17 replies to this topic

[blueprint2002]

As everyone on this forum knows, the engine of every F1 car doubles as the rear part of the chassis, and has done so since the Lotus 49 and the Ford DFV engine of 1967. Or, pedantically speaking, since the Ferrari Flat-12 of 1965 and the BRM H16 of 1966.

What is perhaps less well known is to what extent the engine contributed to increased torsional stiffness of the assembly: I for one have not been able to find anything on this subject. Of course, the figures for the cars of that era have surely been improved upon, by at least an order of magnitude, with the developments since then: the substitution of bonded aluminium sandwich for plain riveted aluminium sheet probably the first major step forward, this material in turn superseded by the moulded carbon-fibre-reinforced plastics that have long since become universal. The contribution of the engine is a lot harder to evaluate, not much relevant information ever having been made public, as it appears.

Torsional stiffness is actually only one aspect of chassis performance, bending stiffness in both longitudinal planes being probably of no less significance. Aerodynamic loads have certainly seen to that, their fore-and-aft distribution just as important as their considerable magnitude. Together with the weight distribution, the aerodynamic forces directly influence the bending moments in the vertical plane, and indirectly, through the cornering loads at each tyre contact patch, those in the horizontal plane. In fact, as the aerodynamic loads are symmetrical across the width of the car, it seems logical that they don’t contribute at all to torsional loading, so that the magnitude of these should not be radically different to those of 50 years ago.

Has anyone seen any figures and/or informed discussion of these matters in any published work? I imagine that it would have to be authored by someone with a background in racing car design and construction (as opposed to the enthusiastic amateurs we are all familiar with): Peter Wright, Gordon Murray and Adrian Newey come to mind, among a host of other true professionals in the field.

 

[Sisyphus]

The overall stiffness of a system of components with varying stiffness is going to be dominated by the least stiff component.  So, if the torsional and beam stiffness of the engine is significantly more than the rest of the chassis, which seems reasonable to me, then exactly what the stiffness is would not be too important.  My guess would be that the stiffness increase from solidly mounting the engine versus not would not be huge but maybe someone has data to prove me wrong.  The stiffness of the connection between the engine and the rest of the chassis would likely be more important it seems to me.

 

I've never understood what exactly was meant by an engine being "designed to be used as a stressed member" since no matter what an engine block is going to be a bit chunk of relatively thick walled aluminum and therefore pretty stiff relative to a sheet metal monocoque.  Where the motor mounts are located on the block could be important, of course, in how much of the block can be loaded, and maybe where the loads are fed in with respect to the crank bearing locations.  Other than providing convenient locating points for the engine to be mounted to the tub and the rear suspension, I don't understand the meaning of the term.

[Joe Bosworth]

blueprint in starting this thread states,

 

"Torsional stiffness is actually only one aspect of chassis performance, bending stiffness in both longitudinal planes being probably of no less significance"

 

I note that this is quite in error.  I say this as having the advantage that torsional stiffness really only became a real design issue in the 1960s, a point in time in which I had substantial practical experience in developing a series of race cars.

 

Both from carrying out paper design calcs, making chassis test measurements and taking the cars to the track to test handling hands on experience provides a good view of the realities of this issue.  To start with, from the time of first race car designs longitudinal strength was adequate despite torsional strengths of less than 2000 ft lbs per degree were prevalent.  The real problem at these low torsional strengths were that the front and back ends of the cars while cornering were doing different things.  At these levels you could change a front or back spring rates by 30 or 40 lbs per inch and make little difference in u-steer/o-steer.

 

By the time you built to say, 3000 ft lbs per degree, spring rate changes of 25 lbs per inch could be felt.  Once you got to plus 4500 to5000 ft lbs a 10 pound per inch spring rate change  became substantial in changing handling.  One quarter of an inch movement of adjustable roll bar made a lap time difference.  Through all of these developments the longitudinal strength was barely changed because making any change in this cost big time chassis weight. lighter was faster as long as you kept chassis twisting under control.

 

I don't have any fix on torsional strength of modern carbon chassis but believe that designs are now based in impact safety rater than handling factors.  I should doubt that modern chassis are good for much more than 20,000 ft lbs per degree as more than about 8000 ft lbs has little affect on handling.

 

Regards

Edited by Joe Bosworth, 21 September 2020 - 19:08.

