Lateral load transfer

[Ayu]

Hello,

My name is Ayush Maheshwari. I am in suspension department of my fsae team.My experience with suspension dept. is of 3 months. I have a conceptual doubt regarding total lateral load transfer.
According to fundamental of vehicle dynamics by thomas gillespie, lateral load transfer arises from two mechanism:1. cornering forces 2. body roll which are independent of each other.
Hence TLLT should be = 2(Fy X Hr / t) + 2(K x phi / t)

But according to RCVD by milliken section 18.4 :
TLLT = (W X Ay x h / t) and it hasn't taken the effect of body roll in the equation. But while calculating distribution of lateral load transfer in front and rear.

My query is that whether :
1. we should take the effect of body roll while calculating TLLT?
2. do the effect of body roll is so small that we can neglect it?
3. Do the effect of body roll is included in the formula (W X Ay x h / t) which means cornering force and body roll is dependent of each other which is against of what written in fundamental of vehicle dynamics by thomas gillespie.

According to the data of our car:
For the case of Ay = 1.5g
I calculated the lateral load transfer of front and rear separately using equation written in milliken (on page 682) and for front it comes out to be 442.7804N and for rear 528.773N total of front and rear comes out to be 971.554 and when i calculated the total lateral load transfer equation (W X Ay x h / t) it comes out 976.8855N and there is difference of 5N.

Is this 5N is the effect of body roll that i am not considering while calculating TLLT equation(W X Ay x h / t)?

My other query is why the effect of cornering force(inertial force) and body roll is independent from each other(according to as they both are the cause of lateral force(fundamental of vehicle dynamics by thomas gillespie).

[Claude Rouelle]

Ayush,


Welcome to the craziness. 3 months already into FSAE? And still enthusiastic? Don't worry that will change :)

1. Which FSAE team?

2. What is Hr and phi? I won't take the time to check myself in the book so I guess you should state your question in a more define way

3. Unless I misunderstand your question, the body roll is a consequence of the lateral acceleration and the tire cornering forces.

A. If you have the same front / total roll stiffness distribution and your front and rear roll stiffness are constant whatever roll angle you have, you will have the same front and rear weight transfer for a given lateral acceleration, whether you have a roll gradient of 0.2 deg/G or 2.0 deg/G. The reality is that you will most always have discrepancies between calculation and reality in your front and roll stiffness for example because your motion ratios are not constant due for example to unavoidable compliance.

B. BUT do not forget that even if the dynamic tire Fz would be the same whether you have a soft or a hard roll stiffness, the camber induced by the roll won't be the same and that could change both your lateral grip and your balance (yaw moment). By the way do really think it is necessary to expose in your questions forces with an accuracy of 1/10000 of a N?

C. However I guarantee you that in transient the weight transfer WILL be different.

[Ayu]

Sir,
Thank you for replying..
I am from Thapar University FSAE team,India
Hr here is the CG Height and phi is roll gradient; t is the average track width of car; Fy is lateral force(sprung mass * lateral accn(g's)
If cornering force and body roll are "cause and effect " of each other , then why it is written in the books that lateral load transfer due to cornering force is independent of roll angle of body?
A. According to me if a car takes a turn then sprung mass inertial force or centrifugal force will produce lateral load transfer due to CG Height distance from ground in combination with the body roll due to the torque produced by moment arm and cornering force. Hence I feel that body roll is dependent on lateral load transfer due to cornering force. I am having problem while finding Total Lateral Load Transfer (TLLT) of car, should i use 2(Fy X Hr / t) + 2(K x phi / t) as specified in gillespie or ( Fy X Hr / t) as specified in millken? Please help me on this?

B. We have use the TLLT Distribution in front and rear 45:55. After calculating TLLT (after clearing above doubt) I will calculate FLT/Ay(Front Lateral Force per lateral acceleration) = 0.45 X TLLT. Further I will calculate front roll stiffness using ((FLT/Ay x front trackwidth)-((sprung mass x sprung mass cg heigth) + (unsprung mass x unsprung mass cg height))). Is this the correct way to calculate front roll stiffness?

B. Our car has different roll stiffness of front and rear.To remove discrepancy between reality and theoretical values we will finalize our stiffness values after dynamic testing and calculating damper travel using Linear Potentiometer. Are we on right way? Please throw some light.

