Mike's really good at revealing just the right amount of detail. You both are doing an terrific job by taking mystery out of your topics. BTW: "Chassis Torsional Stiffness" may be too simplified a description. This number would be the sum of all the compliances from each of: at least 3 subsystems: front suspension area, motor compartment area, and rear suspension area. Just like there is really not a Roll gradient because you need to define the position from which your measurement device is reading. If your driver says "the front rolls more than the rear", they are either an idiot or a VERY perceptive evaluator. Windshield & sidegass became torsional stiffness players. Then there is the measurement technique. Which is best ? GM used a "Pull down test" to hold 3 corners of a frame, car, cradle, etc, and metrifying the moment that resulted. Spring centers used as support points if total frame or car is being measured. Then the K&C rigs brought their own issues: How to secure the vehicle ? 4 mounts, 3 mounts, 2 mounts< and even 1 mount point. For inertia relief simulations on a K&C rig, many iterations are often needed (called "bootstrapping") because of the mounting type and "Chassis Torsional Stiffness", tire forces & moments had to be recalculated several times because the nonlinear properties altered all the other compliances like roll/steer, roll/camber, tierod loads, etc. Dog chasing tail comes to mind. Panhard bars are more difficult to deal with. Then there is the durability trade-off: more stiff: more likely to break. The solution to this in the '80s was a change from "stiffness" to "frequency". With HP Fourier Structural Analyzers, the 'machinery' assembled an exaggerated Normal Modes animation so you could now visualize WHERE weaknesses came from. BIG weight savings because now you could remove metal that was not a player in the localized strains. Definitely stiffen the locations where it was needed: suspension attachment points, roll bar fixturing, spindle size, steering gear mount, steering shaft attachment. (This one was cool to find: The steering wheel is in a roll flexing cage but the steering gear is tight to the front cradle. So is this Roll/Steer, Roll-Steer, steer from roll, ???). So the bottom line was then adding the normal modes of the front, motor, driver position, rear, body, and even the seat(s) to get a total frequency. Then set targets based on driver evaluations and strength of materials considerations (That's the mass trade-off). When it's all assembled from measurements or NASTRAN results, there may be 50k - 100k modes involved, but not necessarily all sequential. In a suspension member (like a twist axle), some modes won't participate at all, so they are left out (NASTRAN identifies "Modal Participation: for each assembled 'package', saving a lot of computer time. As a fact, tires are now done this way. Bead, belt package, liner, tread base, tread blocks are assembled as a frequency package and targets compared. I wrote an SAE paper back in the mid '70's on how I did handling tests with NASTRAN 'ride' models' because a racecar is still a racecar even if you spell it backwards. It shows the positionally based torsional stiffness of a car. They actually let me publish this before it (or I) died of old age !

The whole of having a stiff chassis is to allow the suspension to do its job and be tune-able. Jumping between a road car and a race car on the same chassis with the same suspension and tires, the caged race car does EVERYTHING better. And once you adjust the suspension to suit the chassis stiffness it drives better and more predictable with the roll cage adding torsional stiffness. A chassis is a spring and the softer it is the more it takes away from the suspension

A Chinese vehicle just claimed the record of torsional rigidity. It's the yangwang U8 and got 54425 N·m/°. It was introduced on the recent Guangzhou car show. I'd like to know the number of some other vehicles, like the porsche 991, the jaguar XE project 8 or the stelvio quadrifoglio

I don't agree with what is said at 2:35, because in corners there is a weight transfer to the outer front (cornering) wheel and the twisting force of the two axles are coming from this cornering force.