[Greg Locock]

I'm glad to see we're using Imperial units. As it happens Nm/deg are nearly the same numbers. OK, Giugiaro Esprit was about 2500 Nm/deg. For the X180 intercooled I fitted a couple of braces to the front of the car (after much deliberation), effectively using the front of the body as a stressed member, which was a bit anti-Chapman, but no one noticed - it actually helped solve a problem with gel coat cracking above the wheel arches. 4500Nm/deg. Massive increase in understeer. They then retuned around the new stiffness and turned the Esprit from a scarily neutral, if nimble, car to one that Joe Blow (me) could drive reasonably well. So when we built SID torsional stiffness was definitely on the list of things to look at. But I can't remember what it came in at. All the bushes on the spine chassis/subframes/body/engine interfaces were identical in size so we could in theory reconfigure SID to almost any architecture that was likely. Sadly I don't think that aspect of the design was ever used. I agree 10000 Nm/deg is where the improvements start to become less noticeable. Some manufacturers fetishize the torsional frequency of the BIW, which is closely connected to torsional stiffness, frankly once you are up above 25 Hz the advantages are negligible.

Edited by Greg Locock, 21 September 2020 - 21:46.

[Greg Locock]

Back to the OP, if you know the length of the engine/gearbox assembly between the mounting points, and the OD of the casing, estimating the torsional stiffness is not hard.

 

i doubt it matters greatly  in a 4g aero car riding on soft tires.

Edited by Greg Locock, 21 September 2020 - 21:58.

[gruntguru]

The overall stiffness of a system of components with varying stiffness is going to be dominated by the least stiff component.  So, if the torsional and beam stiffness of the engine is significantly more than the rest of the chassis, which seems reasonable to me, then exactly what the stiffness is would not be too important.  My guess would be that the stiffness increase from solidly mounting the engine versus not would not be huge but maybe someone has data to prove me wrong.  The stiffness of the connection between the engine and the rest of the chassis would likely be more important it seems to me.

 

I've never understood what exactly was meant by an engine being "designed to be used as a stressed member" since no matter what an engine block is going to be a bit chunk of relatively thick walled aluminum and therefore pretty stiff relative to a sheet metal monocoque.  Where the motor mounts are located on the block could be important, of course, in how much of the block can be loaded, and maybe where the loads are fed in with respect to the crank bearing locations.  Other than providing convenient locating points for the engine to be mounted to the tub and the rear suspension, I don't understand the meaning of the term.

I think that is a good summary of the factors involved in the OP's question. Beyond that:

 

 1. I would think it reasonable to assume the engine to be infinitely stiff. Firstly because, as you say, the least stiff components contribute the vast majority of the combined (series) stiffness. Secondly because you cannot afford for the engine block to deflect significantly. The internal clearances and tolerances are very fine and are critical to its performance.

 

 2. An engine "designed to be used as a stressed member" would require a) suitable external features for loads to be applied and b) sufficient stiffness through the resulting load path to ensure that operating tolerances of the engine are not exceeded under chassis loads.

Edited by gruntguru, 21 September 2020 - 22:32.

[Greg Locock]

25 years ago it wasn't uncommon to see external structures on longitudinal engines to improve their bending stiffness, for NVH reasons. MB went one step further and started casting their bellhousing integral with either the sump or the the block, I can't remember which.

[Charlieman]

The DFV is mounted to the bulkhead behind the driver's seat and/or fuel tank. There are two screw mounts at the bottom, with two triangular plates at the top. The triangular plates are intended to permit thermal expansion. There are bolt holes at the rear of the cam covers on some DFVs for suspension mounts. My guess is that the mounts are one of the least rigid elements in the chassis.

 

Half a century earlier, Ettore Bugatti mounted his engines rigidly to the chassis. Early versions used lugs on the block/upper crankcase; others used lugs on the lower crankcase/sump. It all saved a bit of weight on crossmembers.

[Fat Boy]

I've never understood what exactly was meant by an engine being "designed to be used as a stressed member" since no matter what an engine block is going to be a bit chunk of relatively thick walled aluminum and therefore pretty stiff relative to a sheet metal monocoque.  Where the motor mounts are located on the block could be important, of course, in how much of the block can be loaded, and maybe where the loads are fed in with respect to the crank bearing locations.  Other than providing convenient locating points for the engine to be mounted to the tub and the rear suspension, I don't understand the meaning of the term.

 

The block is designed to take certain external loads with minimal distortion of the cylinders or bearing bores.

[Fat Boy]

 

i doubt it matters greatly  in a 4g aero car riding on soft tires.

It's noticeable by the bloke driving the 4g aero car on soft tires, even if the stopwatch can't see it.

[Greg Locock]

I've got evaluators who claim to be able to feel alloy vs steel wheels. I once did an exercise with one, where we made a hardpoint change repeatedly, he could spot it, statistically, and we could measure it on the K&C, but we couldn't see it on any of our standard steering plots.

[Joe Bosworth]

Greg

I would be fascinated to know what factor he was sensing to allow him to recognise between steel and alloy wheels.

