They work. It is probably not feasible to calculate the loads by hand, but you can get an idea by bolting a steel beam from one spring tower to the other, and then jacking one side of the car up. Try and take the bolts out, and when you've done it, try and pull the beam hard enough to get them back in.

I've never done this, but would be very interested to see the results.

From work I've seen you'll need about 2 kg of metal in the brace (if it is straight) to make a reasonable difference.

One problem is that the improvements are in a host of driver-feel related things rather than any great /measurable/ improvement.

Cheers

Greg Locock

A larger issue than the strength of the bar in most cases is the method of attachment to the towers.  Many times the attachment is with undersized bolts in single shear, plus the attachment location in relation to the bar induces bending moments.  Many of the aftermarket bars sacrifice functionality for appearance.

A well designed bar can definitely  improve steering response in turn in.  Tying the towers together  minimises their relative motion which allows the springs shocks and antiroll bar to do their job more precisely.

In many street applications there is a minimal amount of room to properly tie the towers together without major surgery, so compromises are necessary.  But as
Greg mentioned above you can do some tests.  
You can quantify how flexable a particular vehicle is, with a jack and a tape measure.

Dave

I have a front strut tower bar on my project car, and it has made a very noticible improvement in the handling of the car.  It is a sufficient enough difference on my car that when I did an inadvertant double-blind test on myself (disconnected the bar to get at the intake manifold and forgot to tighten it back down), I noticed something was dramatically off in the first turn I took.  I've tried a 3 or 4 different bars and it has been my experience that the simpler and straighter the design, the more effective (the $40 straight rod with two brackets is much more effective than the 3-part adjustable bar that cost $200).  

In terms of measurable effects, the first time I jacked up the car after installing one, I was quite surprised to find that the other side came up about 1/2" -- sort of irritating since I have to either disconnect it or use jack stands every time I jack the car up now.

star for that, that's the kind of number I was expecting.

Cheers

Greg Locock

If you need more justification for fitting one then the improvement to the local dynamic stiffness (laterally in particular) will usually result in lower levels of road induced noise!

Strut tower bars are very usefull.  When used in the front, it makes the initial turn-in response, much quicker and more linear.  It can also improve high frequency small impacts for ride comfort.  Basically, the bar holds the tower in place, so the damper can work much better.  The stiffer strut tower structure allows the low speed damping force to work better, because the force is not bending the tower.

In the rear, they are used on hatch vehicles with a high mount strut.  Here, a rear tower bar can make a big difference.  If the vehicle is a open hatch constrution, it is very hard to control the flex of the tower if the shock is mounted high in the vehicle(near the top of the rear seatback).  The bar can make the rear feel more firm during cornering.  the strange thing with a rear bar is that most of the time, there is actually very little stress on the bar.  We have put a strain gauge on a rear tower bar, and basically no values were produced during handling manuvers, But it makes a big feeling change.

I dream that tower-to-tower braces might increase one wheel's stiffness at most 2X (if the wheel stiffness in question is 100% due to tower flex).  In that same dream a brace from each tower to 3-D a stiff corner point toward the middle of the firewall is necessary to get the main part of the chassis involved.  As installed in early Z-28 Fireb-aros, and in a host of US passenger sedans ever since.

You are unlikely to achieve a factor of two in static stiffness just by a simple bolt-on part, but I have seen improvements almost as large as that from a 3 point brace.

In production we try to get the same effect by building a horseshoe around the strut caps, back to the firewall and across there.

This is obviously a less efficient use of metal.

Cheers

Greg Locock

Volvo provided a set of strut tower braces that went from the upper strut bearing to the firewall, at approximately 45 degrees.  These work very well.  They also provided a pair of bars to link the cross-member rearward to the subframe.  All four bars made a very noticeable improvement on my 1985 245 wagon.

Great question

The torsional and bending stiffness are global properties of the body, and bracing the towers makes only a very small contribution to those properties. Therefore when you run your optimisation the 'correct' place to strengthen the shell is not at the towers.

Welded would be better but whoever is servicing the engine might have a few choice words if you use a welded brace at the front.

Cheers

Greg Locock

I agree with Greg there. We already have a few "choice" words for auto engineers when it comes to certain things such as the placement of oil filters! I'm thinking of haveing an extra elbow installed on my forearm...

-Jon

"How come achieving higher torsional and bending stiffness for the structure does not take care of the problem of tower deflections?

I picture the young Ahnold Schwarzenegger wearing a set of those novelty reindeer antlers.  The main car body >>may<< be as rugged as Ahnold, but the suspension towers are mounted out on the tips of the antlers.

High torsional stiffness is best achieved by a large diameter circular tube. We make cars by squaring the tube, then we add gaping holes in the form of doors, trunks, and upper and lower engine bays, further whacking the torsional stiffness.  The stiffness that remains comes  from the relatively and absolutely puny contributions from box/tube frame rails and "torque boxes" hidden in the body structure.  The body and maybe even windshield contribute to stiffness of cars, even (especially?) ones possessing the widely admired "real frame".  I heard Corvette converibles get an X crossmember to try and make up for the missing roof. Studebaker increased their frame thickness about 50% when they added V-8 power, and added an X-member on convertibles and the sliding top wagons.

The importance of the door holes is suggested here.
http://www.gmgoodwrench.com/gmgoodwrenchjsp/gmspo/Ponti...
The stress is high at the corners, suggesting large local strains there, making these areas useful for slight stiffening.  Adding REAL stiffening via diagonals would be the logical improvement, but as others have suggested, would meet with consumer objection.