Thank you

[Claude Rouelle]

Ayush,

A. You are mixing 3 things: roll angle lateral force and weight transfer. More lateral G = more roll. Same lateral G but softer suspension = more roll but if everything is linear same weight transfer. Independently of what RCVD or Gillepsie says what do YOU think is right? What is the part you do not understand?

B1. Way to simplified. You need to get deeper in the available resources books and this forum before you ask this kind of questions. You miss at least a basic tire model here
B2. Good luck to figure out where the front and the rear suspended mass CG and CG heights are.....

C. Your linear potentiometers will give you an approximate idea of the forces coming from the springs (approximate because spring stiffness are never constant and there is always compliance) but it doesn't take into account your ARBs and your dampers. Of course you can calculate that but it won't be accurate. I have compared those calculations with the the ones coming from strain gauges on pushrods and they are quite different, especially in transient.

Note 1 Remember in VD or in racing you work in DELTA, There are too many parameters and too many parameters innacuracy to be spot on in absolute values

Note 2 Your car kinematics does not seem to have been taken into account: whether you use te n-lines theory or the roll center height (kinematics of force based) your front and suspended mass weight transfer will be influenced by your suspension kinematics design.

[Z]

Are we on right way? Please throw some light.

Ayu,

No, you are on the WRONG WAY.

There are many mistakes in above three posts, including Claude's. (Edit: Claude just posted, and I haven't read his second post yet...). Here are some of them, and imagine a big red pen with many "!!!XXX!!!"s...


... According to fundamental of vehicle dynamics by thomas gillespie ... TLLT should be = 2(Fy X Hr / t) + 2(K x phi / t).

No, that is NOT what Gillespie says! READ THE TEXT!!!
~o0o~


... I calculated the lateral load transfer ... it comes out to be 442.7804N ... 528.773N ... 971.554 ... 976.8855N

Groooaannn!!! You are supposed to be doing Engineering calculations, not working out India's national debt!
~o0o~


(In Claude's post...)
A. If you have the same front / total roll stiffness distribution and your front and rear roll stiffness are constant whatever roll angle you have, you will have the same front and rear weight transfer for a given lateral acceleration, whether you have a roll gradient of 0.2 deg/G or 2.0 deg/G.

Wrong!!! Claude, you should know better.
~o0o~


If cornering force and body roll are "cause and effect " of each other ...

How can two things be BOTH cause AND effect of each other? Stupid!!!
~o0o~

Ayu, the two biggest mistakes you are making are these:

1. You are browsing some Vehicle Dynamics texts, pulling some equations from said texts, and then using said equations to calculate numbers to SEVEN significant digits (!). But you are making NO EFFORT TO UNDERSTAND what those equations represent, or how they are derived.

2. You are relying on Vehicle Dynamics texts to help you! Sadly, these do a very poor job of explaining Vehicle Dynamics.

So... what to do?

Your FSAE car is a VERY SIMPLE mechanical gadget. It is just a bunch of massive-links, and joints, and springs, and stuff... Your best chance of understanding it, and making it work well, is to simply apply the principles of age-old "Classical Mechanics" to it, starting from first principles. To learn these principles I suggest you find a very old book on Mechanics.

To restress the above, namely that modern VD teaching is a failure, neither of the two VD books mentioned above include a factor that is a "significant" one in Lateral Load Transfer calculations (*). I guess that Claude's Optimum-? software also misses it. I have mentioned this factor before, shown how easy it is to calculate its influence, and thus shown just how significant it can be in any given situation.

(* The wind blowing a downy-feather onto your nose = "IN-significant", while getting hit in the face with a shovel = "VERY significant". The above factor is somewhere between these two extremes. So, roughly between 1% -> 100% of the normally calculated TLLT. And thus good justification for calling above seven-significant-digits STUPID! :))

Z

[Tim.Wright]

How can two things be BOTH cause AND effect of each other? Stupid!!!


Mechanical feedback?

Slip angle causes lateral force which causes slip angle which causes lateral force which causes slip angle.

Lateral force causes compliance steer which causes lateral force which causes compliance steer which causes lateral force.

Bumpsteer causes lateral acceleration which causes bumpsteer which causes lateral acceleration which causes bumpsteer.

Yawrate imposes slip angles on the front/rear axles which changes the yaw rate which changes the front/rear slip angles which changes the yawrate.

If you draw up a control systems style block diagram of the lateral dynamics of a vehicle you will see several such feedback loops which do no have a defined start or end point. In these loops there is no distinction between cause and effect.

