Slightly different footage of Senna's crash...
Discussion
Personally I think it's near impossible to draw a definitive conclusion but my general gut feeling is something whiffy about it.
The steering column being broken at the the welded section after the crash would at the least show that the welded section (or joins) was the weakest link in the column (excuse the pun). A quality weld would be stronger than the other metal.
I also find a slightly mysterious lack of available footage of key moments adding to the questions.
Whatever the cause, everything since the crash has allowed the speculation to continue over 20 years after the event..
The steering column being broken at the the welded section after the crash would at the least show that the welded section (or joins) was the weakest link in the column (excuse the pun). A quality weld would be stronger than the other metal.
I also find a slightly mysterious lack of available footage of key moments adding to the questions.
Whatever the cause, everything since the crash has allowed the speculation to continue over 20 years after the event..
Edited by Fire99 on Friday 24th March 10:47
Looking at the tech drawing of the column modification, why did they do it in such a complicated way?? Why not just replace the column with a longer one in the same tube material??? The spacer piece was a machined item and the bush must have had to have been modified to take the wider spacer piece too, plus the welding at each end, its just seems to be such an overly complicated solution to solve a relatively easy problem????
number 46 said:
Looking at the tech drawing of the column modification, why did they do it in such a complicated way?? Why not just replace the column with a longer one in the same tube material??? The spacer piece was a machined item and the bush must have had to have been modified to take the wider spacer piece too, plus the welding at each end, its just seems to be such an overly complicated solution to solve a relatively easy problem????
These were exactly my thoughts too when I first time realized how they had actually made the insert piece. Before this I always thought that it was just welded smaller tube inside the larger tube (which I think is still a very common misinterpretation of the structure). I have thought this many times, but I haven't come up with any reasonable justification why to make the insert piece as a machined piece like that.skwdenyer said:
Hi there. Interesting. You've clearly gone to a lot of trouble.
Reading through your paper, I have to ask a few questions:
(1) Is it reasonable to consider the boundary condition at the foot of the column fixed? I'd have to assume that there was some sort of joint there - perhaps a UJ perhaps? In that case, depending upon how the rack and/or joint are mounted, you'd get a rather different boundary condition, allowing some vertical deflection of the column and, therefore a different stress result further up.
(2) What was the fit condition between the inner and outer parts in the bush area? I'm surprised it wasn't an interference fit of some sort - there's no tolerancing on the drawing, so I can't tell. Do you have any information about that? It would seem that that would have made a difference.
(3) I'm a bit confused by your modelling / meshing of the steering wheel. Clearly the moment is going to be critical in establishing the correct stresses, but the models shown in Figures 5,7 show an elongated cylinder quite unlike the real thing. This appears to signficiantly lengthen the moment arm. I realise that modelling is time-consuming, and solid meshes are computationally expensive, but this is 2017 (lol) - I'm old enough to remember doing a lot of meshing by hand (as in text editor to create the deck for the solver), and then laboriously moving things around with Patran. Have you evaluated the sensitivity of your solution to a variation of only a few % in the exact positioning of those masses?
I think your approach is terrific - well done for all this effort so far; I wish I (any more) had access to modellers / meshers / solvers to do this stuff (out of that loop these days). But I think that what you've shown is so clearly very sensitive to precision that, to draw any serious conclusions, you're going to have to go back and make the model rather more precise.
Do keep it up!
skwdenyer, Thanks for your comments. Its always nice to have comments from someone who clearly has knowledge on the matter.Reading through your paper, I have to ask a few questions:
(1) Is it reasonable to consider the boundary condition at the foot of the column fixed? I'd have to assume that there was some sort of joint there - perhaps a UJ perhaps? In that case, depending upon how the rack and/or joint are mounted, you'd get a rather different boundary condition, allowing some vertical deflection of the column and, therefore a different stress result further up.
(2) What was the fit condition between the inner and outer parts in the bush area? I'm surprised it wasn't an interference fit of some sort - there's no tolerancing on the drawing, so I can't tell. Do you have any information about that? It would seem that that would have made a difference.
