15 replies to this topic

[Greg Locock]

OK, heres the way it works

1) totally insensitive to shocks

2) not very sensitive to springs, sta bar

3) not sensitive to suspenson geometry

4) doesn't use ride heights to work out aerodynamics

5) not very sensitive to CGZ

so... I think it is using a bicycle model - ie weight transfer is not important

As the manual states - it is quasi static so shocks dont matter.

OTOH it is good at - changes in weight,mu,downforce, drag, power and gearing.

I like the way it works out laptimes

[Ben]

Someone with some time to interogate it.

At the moment my LTS is based on a particle model and a g-g-V surface. The surface is defined using the 'simplified cornering with aerodynamic downforce' section from milliken and a basic tractive effort curve.

I did it more to prove the program architecture but it's a fairly good first order look at fundamental things like power:weight, peak mu, and Cz:Cx.

I'm working on points 2 - 4 with a full vehicle and tyre model. I'm keeping it steady state so dampers aren't in the picture. I have ADAMS for that

Ben

[Greg Locock]

yeah, i've got ADAMS for that as well.

Having said that project number 17b is to try and model our ride and handling circuit and see when the electronic stability control kicks in.

At the moment the car just drives round in a spiral, so I think I need to add a throttle controller.

Hey, I like Lapsim, I guess I'm just disappointed that it doesn't cover weight transfer

[Ben]

Originally posted by Greg Locock


Hey, I like Lapsim, I guess I'm just disappointed that it doesn't cover weight transfer

I'm not disappointed, just baffled. Seeing as adjustment of the lateral load transfer distribution is one of the most fundamental tuning methods in racing.

Ben

[Fat Boy]

I think you guys might be selling it a little short. It does cover lateral load transfer. Some of the data that can be displayed after running a sim is roll, pitch, vertical tire load, lateral weight transfer, etc. I've played around with it for a little bit now, and it's quite close to measured data. I suppose it's all a question of how good your model is, but with just a little bit of work it seems to produce quite accurate results.

I spent a little time talking to the man who wrote the code. He reckons for simple tasks like sorting out gears or doing a downforce/laptime sweep, there's really no need to get the expert software. He also said if you are working with a non-aero car, that there is probably no need to pay for the expert version. I was suprised to hear him say this.

In the 'Help' menu there is an option to 'show EXPERT mode'. It lets you see the things that can be tweeked in the pay version without being able to actually tweek it. Anyone who is spending some time on the sim would probably want to check this out.

As far as dampers go, the software guy doesn't believe that anyone can accurately simulate damper reactions (I assume he meant outside of ADAMS), so he says if he can't do it right, he doesn't bother.

He also doesn't recommend using Pacejka tire data for the model. He says that while good for street tires, it is meaningless for race tires in limit handling situations. The recommendation was to get the car model right and then back into the tire model until the simulation runs look good. I've found that the F3 tire with minor grip changes seems to work well for other single seaters.

Anyone have other comments?

Fat Boy

[Lukin]

Don't ever say I'm slow to respond.....

The 2005 Lapsim is out now, with the small engine simulator described in March Racecar Engineering.

Has anyone input data to try and generate a track map? I have MoTeC data required to generate a track map (wheelspeed, lat G) but I can't seem to generate a track map.

With the map generated I'm thinking I should be able to input the basic FF setup/characteristics such as chassis and engine etc into the F3 (as Fat Boy said) car model, and 'tweak' the grip characteristics until the simulated results (mid corner & straight speeds, max lateral and long g) closely resemble the data.

Could it work? If the answer is yes. Can an idiot like me do it?

[Ben]

Originally posted by Lukin
Don't ever say I'm slow to respond.....

The 2005 Lapsim is out now, with the small engine simulator described in March Racecar Engineering.

Has anyone input data to try and generate a track map? I have MoTeC data required to generate a track map (wheelspeed, lat G) but I can't seem to generate a track map.

With the map generated I'm thinking I should be able to input the basic FF setup/characteristics such as chassis and engine etc into the F3 (as Fat Boy said) car model, and 'tweak' the grip characteristics until the simulated results (mid corner & straight speeds, max lateral and long g) closely resemble the data.

Could it work? If the answer is yes. Can an idiot like me do it?

I'll try with some data tonight, but is it a problem with the format or that the track looks wrong in lapsim? I've used some of the data you sent me and generated a track map in Matlab that looks pretty close to the Motec Interpreter map.

Ben

[Lukin]

I can't seem to generate a track map. I have tried opening the CSV file in Lapsim, matched the channels from the data to the channel in Lapsim, but it just doesn't seem to work. I think it's almost there.

[gary76]

Since the evenings are getting dark and the garage is cold I decided to down load the Bosch LapSim freebie and have a play. Problem straight away, I could not make an engine file to read 13000 rpm. It appears the freebie limits the max. engine rpm to 10000. Anybody got any idea how I might extend the engine speeds? There must be a trick somebody has found! I am trying to analyse a simple 4 wheel car with a motorcycle engine. Any help would be appreciated, thanks.

[Greg Locock]

Well, you could try running an 8500 rpm engine, and then scale the torque and axle ratio to suit.

