Lapsim by bosch

[Silente]

Hi,

i am playing with lapsim since some days and i have seen that it uses a particular tyre model with a short pacejka set of coefficients.

Some time ago, BillCobb said that it is actually possible to find from tyre data the coefficients to model the tyre in lapsim way.

Lapsim Vehicle and tyre parameters can be opened with matlab. So if anybody knows how to extrapolate this coefficients we could play around with a free lapsim.

I have seen, in particulare, that the coefficients the software has inside seem to be not so much sensitive to load. So, for example, you change track lenghts and/or CG height, results doesn't change so much.
But, the same changes produce a sensible difference in maximum cornering performance estimated with a corner simulation i made in matlab..

Anybody can help?

Maybe this is a question for Bill...

[BillCobb]

I believe I posted the code for the Lapsim tire fitting routine on the TTC site. In fact, it produces coefficients for additional surfaces (some of which Lapsim doesn't use, but are useful for other non-Lapsim modeling, synthesis and analysis projects.

I'll go see where I stashed it...

[BillCobb]

I put some files out on the TTC forum for you to work with. One is a GUI to show you what the 4 terms do to the tire Fy characteristics.

The other file uses the 4 term model to compute the 4 term FY, MZ and MX coefficients from TTC data. Lapsim does not use the MZ or MX data, but you SHOULD.

Note that I've used this tire model to perform some rudimentary vehicle testing on a FSAE sample car in an ISO procedure. This Matlab model is also posted in that forum. The point is that this type of analysis should be guiding your vehicle parameter specifications and tire selection process.

[Silente]

I know your posts with example of how to work in matlab on tire dta. I found them very useful.

I actually have done a simple constant radius cornering sim to find maximum grip and the effect on cornering perforamnce of car parameters, considering also Mz (not Mx for now).

And i have seen that where there is no difference with lapsim, i see differences in maximum lateral acceleration performed by my model.

Is this something that makes sense to you?
Do you know how lapsim find cornering performance at the apex? Because it is very fast in finding a lap solution, doesn't seem to iterate till a convergnece point.

[exFSAE]

Don't know specifically how Bosch's software works.

You could always write your own lap time simulation program, won't take more than a couple days.

[BillCobb]

Run your model without including Mz effects. Ruud feels that Mz effects are very small at max lateral g, but that's usually not the case because of the net rigid body aligning moment.

While tires are usually insensitive to slip angle changes at max lat, weight transfer and camber influences are still strong players. Implementation of these factors would produce a difference.

BTW: Differences are good. Findings of no difference would make me suspicious of the model. There may be effects caused by the use of the free version vs. the purchased version, too.

[Silente]

it won't take more than a couple of days if you know how to do it!

I hae not so much time at the moment, but just now i don't know the logic behind the driving the car around a lap.

Till now i have only iterated on a corner, but lapsim seems to find cornering sped and accelerations at the apex like if it was something given immediately by a formula..

[Silente]

Anybody has experience on doing a lap simulation?

Or anybody can direct me on something about it?

I guess the first step is to find, as i said, minimum corner speed, and then to elaborate the rest, but what about the way to do it and what, for example, about braking points etc?

[exFSAE]

There are corners, and there are straights. Sounds like you have the corner part down.

In a straight, there's acceleration and there's braking. Make something simple for each, then think about how to splice them together.

Then figure out how to splice a corner onto either end.

[RobbyObby]

http: //fsae .com/ eve/forums/a/tpc/f/125607348/m/217101453?r=31620500941#31620500941

Read this post from the "Reasoning your way through the FSAE design process" thread.

[RobbyObby]

Apparently I didnt put enough spaces in my link.

Read through the "Reasoning your way through the FSAE design process" thread, and on the second page Big Bird gives an excellent breakdown of designing a lap sim from scratch.

[thewoundedsoldier]

My "lapsim" utilizes a loop. The end output of each loop is the vehicle speed at the exit of the turn (the start of the chute). Once this output becomes the input for the next loop, the initial velocity of the vehicle in the chute is known, and combined with a turn radius and chute distance, the required braking distance is easy to figure out. It's all basic dynamics stuff. You can make it more or less complicated based on if you want to use constant-acceleration equations, or use more in-depth models that simulate your car in a variety of ways.

By using the loop method where I have an input/output of vehicle speed, I break down the track into "X" amount of 1-meter long turns to get the overall picture. It seems to be the pretty common way to do it.

[murpia]

My guess on how the 'free' Bosch Lapsim works is this (partially based on observing the graphics while it solves):

Track curvature is used to find apexes of maximum curvature. A steady-state solver finds a maximum speed that this curvature can be driven at. The solver works forward / backwards from there for each step by steady-state solving for 'spare' braking / traction grip and using that to brake / accelerate the car (subject to engine torque limits). Where the solutions overlap dicates the transition from one acceleration region to the next braking region.

If the lap boundary is in a region of significant curvature, there may be some algorithm to smooth any discontinuity.

