strong impulses when actuating joints

A place to discuss everything related to Newton Dynamics.

Moderators: Sascha Willems, walaber

Re: strong impulses when actuating joints

Postby Julio Jerez » Thu Oct 18, 2018 11:42 am

do you mean you are trying to model a suspension without a damper?
if so how are you doing that, with a joint?
or by adding. f=-k * x. as external forces to the bodies.

if the later is the case them I can guarantee you the the source of the problem, when you said that doing sub step increases the vibration a light click on me right away and tells me there is a heigen value in the system that is too close to one and making the system response diverge.
please tell me how are you making the suspension.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: strong impulses when actuating joints

Postby blackbird_dream » Fri Oct 19, 2018 2:03 am

No. The vertical force is 2nd order vs deflection with coefficients fitted from data.
Fz1=-(k1*dx^2+k2*dx)
The damping is viscous i.e. proportional to vertical velocity.
Fz2=-C*dv
The external force applied to the RB wheel is then Fz1+Fz2+ Self aligning moment+ overturning moment + lateral force.
I meant I do not use additional damping proposed in newtonbody editor. It's not suspension. There is no suspension on this type of vehicle. It's tire's softness and damping
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: strong impulses when actuating joints

Postby Julio Jerez » Fri Oct 19, 2018 6:51 am

blackbird_dream wrote:No. The vertical force is 2nd order vs deflection with coefficients fitted from data.
Fz1=-(k1*dx^2+k2*dx)
The damping is viscous i.e. proportional to vertical velocity.
Fz2=-C*dv


why does is use a K1*dx, I assume that that try to model teh spring non linearity and that k1 is extremely small while k2 is a very large spring in newton/m constant, but in any case the equation of a spring system when applied to a collsion of rigid bodies does no not behave have liek that simple equation that is for an simple mass
here you have the spring acting at soem arbitration point and there are more than one.

applying the force as if they were independent lead the that kind of behavior you see. for many reason, 1 is the will be numerally unstable, as the time step is smaller. 2 is also incorrect.

I can give you at more stable and acurate way to apply the spring force that converge to the exact result in the limit but that is does the best possible job on discrete time step. and it is still the correct spring system.

Many papers and many books is tire dynamic have that wrong whne they speak of the sprung mass, there si no such thing as sprung mass of weigh transfer of is a continue simulation thsoe ternm are for what is call state state, which is nothing by a point in a dynamics system where system is stable.
teh do that because forme there the bevavior can be linearized and the behaviodes is lionear for a small deviation form that point.

basicall the only time you equation is correct is whn ethe chassi and the tire is a rest. and is moves by a very small amoping forn teh equailrium point, but since the constact of you system are so extreme, a deviation form teh equilbrium point is tiny.

to fix that you need to apply the force and tiem t + dt liek this

Fz(t + dt) =-(k1*(dx + ddx)^2+k2*(dx + ddx) -C*(dv + ddv)[/quote]

but that requireres solve a system of equation.

I can give you the derivation and code that you can use if you wnat then the spring will alway be statbel at any rate and avoid that vibrarion you see, there is code in newton that does that already.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: strong impulses when actuating joints

Postby blackbird_dream » Fri Oct 19, 2018 7:46 am

the load/deflection curve of the tire alone is non linear. 2nd order as 1st approximation.
It's not about collision but road/wheel interaction.
k1=30110824
k2=618309
I agree that it may not be accurate but this is second order in terms of vertical oscillation. We can check the natural mode is correct i.e. around 4Hz. I don't need more precision.
there should be vibration. The load here is vey heavy and the lever arm (distance from the load CoG and the front wheel centre) huge.
The rest (lateral force and moments) is only Pacejka approach.
Anyway any suggestion is good, if you think I can make better, I'd b glad to read it
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Re: strong impulses when actuating joints

Postby blackbird_dream » Fri Oct 19, 2018 8:12 am

Yes I'd like to know more about the way to model the load/deflection law:
f=-k1x-k2x^2-cv
and its derivative
User avatar
blackbird_dream
 
Posts: 354
Joined: Wed Jun 07, 2006 3:08 pm
Location: France

Previous

Return to General Discussion

Who is online

Users browsing this forum: No registered users and 16 guests

cron