 

I know that 1 PSI difference in tire pressures can be felt but believe this is mostly due to side wall compliance and contact oint changes.  If he was sensing wheel compliance I would have thought that would be hidden in the tire compliance but maybe not.  Could it have been in frequency difference due in part to sprung/unsprung ratio weight change?

 

I have never come on this being an issue despite many hundreds of opportunities.

 

Regards.

[Greg Locock]

We /think/ it was camber stiffness of the wheels. When they want a cop-out when reporting stuff they usually classify it as compliance-feel. For some reason the rear suspension is often more important. My pet theory for that is that the driver has direct physical control over the front wheel steer angle and so can rapidly compensate, but what the rear gets up to is more of an outcome of a chain of events. Anyway I don't have a good steering test for it, the best we can do is chuck the car on K&C and try and reduce the toe/MZ compliance.

[Kelpiecross]

  I remember reading in the sixties  about an Italian  single-seat car  (I think for Formula Junior or something similar)  from a fairly obscure maker that used an unmodified  pushrod Ford  engine as  the rear part of the car.   Apparently it had problems with the crankshaft main bearings due to the block flexing.  

Edited by Kelpiecross, 23 September 2020 - 05:01.

[Greg Locock]

Yup, bearing ladders and structural sumps aren't there just to keep the machinists busy.

[mariner]

My practical knowledge is zero versus Greg and Joe but I can add some comments from one of the Multimatic shaker rig engineers who worked with Dave Windsor and John Mile at Thetford.

 

He said they routinely instrument the cars at intermediate points along the chassis when on a shaker rig so as to separate out mounting and chassis intermediate point stiffness. 

 

His comment was that they had tested some carbon fibre F1 type cars and there were distinct changes in stiffness at the engine to fuel tank joint . I think he also said that was true of gearboxes to engine joints. His general comment was that modern F! cars were not quite as perfectly rigid as the teams like to claim

 

In terms of engine stiffness  nothing is a much bigger lump than a big block Chey of 427 ci to 500+ ci. It weighs 240lb!  They do have reputation for twisting if used as chassis member. The CanAm Mclarens used an alloy block as a stressed member and they had triangulation tubes from the chassis bulkhead to the bell housing to help. Also when GM used one in its clever 4WD sports experimental car they had a brace over the big block and they had some very good measuring abilities.

[Greg Locock]

Back in the olden days (before FEA was reliable- we should still do it!) we got a big surface plate, mounted the body on that, and used dial gauges down the length of the car to measure the deflection as it was twisted. This led into the biggest single change in BIW design i know of - proper analysis of the joints, say A pillar to roof to header and cant rail. This started a fashion for chopping the structure up into bits about 500mm long around each joint and trying to measure the stiffness of that joint. We stopped doing that, it was incredibly difficult to get meaningful results. 

[DogEarred]

I remember back in those Great Days, when I worked for a design house in London.

 

One of the projects in house, was to convert a Vauxhall Vivaro van (Transit size) into a sandwich van.

 

After much chopping & cutting of body panels, it was decided to check the chassis stiffness by bolting a long RSJ to the rear of the chassis, bolting down the front of the chassis, then loading up the end of the RSJ with heavy weights.

 

Myself & another huge guy were commandeered to be the heavy weights sitting on the end of the RSJ.

 

Our weights were recorded & the chassis twist was carefully measured.

 

On removing ourselves, it was checked whether the chassis had returned to its original state.

 

It had not & was bent measurably.

 

 

'High fives' from my big mate & I for a RESULT!!!...

 

 

I also remember, not so long ago, in a quite successful F1 team's R&D dept., they used a huge, hollow alu or steel cube as a stable base for measuring such as twist bending etc. for various components.

 

I recall that a complete monocoque was bolted to the cube, using an adaptor, for the front bulkhead, with the chassis sticking out into fresh air.

 

The rear bulkhead could also be adapted & hydraulic rams used to introduce loads. 

 

A dummy engine & gearbox could also be installed & measurements similarly taken.

 

 

Edit:

 

Talking of contemporary Formula 1 (if we were) all the engines are nowadays mounted to the back of the monocoque by 6 studs, either attached to the tub or engine. 

(The technical rules specify the spacing in order to make any change of manufacturer a simpler affair to redesign but the actual components are free.)

 

The chassis metal inserts for the studs, bespoke studs & engine materials differ so great care is taken in setting the geometric tolerances of each, to account for the different coefficients of expansion such that the best fit, (hence torsional stiffness) is at the engine running temperature.

 

Similar antics go on at the engine/gearbox interface but I don't think any stud spacings are mandated. (I remember having to design an adapter plate for a change of engine manufacturer)

Edited by DogEarred, 24 September 2020 - 22:51.