Constant Longitudinal Velocity in Adams/Car J turn Manoeuvre for a mini-Baja model 2

[josh m]

I've been analyzing a vehicle and the Full vehicle analysis>Vehicle Set Up > Powertrain isn't active or it is hidden ...

in the J-Turn Test , I've been trying to make the longitudinal velocity constant during the j-turn manoeuvre ...

what i did was to make that happen by means of Event Builder ... but it doesn't happen exactly as it should ...

i chose the 10.5 m/s velocity as the longitudinal velocity but as you can see the vehicle's velocity does two things :

1- it lowers by a sudden wave from the velocity that it has been defined in Event Builder

2- i can't raise the longitudinal velocity because for one thing it gives an error as to the vehicle either rolls over or spins around or gives a halt error that stops the test , for another thing , no matter what the steering angle is , the longitudinal velocity cannot go beyond 10.5 m/s or it gives the same error which i think it's really odd and wrong and i'm almost sure that it's because of a software thing or the model issue or something else ...

the error is this :

---- START: ERROR ----
At time 2.990053083E+00 the integrator is unable to proceed. Possible Causes:
(1) The accuracy required for the numerical solution can not be attained.
Relax (increase) the value of the acceptable integration ERROR.
(2) Incompatible redundant constraints, a lock up, or a bifurcation
situation. The latter two indicate a mechanism design problem.
(3) The system includes a zero (or relatively small) mass on a part with
an unconstrained translational degree of freedom.
Make sure you have mass on all parts with translational degrees of freedom.
(4) The system includes a zero (or relatively small) inertia on a part
with an unconstrained rotational degree of freedom.
Make sure you have inertias on all parts with rotational degrees of freedom.
(5) An Adams element has a function expression that equals exactly itself.
For example,
SFORCE/id1,I=id2,J=id3,ROT,FUNC=SFORCE(id1,jflag,comp,rm),and
DIFF/id4,IMPLICIT,IC=0,FUNC=DIF1(id4).
Similarly,
VARIABLE/id5, FUNC=VARVAL(id5)*TIME
equals itself at 1 second. Avoid setting an Adams element equal to itself.


---- END: ERROR ----

Dynamic Solution stopped.


End Simulation

Simulate status=-124



---- START: ERROR ----
SIMULATE command failed for minimaneuver: J_TURN
Aborting Execution.

---- END: ERROR ----


as you know in the j-turn manoeuvre , the longitudinal velocity goes down during the j-turn mini-manoeuvre ... so if you want to do a steady-state test you have to fiddle with the Event Builder or the Engine Braking Torque or other resistance forces that make the vehicle's velocity go down ...

mind you , in the event builder i put the throttle option on the machine and maintain velocity and also in the static set up i change the task from straight to settle because the straight makes the velocity to change with unbalanced oscillations around the defined velocity during the test :

the mini-baja model is at :

http://simcompanion.mscsoftware.com/infocenter/index?page=content&id=KB8017706&cat=15M9BU&actp=LIST

appreciate your help in this ...

 

[BrianPetersen]

I know nothing of these simulations. But I do have real world driving experience.

If you attempt a real world maneuver like this, the vehicle slows down in the process of doing so because it's sliding.

It seems to me that your simulation is correctly telling you that what you're asking for, can't be done.

If the simulation correctly models real world physics, then telling it to violate the laws of physics will correctly result in "nope".

Why are you attempting to tell the model to violate the laws of physics?

 

[VEBill]

Speaking from experience, a 1972 Dodge Dart can do smooth J-turns, back-and-forth endlessly, with minimal loss of speed during the 180° half-spin.

Well, endlessly right up until the nice policeman politely asks me to stop.

Skinny tires with little grip are very helpful.

For clarity, start in Reverse, at max speed select Neutral, spin around, into Drive and end up going forward at about 30 mph.

 

[GregLocock]

1) Check that you are performing the J turn maneuver correctly. The spec I work to says the vehicle should be in neutral.

2) Let the car settle before entering the maneuver, you may be seeing a switch -on transient where the controller is sorting itself out. I'd run for 5 seconds at constant speed before ramping the SWA.

3) If you do need a constant speed according to your test spec, then you need to modify the speed controller's PID, or increase the power of your engine.

4) check the marker for the longitudinal velocity request, if you start to drift the car then its speed won't change much, you've just spun the car around its own axis , and so longitudinal speed isn't quite what you'd expect.So look at lateral velocity and mag velocity as well. Also on your plots include swa and latacc and yawv.

5) "thing it gives an error as to the vehicle either rolls over or spins around " yes, that's the end of a j turn! You may have some vehicle limits set which terminate the test.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[josh m]

thank you for responding so fast GregLocock

but as I said , no matter what the steering wheel angle would be , the longitudinal speed upper than 10.5 m/s makes the vehicle give the aforementioned error …

would that problem also be solved with speed controller's PID ??

is there a rational explanation for this happening ?

 

[GregLocock]

Look very carefully at the animation and see if you can see something odd just before the event fails. Switch eprint/debug on and see if you can understand the report (good luck). Try changing the integrator from SI1 to SI2 and switch mod on or off. try altering the error value. Sorry, debugging somebody else's model is always tricky.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[josh m]

thank you so much GregLocock

 

[josh m]

FYI BrianPetersen

i'm trying to do this because I need my model to perform this test in a steady-state condition … I can't achieve that goal with a variable longitudinal velocity ...

 

[BrianPetersen]

I can't help it that you are trying to simulate a violation of the laws of physics. I still say the simulation is telling you the right result ... "not possible".

 

[josh m]

I know BrianPetersen . you are right …

with the steering input there would be an acceleration in reverse direction of the vehicle moving direction … that equals the sine of the wheel steered angle multiplied by the lateral acceleration … you are completely right . Fy . Sin(alpha) . alpha : wheel angle Fy : lateral acceleration

that is if you say you would eliminate every other resistance forces ...

but what i'm trying to do is to create a steady-state condition by fiddling with the engine powertrain and throttle so it could neutralize that opposing acceleration …

I thought maybe I could stabilize the longitudinal velocity without touching the PID … just that

 

[enginesrus]

Shouldn't this thread be in the AUTOMOTIVE OTHER TOPICS section?

 

[GregLocock]

No, it should be in the ADAMS forum!

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?