Since Bosch Lapsim says that the solver sensitivity to yaw inertia is small, it suggests that it solves without a strong yaw angle / yaw moment / yaw acceleration correlation from step to step.

Such a solver would work well for tracks derived from real data where the curvature is continuously and smoothly increasing as the apex is approached, and decreasing as the apex is passed. It might work less well for constant radius tracks made from scratch. Bosch Lapsim requires a logged dataset to generate its own track map, although in theory you could fake it.

Note, the 'paid' version may work differently...

Any reasonably competent vehicle dynamicist could find many scenarios where such a solver would not predict the optimum car performance. However, it is still extremely useful and it is a credit to Bosch that they make available a free version. No-one else does that to my knowledge.

Regards, Ian

[Zac]

The free and paid versions are the same. The only difference is the number of knobs you can play with. The paid version offers a lot more chassis and engine parameters that you can tweak.

If I remember right you can use matlab to edit some of the parameters that are grayed out in the demo version. It's been a while since I've played with it so they might have "fixed" that by now.

Another program you could give a try is chassis sim. It does more or less the same thing, but the demo version is a full release on a 30 day timer.

[Silente]

i have tried to change some locked parameters from matlab but it doesn't seem to work. It'strange, because vehicle files have .mat estension, so changing them from matlab should have some effect.

I see the parameters in the workspace changing when i move them, but i don't see differences in lapsim.

I was hoping to have the possibility to move them a little.

Zac you were doing the same to change them?

[Zac]

yeah, that worked 3-4 years ago.

[BillCobb]

I'll post a LapSim Crutch on the TTC.

[BillCobb]

Look in the General TTC Concussion section. I suspect there is a more recent version, but it will take some Red Bull to find it...

[exFSAE]

Or in the time spent screwing around with this, could have had your own lap sim up and running. Promise it's not as hard as you think once you sit down and think it out.

[flavorPacket]

building the lapsim isn't the hard part. it's implementing it into something robust enough to believe and design around.

knowing what you want to get out of sim is the critical first step. Even using nonlinear tires may be overkill for doing some of the more basic analyses.

[exFSAE]

Originally posted by flavorPacket:
building the lapsim isn't the hard part. it's implementing it into something robust enough to believe and design around.

knowing what you want to get out of sim is the critical first step. Even using nonlinear tires may be overkill for doing some of the more basic analyses.

Definitely agree.

My opinion is that I'd rather have my own set of basic tools - in that I have a feel for how it works, what it's doing, and the magnitude of uncertainties I'll have in my output. Getting a license of ADAMS is a waste of time if you aren't comfortable with what it does and how, and if you don't have sufficient input data to really define the model. As I'm aware, a fair amount has to go into Bosch's tool.

Or, we can look at the question like this - Can screw around for weeks with Bosch trying to get it to work with your kind of car, and maybe at the end of it get some answer out which you have no validation against anyway.

OR, you can spend a couple days getting a rudimentary lap, corner, or straightaway sim together and have some questions answered quickly which can be immediately fed into the design process.

Which is going to have the most positive net impact down the road?

[Silente]

ExFsae what you say is correct but not everybody has so much experience in programming with matlab, for example.

I was able to do a simple sim, but i realise that to do a complete lap simulation i would have a lot of problem beacause of my lack in programming.

Moreover, in my steady state corner sim i find a solution iterating, but you need some time to find maximum grip. If you have to do it for each corner of a track...Not to say the rest of the track.

Maybe my concern is beacuse i am not so good in programming, but i think i would not need just a couple of days!

I still don't understand how it finds immediately apex speed corner!

[ibanezplayer]

Originally posted by Silente:
...i would have a lot of problem beacause of my lack in programming.

If it is hindering you that much surely your school has a computer/software engineering department where you could recruit some people.

[thewoundedsoldier]

It finds vehicle speed at the corner apex by very simple constant acceleration equations. It assumes the tires have trimmed out to steady state conditions.

The iterative process is extremely easy to build once you build your first cornering model. You can then use that one single corner model to create fake, random laps with several simulink tools, or you can accurately use it to recreate a lap based off of GPS data.

I think it is much easier to find maximum grip than to create the lap. At every point you should have an output capturing what the cars acceleration is in each of the x- y- and z- directions. You can use those to pretty easily calculate a rough estimate of what the tire can do at that exact point (assuming you believe in combined loading tire data haha).

Lastly, programming is hugely overrated in formula student for some reason. A lot of people at my school are afraid of things like matlab, cnc code, html, etc. Learning code is easier than learning anything else because it doesnt require much critical thinking--just plug and chug. (ie critically thinking about how you want to simulate your vehicle and/or lap is much more interesting than figuring out the exact tools with which to do it...i've seen awesome lapsims done completely in excel).

[BillCobb]

GuitarMan has it right on. A team should be composed of individuals who bring some specialty to the project. Having a Matlab wizard on a race team is as important as having a tire or engine specialist. There ought to be plenty of work to keep them busy. This is New School Racing theory. Old School is struggling...