(3) I'm a bit confused by your modelling / meshing of the steering wheel. Clearly the moment is going to be critical in establishing the correct stresses, but the models shown in Figures 5,7 show an elongated cylinder quite unlike the real thing. This appears to signficiantly lengthen the moment arm. I realise that modelling is time-consuming, and solid meshes are computationally expensive, but this is 2017 (lol) - I'm old enough to remember doing a lot of meshing by hand (as in text editor to create the deck for the solver), and then laboriously moving things around with Patran. Have you evaluated the sensitivity of your solution to a variation of only a few % in the exact positioning of those masses?
I think your approach is terrific - well done for all this effort so far; I wish I (any more) had access to modellers / meshers / solvers to do this stuff (out of that loop these days). But I think that what you've shown is so clearly very sensitive to precision that, to draw any serious conclusions, you're going to have to go back and make the model rather more precise.
Do keep it up!
Here are some considerations on your questions.
1. I have assumed that the column was fixed to the pinion rigidly. There are some photos which seem to suggests that. However, the drawings are not exact on this matter so its uncertain how it was fixed. I actually opened one of my more simple models to test this thing (the accurate model takes too long to run
). It would seem that if the pinion was not completely fixed but free to rotate (only translation fixed) the flexing of the steering wheel would increase about 10%. However the effect on the stress levels at the fillet area are negligible, under 1%. Most of the constraint forces are anyway going trough the bracket. 2. The exact fit is not known. I have used 0,2 mm initial gap between the bushing and the column. I would expect it to be quite tight fit as you suggests, so 0,2 mm might be on the safe side when considering which gives more flexing.
3. The steering wheel is very roughly modeled because it is in essence only a point mass in the model. It does not lengthen the column, the dimension are the same as seen in figure 2. The cylinder which models the steering wheel is the last part with length of 65 mm. It sure does not look correct, but in my opinion it does not effect the accuracy of the model as the mass of the steering wheel is in the correct position.
So in my opinion there is not that much wrong with the model. In any case there are uncertainties for example in loads, so its very hard to come to any exact answer.
VladD said:
Does anybody apart from me think that's a bit odd? I've never been a professional F1 driver, but if my wheel started moving about as much has been claimed, I think I'd probably slow down and head for the pits and at the very least mention it to my team.
I understand what you're saying and agree your instinct would be to slow down if you thought something had gone wrong with the car (or at least mine would!). However, if the movement only became excessive immediately prior to the accident perhaps by the time he realised there was something wrong he didn't have an opportunity to radio the pits? The car would have been travelling at the best part of 200mph approaching Tamburello....-Ilu- said:
Hello everyone,
So here is a link to my analysis of the FW16 steering column.
Interesting read...So here is a link to my analysis of the FW16 steering column.
You said the interface between the column and the bush was rigid, but between the bracket and bush was free to slide. Surely it should be the other way around, with the bush presumably being pressed into the bracket, and the column being free to slide and rotate within the bush?
If the assembly is modelled as you say, with rigid connection - rightly or wrongly - between bush and column, how come there are no significant stresses transferred into the bush (or indeed into the bracket - Figure 9)? It's as if the bracket/bush assembly is fully constrained, and there is some kind of error in whatever connection exists between the elements there. If I'm interpreting your deflection plot correctly, the column appears as if it could be crashing through the bottom of the bush, indicating a contact element error (although I know scaled deflection plots can sometimes do this even if modelled correctly).
For the magnitudes of stress and deflection you're talking about, I'd expect to see at least some kind of stress distibution in the bracket - even though it's a coarse mesh - particularly around the boss.
It seems you've got a load of different element types and connections in a fairly small area; are you sure you're not over-constraining your model somehow, which could give artificially high peak stresses in the fillet radius (since the loads don't have a realistic path through which to distribute)?
Just some observations...
