software from the same breed, such as AeroSusp or AeroContour to literally any kind of motorsport division. 

Thanks for attending, questions are always welcome.