Actually I couldn't get the latest version to work AND it trashed the old version that did. grr

[ben38]

For the maps, I tried to make a copy of the existing tracks file and edit them to change. Even that did not work. The free version seems pretty well locked on that point.
Maybe older version allowed that but the current just doesn't seem to.

I used LapSim for the 2 race cars i have this year.
The manufacturer doesn't give enough data along with the car (ie no engine curve and no aerodynamic values).
Playing with my data and adjusting the results with LapSim first with the straight line simulator proved quite good in the end.
I get a whole set of values that surprised the engine guy and the aero guy enough to admit that they were very very close to the real values.
The power/aero model used is quite good and very usefull even in the free configuration.
Considering the lap time simulator, i tried last year with the values of the Proto on Silverstone. The results were not so bad.
I used at start the Pacejka model for tires at start then modified it a bit to finaly get very close to my data.
But some set up changes did not have effect as said before.
Another point which was not that accurate was the differential. In the software stiffer and stiffer was only gains while in reality not really.... Does it have problems to simulate it as the soft doesn't include bumps? Maybe.

Overall I find it usefull with aero/power simulation, top speed, gear ratios and this kind of stuff.
But for the real balance of the car maybe i'd take its information carefully

[Fat Boy]

Originally posted by Greg Locock
OK, heres the way it works

1) totally insensitive to shocks

2) not very sensitive to springs, sta bar

3) not sensitive to suspenson geometry

4) doesn't use ride heights to work out aerodynamics

5) not very sensitive to CGZ

so... I think it is using a bicycle model - ie weight transfer is not important

As the manual states - it is quasi static so shocks dont matter.

OTOH it is good at - changes in weight,mu,downforce, drag, power and gearing.

I like the way it works out laptimes

Holy old thread, Batman.

1: Yep.
2: This is a function of the tire model. If the tire model is more sensitive to load, so will the car model.
3: Pretty true. Being a quasi-static steady state model means no proper 'transients'
4: The version that you buy does.
5: relates back to tire model.

This actually uses a 4 wheel model, not just a bicycle. I played with the free version and then spent a good chunk of time on the buy version. It's a pretty good piece of kit for the money. There are plenty of quirks, but it has a good bang for the buck. If you don't change the aerodynamics of the car much, it is pretty good at going from track to track for gearing, laptime, corner speeds, etc. Coming up with an entirely new aeromap for each wing setting is a bit tedious.

I have spent some time now working on a fully dynamic sim. They're definitely a different beast. A lot of the ideas transfer over, but the ability to simulate curb strikes,driving line changes, pavement types, a real driver, etc. makes a pretty profound difference. The big deal with a quasi-static model is that it will always be on the perfect line and using the absolute limit of the tire. It never picks up push, gets wheelspin, or goes loose (which is why the diff isn't such a big deal).

I don't think Lapsim is something that you can use to effectively set up a car with, but if you acknowledge it's limitations and use it's strengths, it's a pretty good deal.

[Greg Locock]

I seem to have come over as unduly negative in my first post - I like LapSim, and have used it ever since installing it, not least as a double check on other calculations.

The tire model with the old freebie is a bit odd so far as load sensitivity goes, with a more realistic model it is much better.

Circuit simulation actually something I'd like to do more of, but the chances of me ever needing to develop a true driver model are about zero.

One thing I've been doing a lot more of lately is feeding the recorded steering wheel and speed into a model and then comparing the real and simulated trajectory of the car. That is a good test of tire models! (and a good test of the instrumentation setup in the car if they aren't using GPS)

[desmo]

If they aren't using GPS?

[Greg Locock]

? What do you mean?

[gary76]

I have some on-car-recorded data from a DigiDash data logging system, this recorded at Snetterton and the Silverstone National circuits in England. Following the instruction I have received fro ETB Ltd (the DigiDash manufacturer) I saved and 'exported' one lap in a "csv" as instructed. This data is generated by 'G' sensor data I understand. The "gear" column in the "csv" data was deleated as DigiDash information was that it sometimes gives a problem in LapSim! This "csv" was saved in a Folder (I made) titled "Data-recorded" on the LapSim Program.

Open the LapSim program, enter a 'car-setup' data set, then 'Load Data File' which brings up a page titled 'Load WinDarab Data File', from this page I select the 'Data-recorded' folder I made and from this folder I enter the "csv" file I have recorded from the car. I get a message on the LapSim page saying " No Set-Up data found in selected data file", "Select setup file to combine with on car recorded data in order to process".
I re enter the ' car setup' file and a page comes up "Load WinDarab Data file" with what I think is a listing of the 'Channel names' as shown on page 7 of the V2004 manual. In all the entry boxes the words "no signal" is displayed and I cannot enter any data into these boxes. There is also a note saying " Select long and lat acceln+ at least one wheel speed". The program also 'freezes' and I cannot continue. I have to switch off and restart!

What am I doing wrong? I have read through the manual and there is one section which I think could be my problem area but I cannot seem to find how I can address them. This is the Step III: Specifying channel names. Section. I cannot seem to find a page to enter this data. Maybe one of users could give me a clue where I am going wrong.
I note that there was a similar question in this thread, did you resolve it?