-Ilu- said:
number 46 said:
Looking at the tech drawing of the column modification, why did they do it in such a complicated way?? Why not just replace the column with a longer one in the same tube material??? The spacer piece was a machined item and the bush must have had to have been modified to take the wider spacer piece too, plus the welding at each end, its just seems to be such an overly complicated solution to solve a relatively easy problem????
These were exactly my thoughts too when I first time realized how they had actually made the insert piece. Before this I always thought that it was just welded smaller tube inside the larger tube (which I think is still a very common misinterpretation of the structure). I have thought this many times, but I haven't come up with any reasonable justification why to make the insert piece as a machined piece like that.In the 70's it was common for them to just use the OD of the tube as the contact face that passed through the support bracket and then use a nylon bush/block as the support, in the 80's you then saw a composite bush used that allowed some articulation still running on the OD of the tube. Both of these give some slop in the column. As manufacturing improved and carbon tubs were made with jigged mountings for the steering racks, you could then use tighter tolerances on the support bush as the column alignment was very accurate (which is also why you don't need to use a U/J at the pinion end), which allowed a bronze bush and machined OD column to be used.
It likely that even on the unmodified shorter column, they still used a machined insert in the section of column that passed through the support bracket bush, as the OD tolerance on a tube isn't accurate or consistent.
Current F1 cars have gone further on this and now use roller bearings for the column supports, to reduce friction and improve clearances further still.
jsf said:
-Ilu- said:
number 46 said:
Looking at the tech drawing of the column modification, why did they do it in such a complicated way?? Why not just replace the column with a longer one in the same tube material??? The spacer piece was a machined item and the bush must have had to have been modified to take the wider spacer piece too, plus the welding at each end, its just seems to be such an overly complicated solution to solve a relatively easy problem????
These were exactly my thoughts too when I first time realized how they had actually made the insert piece. Before this I always thought that it was just welded smaller tube inside the larger tube (which I think is still a very common misinterpretation of the structure). I have thought this many times, but I haven't come up with any reasonable justification why to make the insert piece as a machined piece like that.Similarly, when I first saw the modified column on Senna's car, I assumed it was a deliberate weak point that would fail in the event of an accident. Obviously the failure would be in the wrong direction to do much good though. Unbelievable it was necked down that way for no good reason.
I only have a yellow book for '92, not '94, but it makes no mention of any requirement for impact absorbing steering columns.
dr_gn said:
Interesting read...
You said the interface between the column and the bush was rigid, but between the bracket and bush was free to slide. Surely it should be the other way around, with the bush presumably being pressed into the bracket, and the column being free to slide and rotate within the bush?
If the assembly is modelled as you say, with rigid connection - rightly or wrongly - between bush and column, how come there are no significant stresses transferred into the bush (or indeed into the bracket - Figure 9)? It's as if the bracket/bush assembly is fully constrained, and there is some kind of error in whatever connection exists between the elements there. If I'm interpreting your deflection plot correctly, the column appears as if it could be crashing through the bottom of the bush, indicating a contact element error (although I know scaled deflection plots can sometimes do this even if modelled correctly).
For the magnitudes of stress and deflection you're talking about, I'd expect to see at least some kind of stress distibution in the bracket - even though it's a coarse mesh - particularly around the boss.
It seems you've got a load of different element types and connections in a fairly small area; are you sure you're not over-constraining your model somehow, which could give artificially high peak stresses in the fillet radius (since the loads don't have a realistic path through which to distribute)?
Just some observations...
Thanks for comments. You are right, the connections between the bushing, the column and bracket should be like you said. And in the model they are (just checked), so I have made a mistake in the report... something to correct for the next version!You said the interface between the column and the bush was rigid, but between the bracket and bush was free to slide. Surely it should be the other way around, with the bush presumably being pressed into the bracket, and the column being free to slide and rotate within the bush?
If the assembly is modelled as you say, with rigid connection - rightly or wrongly - between bush and column, how come there are no significant stresses transferred into the bush (or indeed into the bracket - Figure 9)? It's as if the bracket/bush assembly is fully constrained, and there is some kind of error in whatever connection exists between the elements there. If I'm interpreting your deflection plot correctly, the column appears as if it could be crashing through the bottom of the bush, indicating a contact element error (although I know scaled deflection plots can sometimes do this even if modelled correctly).
