Braking during cornering

[simulator78]

Hi,

I am working on a driving dynamics simulation and have a problem with braking during cornering. I don't have automotive engineering background so obviously I am missing something. Here is the story:

If I brake during a turn, turn rate increases considerably. E.g. when driving on highway at 130kmh in a left turn, if I try to brake gently (so no hard braking, no wheel locking) to 100kmh the car pulls towards left and before I can compensate the car changes lane.

I double checked the tire model implementation and the forces&torques seem to be ok. I couldn't locate any bugs. In other words, it looks like what I am experiencing is correct behavior .

With a real car this does not happen. The simulation model does not include ESP but braking on a highway should be possible without it too. Could anyone give an idea?

Thanks

 

[simulator78]

Some extra information:

The brake I am talking about is not a hard-brake and I am not sure that I am reaching the traction limits. I was also careful to implement a forward bias in braking. Some numbers might clarify:

Weight of car: 1560kg
speed : 130kmh
front wheel deflection: 0.35 degrees
applied front brake torque: 440 Nm
applied back brake torque: 240 Nm
turn rate w/o brakes: 3.1 deg/sec
turn rate after braking: 4.6 deg/sec
deceleration 0.28g

 

[MikeHalloran]

Does your model include body roll and roll steer?




Mike Halloran
Pembroke Pines, FL, USA

 

[simulator78]

Hi Mike,

Yes body roll is included. On the other hand roll steer is not implemented. I started to read about roll steer and it might just be the thing that I am missing. But I am not sure whether I understand the concept. Is the following scenario correct?:

We drive 120 kmh on a highway and turn towards left. As a result a slight right roll is present. At some point we start to brake hence the body rolls a little bit more to right and nose pitches down.

What I observe at this point is an increased turn rate towards left due to increased lateral forces on front and back tires (front laterals increase more). But now with roll steering it should be different.

Due to roll steering, more compressed front right suspension turns the right front wheel slightly to right. Left front is a little bit complex because increased right roll reduces the suspension compression whereas nose pitch down increases it. Still, in total we have to get less steering towards the left and turn rate should remain more or less constant.

 

[MikeHalloran]

At this point, you can either add a bunch of empirically derived cross- coupling coefficients to your model, or just go ahead and model the suspension linkage of whatever vehicle you're modeling.

To be accurate, you'd need to include spring rates in three directions at every rubber bushing, too.

Either way, the complexity of your model goes up.

My favorite example of intentional roll steer is the later Corvair's rear suspension, which had a fore/aft radius arm locating the rear spindle, and a small lateral bar near its front bushing that toed the outside wheel in as the car rolled.





Mike Halloran
Pembroke Pines, FL, USA

 

[GregLocock]

what is your latacc in this event?

what is your cgx location?what is your rolling radius? wheelbase? track? cgz?

what tire model are you using? how are you modelling combined slip? how have you established the limits of the friction ellipse?

In a swept steer event at 120 kph what latacc can you achieve?

In a braking event what longacc can you achieve?

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[simulator78]

Thanks for the answers. I am looking to roll steering issue and in general to the suspension system which is not modeled with adequate detail at the moment.

I can't exclude an error in tire model but my first priority at the moment is getting something drivable as soon as possible .

 

[imcjoek]

I worked with computer simulation of cars a bit.

All of the tire models you could hope to run in real-time have flaws and/or band-aids to fix certain situations. (Try parking on a hill with a pacejka model!)

So I would suggest you output some debug slip angles, loads, side forces, etc and watch what they do. I would suspect your tires operate at a high-ish slip angle even under light lane-changes, and as a result, changing the load on that tire (braking to transfer weight) results in a significant side-force change. Meaning your tire slopes are kind of "squishy" [another technical term], often necessary to acheive stable numerical integration.

 

[simulator78]

imcjoek:

I actually spent quite a bit of time with debugging the model before sending here the question. I can't exclude a problem in the tire model with 100% certainty but it seemed ok to me. Probably we should isolate the tire model from the rest of the system and validate it using with real life data to be sure.

Just a question though. Were you able to model a car that has realistic handling without implementing roll steering etc. ? With which tire model (I am using TMeasy)?

 

[NormPeterson]

Have you looked at what the rear wheels are doing in terms of toe (or axle roll steer if it's a solid axle vehicle) during this time?

Does your model allow you to adjust the front vs rear tire pressure differential and if so, does that change things any?


Norm