Is multilink suspension advantageous over double wishbone?

[PlatinumZealot]

Think of the double wishbone as a 4 axis CNC machine.. The multi-link would be like 5 axis. (I am not saying to compare anyting mathematical here). They can both do the same job but the 5 axis will give you that extra layer of refinement and delicacy.

The BMW seven series is a good example. BMW could have put a solid A-arm on the bottom of the upright, but they chose the "double arm" for this reason... (I am still looking for a more technical explanation)

The double-arm front axle ensures wheels grip the road tightly. The suspension reacts quickly to absorb any uneven surfaces while the wheels follow every steering movement exactly. The aluminium structure eliminates unnecessary weight while increasing rigidity, and in the case of a collision, it absorbs the forces of the impact to provide maximum passenger protection. Sporty manoeuvres can be carried out smoothly in every driving situation without any unnecessary jolting of passengers.

I think there is a bit more to learn, don't want to just sweep it under the carpet.

[Caito]

5-link rear suspension. 1970 f1 car.

http://atspeedimages.com/image.php/4c9a ... ension.jpg

To me that looks more like a double a arm with toe link.

[Caito]

Again.. a double a-arm is just a specific implementation of a multi-link suspension.

In that case, all suspension are, in some way.

Not quite true. I believe multi-link only refers to independent suspensions, and there are quite a few types of suspensions that do not qualify as such, for example beam axles.

If we're talking about independent suspensions, unless we start naming sliders, we could say that they're all 5link specifics.

[munks]

If we're talking about independent suspensions, unless we start naming sliders, we could say that they're all 5link specifics.

Basically. As you said earlier, you have to restrict 5 DOF, which 5 links do. But sliders and even MacPherson struts can't be represented with 5 links of finite length, IIRC.

BTW, my interpretation of multi-link only referring to independent suspensions may not necessarily be correct, although the Wikipedia page on multi-link suspensions seems to back it up.

[Jersey Tom]

Was thinking about this a bit last night. One potential advantage of a pure 5-link suspension is that it's easy to change things pretty dramatically. Pretty easy to make links of various lengths precise, certainly easier than fabricating a full control arm.

Even if one were to use the links to create virtual control arms with pairs of links being coplanar, it gives some interesting options for adjusting jacking coefficients, camber curves, etc. You could even take a car that's currently set up for a double a-arm suspension (FSAE car is an easy example) and easily swap it over to a 5-link design just with a different upright.

Ultimately, analysis of a completely open 5-link suspension isn't trivial. But, by constraining it down to effectively work as a double wishbone, you do get those adjustment options.

Maybe I'll do a blog entry about it later today.

[Caito]

What I read somewhere is that 5-link lets you tune parameters "independently" which I haven't tested in any suspension software.

In a double wishbone you alter many parameters at the same time. You can't just change a camber curve without altering other stuff.

[Jersey Tom]

That's certainly possible. Personally I've never done 5-link design, but without the constraint of having two pair of coincident points.. I would imagine that there is more freedom in having less coupling between parameters.

I'll be honest, I'm not even entirely sure how I'd go about doing it.

So then, what is going to get you the most performance at the end of a design cycle? The suspension that's maybe not as free, but easy to tweak and refine.. or the one with all the freedom that's a real challenge to refine.

[PlatinumZealot]

So if multi-link was allowed in F1 do you think teams would use it?

[Jersey Tom]

Maybe. Something to consider is.. just how much advantage is that extra kinematic freedom going to get you on such an aero-dominant car?

What's the effect of having 5 control links all adding to the frontal area of the car.. as compared to two sleek, aerodynamically efficient control arms with just one push/pull-rod?

[thisisatest]

5-link rear suspension. 1970 f1 car.

http://atspeedimages.com/image.php/4c9a ... ension.jpg

To me that looks more like a double a arm with toe link.

http://atspeedimages.com/image.php/4c92 ... _pipes.jpg
here you can see the two trailing links. combined with the one upper and two lower lateral links, you have a 5-link.
the trailing links can be tuned to give you anti-squat, anti-dive, and have it decrease or increase with suspension travel. the three lateral links do the usual camber control along with toe control.
i used to have an R/C car with a 5link rear suspension.
http://www.apexspeed.com/doug/rc/vintag ... x2_005.jpg
the company then went back to double wishbone. seemed better to be able to maximize what you can really grasp than have more ways to have the setup wrong.

edit- forgot to post the second photo of the f1 car

[PlatinumZealot]

Today with all the improved computer technology, I can imagine a multi-link car running Free practice 3, the engineers gathering data and then the computer iterating the lengths of each linkage that will improve the laptimes, as the car comes in for the final stint the engineers change one of the links to a shorter one and send the car back out on its way to a faster lap time...
As I understand it right now, that the teams usually arrive at the geometrical suspension settings quite fast (correct me If I am wrong). Could the on track setup disadvantages of the multilink be negated today with the advances computer technology?

[riff_raff]

In terms of kinematics, the typical multi-link (5 link) suspension on a production car is a far more complex analytical case than the double wishbone on an F1 car. While more links and nodal points theoretically gives the 5-link configuration more adjustability, it also makes the set-up far more complicated.

In order for any suspension linkage to function properly, its kinematic solution must be determinate in nature. This means that it cannot have any unconstrained kinematic DOF's. Analyzing the typical automotive 5-link suspension with elastomeric bushings, is incredibly complex. Especially due to the non-linear spring rates in the elastomeric joints.

Even with the massive digital computational capabilities currently available to most F1 teams, resolving for multiple non-linear (ie. aero plus structure) conditions simultaneously is still not practical.

riff_raff

[PlatinumZealot]

Making some of the linkages have rigid joints should reduce complexity. You can reduce the number of linkages from 5 to 4 as well. Only a few street cars have the full five links.

This might be possible on the rear suspension, since teams use spherical joints on the lower suspension arms. You keep the flexture joint on the top inner mount for aero reasons, and use spherical joints for the other joints.

For the front suspension, I guess you would want to keep the flextures, but flextures are designed to rotate one axis only so that would not work for all the joints (only one joint can be allowed one DOF with 4 linkages). So, for the front suspension one flexture type control arm, preferably at the bottom, and the top has two rose joint linkages plus steering rod. A four linkage multi-link.

What do you think?

[Jersey Tom]

Lofty ideas, but impractical.

Let's assume that we limit ourselves to changing 4 link lengths front, and 4 rear.. and three options for each (standard, incrementally longer, and incrementally shorter).

That's 3^8 combinations for a full factorial DOE, = 6561 combinations. If we're doing a full dynamic lap sim with a model with decent fidelity, let's assume we can run at about real time.. and we're looking at a 1'30" lap. We'll even assume that you have distributed computing and 8 machines to split the workload up on.

6561 combinations, 90 seconds each, over 8 solution centers... 20.5 hours. Effectively a full DAY to get an answer just for a handful of links.. before touching any additional effect of springs, bars, camber, whatever.

The idea of just letting a computer sim play with parameters to find a solution by the "shotgun approach" of some big DOE, is frequently impractical.

[PlatinumZealot]

Are you sure the simulation is that long? I think you would "fast forward" the race laps? Or do you mean the lap simulations on the post rig? I am not familiar with full on suspension simulation.

Nonetheless, I can't remember the last time I saw a front control arm change in the pits? Wouldn't your front control arm geometry be sorted before you reach the race track?