[Z]

... Remember in VD or in racing you work in DELTA, There are too many parameters and too many parameters innacuracy to be spot on in absolute values.

Claude,

Your above comment is deeply troubling.

On another similar and recent thread, the 40+ year old "James Bond Spiral Jump" was said to be simulated with a punched-card Fortran program, and then carried out perfectly FIRST TIME (no doubt pleasing the stunt driver!).

http://photos1.blogger.com/blogger/7865/1286/1600/amcII.jpg

So, can you please explain why you think much simpler things are impossible today?

That is, why is "VD/racing" analysis so shonky these days that you say we cannot expect reasonably accurate "absolute values"? And what would Prof. Walter Lewin say about your approach to calculations, specifically regarding the issue of "accuracy"?

Most importantly, is promoting the acceptance of such third-rate standards to today's Engineering students a good thing to do?
~~~~~o0o~~~~~

Tim,


In these loops there is no distinction between cause and effect.

All your examples above describe "causal chains" (ie. "A, which causes B, which causes C, ..."). Of utmost importance in understanding such chains (or better yet, "causal nets") is that THE "CAUSE" MUST COME BEFORE ITS "EFFECT".

Notwithstanding some interpretations of QM theory, anyone trying to make sense of the real world, say an FS-student trying to figure out how to make their car go faster around a corner, really must have an appreciation of this temporal order. If, say, said student is running a simulation of their car and they want to alter the Yaw-rate at a given time "t", then they MUST make changes to the "causal" stuff, perhaps the front-STEER-angle, at a time "t MINUS dt".

The current fashion of conflating "cause and effect" and saying that they are "the same things, indistinguishable from each other, blah, blah..." is yet another step in the downward spiral of the modern education system. It is the lazy (and stupid!) teachers' excuse for not bothering to teach the fundamental importance, and difference, of the variable "t" in their cogitatio caeca equations. (Tech note: The equal sign ("=") used in modern algebraic analysis gives NO indication of the direction cause => effect. It should.)

In short, too many modern teachers have never learnt the difference between cause and effect, so they teach "...there is no difference".

Z

[DougMilliken]

... On another similar and recent thread, the 40+ year old "James Bond Spiral Jump" was said to be simulated with a punched-card Fortran program, and then carried out perfectly FIRST TIME (no doubt pleasing the stunt driver!).

Z -- Two comments:

1. While it's true that the spiral jump was developed using Fortran on punch cards, the work was done by a master at what was called "math modeling" in those days. Ray McHenry brought his years of experience in modeling pre & post accident situations to the development of the jump.

2. The jump that was filmed in Thailand for the Bond movie was the first jump on that set of ramps. But the stunt driver, Bumps Willard, had plenty of prior experience on the original set of ramps that toured with the All American Thrill Show for several years prior.
He wasn't the first driver either, there was at least one other who did the initial tests, after the car had gone over the jump under automatic control a couple of times.

Edit -- just came to my attention that there are demo videos showing the McHenry's latest software.
This is for use in reconstructing accidents (licensed to law firms and used in court) -- it does not have the detail of a full-on model for race car dynamic simulation -- but it captures the behavior over rough terrain and from impacts quite realistically.
https://www.youtube.com/watch?v=1qWwej-AQWQ&list=PLBh3OTTiCozPjBKb8gPV8DWDPG_sv_AR4
All strung together there are about 15 minutes of demos, the first one is only five minutes.

[Ahmad Rezq]

Ayu

I think what you miss is to define the system you are studying and its boundaries, differentiate between system external and internal loads.

- The equation you mentioned from Milliken is the force and moments analysis if we compress the whole car into a single axle ( as expressed from Milliken ) which has a CG where the car weight and centrifugal force act and the reactions at tires.

- In Gillespie: The system was the car axle (suspension) (Free body diagram, free the body and define the external forces acting on it) the forces on the system are reactions at tires the body roll action on springs. Gillespie wrote the moments equation at the rolling center.

both equations are correct. ( i.e Milliken and Gillespie ) but your written Gillepsie equation is not.

Fzo-Fzi=2Fy*hr/t+2K*fi/t

Fzo=Load on outside wheel
Fzi=Load on inside wheel
Fy=Lateral force
hr=Rolling center height
t=Tack width
k=Suspension roll stiffness
fi=Body roll angle