Bosch Lapsim

[oz_olly]

Has anyone got experience with Bosch Lapsim? I am trying to use it for some simulation work for my thesis and am getting absolutely terrible correlation. I have adjusted just about every parameter I can think of. From my experience with simulation software the laptime is highly dependant on lateral accel and to me good correlation is the recorded lateral accel and the simulated having approximately the same frequency but but the sim having higher or lower amplitude. This is most certainly not the case with lapsim. I hope I can get it figured because $9000 AUD is a stupid amount for lapsimulation software that doesn't perform very well.

Cheers

Olly

ACME Racing
UNSW@ADFA

[oz_olly]

Has anyone got experience with Bosch Lapsim? I am trying to use it for some simulation work for my thesis and am getting absolutely terrible correlation. I have adjusted just about every parameter I can think of. From my experience with simulation software the laptime is highly dependant on lateral accel and to me good correlation is the recorded lateral accel and the simulated having approximately the same frequency but but the sim having higher or lower amplitude. This is most certainly not the case with lapsim. I hope I can get it figured because $9000 AUD is a stupid amount for lapsimulation software that doesn't perform very well.

Cheers

Olly

ACME Racing
UNSW@ADFA

[B Hise]

have you emailed Chris yet?

[oz_olly]

Yeah I have, I just thought the issues I am having are related to the corner radii and the closeness of corners which is pretty unique to FSAE and especially last years FSAE Australasian Enduro course.

Cheers

Olly

[murpia]

Hi Olly,

Some of what I'm about to say is from memory, so apologies to Bosch if I have it wrong...

I looked at the Bosch LapSim demo last year and it appeared (by watching the graphics when solving) that the solver is a forward-backward solver from a minimum corner speed (the 'apex' of the trajectory, if not the real physical track apex).

This is a common simulation solver type that uses track data of speed and lateral g to derive a trajectory (series of curves) for the car to follow. The simulation looks for a number of clear minimums in the speed data (which represent the corners). If these minumums are not present sufficiently clearly the simulation can have a problem.

The typical speed trace for an FSAE style autocross / endurance circuit may not present these minimums sufficiently clearly, especially the slaloms.

Can you get a good correlation for a very simple 'oval' course, if you can get data?

Of course, this may not be what you experience as a problem... Also let me know if I'm explaining things clearly enough...

Regards, Ian

[oz_olly]

Hi Ian,

Thanks for your reply. From talking to Chris Van Rutten at Bosch I am fairly confident that I agree with you on the type of solver it is. He has told me it is a quasi steady state simulation, although I am not quite sure what that means. The 2006 Aus Enduro course was very wiggly which is what is causing the problems and I think it is due to your explanation. The simulation is seeing the data as a bunch of wiggles with a few straights. That is why in one of my other posts I have asked if anyone would like to share data from one of the other circuits as I understand from recent threads that the courses have been a bit faster and open. I believe from talking on the forum that the Detroit track was particularly open which would be great for the simulations I am doing. It would then be even better to be able to present data from both types of track but if noone shares I will not be that lucky (Pretty please share I will acknowledge you and your team in my thesis http://fsae.com/groupee_common/emoticons/icon_smile.gif).

I am getting quite good results with E-Lap provided by ARC apparently it began its life as a Minardi simulation. To give my work a sound engineering backbone I believe it is important to use more than one simulation, hence also trying to use LapSim.
For anyone that is interested I am studying the effect of final drive ratio, shift delay, efficiency and gearbox ratio spread. It great fun but takes ages to get results.

Cheers

Olly

[murpia]

Quasi-static refers to the type of solver used to determine the performance of the vehicle at each individual time or distance step. It means that each step is treated as isolated in time and a complete vehicle solution found for the generated constraints at that time or distance step. Practically speaking it means that dampers, acoustics, aerodynamic vortex shedding, tyre relaxation length and other transient phenomena cannot be included in the model. However it is still a very successfull approach to use.

This is distinct from forward-backward which is the type of solver algorithm used to connect together the individual quasi-static solutions into a coherent lap, by defining the constraints for each individual solution.

It is this second factor that is probably causing you difficulties. Also, search for an SAE paper by Blake Siegler (a former colleague of mine) which evaluates the various modelling approaches. There are others co-authored by Pat Symonds of Renault, also.

Regards, Ian

[oz_olly]

Thanks for the explanation. I haven't had much success trying to find papers on race car lap simulation so hopefully those leads you have given me will also be very useful.

Cheers

Olly