For the magnitudes of stress and deflection you're talking about, I'd expect to see at least some kind of stress distibution in the bracket - even though it's a coarse mesh - particularly around the boss.
It seems you've got a load of different element types and connections in a fairly small area; are you sure you're not over-constraining your model somehow, which could give artificially high peak stresses in the fillet radius (since the loads don't have a realistic path through which to distribute)?
Just some observations...
There are of course some stresses in the bracket too, but because the plots are scaled with the maximum stress there is not much to see in the bracket area. The bracket, with the dimensions from the available drawings (there is some uncertainty there), is much thicker material than the tube so I am not expecting to see much of a stress there. The forces acting on it are quite low anyway (about 100 Newtons vertical).
The model is quite complex with many contacts (although the glued contacts are very easy to work with), so it has taken some time to get it work correctly. I have tested the same case with simpler shell model and also with very simple beam model, and got quite the same same results. So I am confident there is not anything major wrong there. The column crashing through the bottom of the bush is due to the exaggerated deformation used in the plots. With actual deformation there would be not too much to see because of very small flexing.
jsf said:
There is no uj on the rack.
Thanksjsf said:
What is the point of spending all that time on this issue?
Me or Ilu? If Ilu has the time, and wants to try to understand what has been said and seen, why not? The world is full of people taking no interest in things, of not asking questions.Nobody is forcing you to read the thread
Edited by skwdenyer on Saturday 25th March 16:25
JNW1 said:
VladD said:
Does anybody apart from me think that's a bit odd? I've never been a professional F1 driver, but if my wheel started moving about as much has been claimed, I think I'd probably slow down and head for the pits and at the very least mention it to my team.
I understand what you're saying and agree your instinct would be to slow down if you thought something had gone wrong with the car (or at least mine would!). However, if the movement only became excessive immediately prior to the accident perhaps by the time he realised there was something wrong he didn't have an opportunity to radio the pits? The car would have been travelling at the best part of 200mph approaching Tamburello....This wasn't a case of driver error . You can clearly see Senna trying to fathom out a problem towards the pedals or column . When entering the curve he knows it's all gone tits up and he won't be turning in , hence the immediate flinch to the left when he realizes that he's just a passenger from that moment forward . He could have walked away but sadly luck wasn't on his side that day .
I'm normally keen on Nigel Roebuck and his thoughts, but this is garbage.
He is only able to speculate, but in doing so, citing tyre pressures, means he has overlooked the obvious, which is where I can't forgive his bland conclusion, in that the cars had already been racing for a whole lap, so tyre pressures surely are out the window as they'd already done a lap at full speed..
Brundle is sitting on the fence, which is at odds with his normal persona.
Things get interesting when Berger speaks, and Piquet's input is even more interesting, given his accident there in 1987 (i think) which was not a driver fault.
In this footage there is in-car stuff right at the end. Have a look at Youtube comments where people seem to understand things better.
Hill is the only F1 driver (a Williams employee remember?) who seems to think Senna just 'lost' control.
https://www.youtube.com/watch?v=YtvBhRTYrZ8
He is only able to speculate, but in doing so, citing tyre pressures, means he has overlooked the obvious, which is where I can't forgive his bland conclusion, in that the cars had already been racing for a whole lap, so tyre pressures surely are out the window as they'd already done a lap at full speed..
Brundle is sitting on the fence, which is at odds with his normal persona.
Things get interesting when Berger speaks, and Piquet's input is even more interesting, given his accident there in 1987 (i think) which was not a driver fault.
In this footage there is in-car stuff right at the end. Have a look at Youtube comments where people seem to understand things better.
Hill is the only F1 driver (a Williams employee remember?) who seems to think Senna just 'lost' control.
https://www.youtube.com/watch?v=YtvBhRTYrZ8
Gassing Station | Formula 1 | Top of Page | What's New | My Stuff