keith

oooohhhh.... good stuff...

would not the car itself deform and absorb energy to some point? Crumple the shell but protect the space within the cage?

Glen

Most likely cause was because during testing the car spends a grea deal of time on the dyno. The right belt was continually laid over to the right/across the trans tunnel area. Was hence exposed to long sustained heat cycles during dyno time causing the belt material to weaken fatigue and tear much easier in the impact.

RedlineMan

Wow...

Thread reinvigorated from the Smithsonian!

Hey guys. Comments well recieved. The 996 is an exceedingly difficult car to build a bar of this type for WITHOUT HAVING TO CUT AND WELD THE CAR. This was the main goal, and it SEVERELY limited my options. The architecture is simply not condusive to a good geometrical layout of tubes as they relate to mounts. If you try and build one, you will see that in very short order.

Having done this a few times now I can tell you that in the end you are left to do little else than make the best compromises you can. ALL of these products have their own set of them. The question becomes how astute are you at recognizing and prioritizing them?

Kurt; you have made some observations based on intuition. They are not any different than the ones I have made myself. Now, take your intuition and apply it to the other products available for these cars (Tech, DAS) and see how they stack up. They too will present compromises. The biggest one I see in the two of them is the main hoop mount. While the mount is closer to the tube itself, it still is not DIRECT, is unsupported by ANY bracing, and the fairly thin material and the layout of the flange does not impress me. Further, the OE offering has pretty wimpy rearward bracing in my opinion. The both also severely limit rearward seat travel for tall guys like me. Those are the things I noted when sussing out my design.

I did in the end produce a slighty different bar. I honestly have never liked not having a direct hoop mount, but there is no way to do it and not cut/weld. I also note that no one seems to note this fact about the other two products available. In any event, I am wise enough to know that nothing is ideal. If you want to call them glorified harness bars, that is not totally unfair. But what a harness bar!!

The prototype (silver) spent the last year in a local car, and fortunately for him, was not "tested!" The first production version (below) is still in a 996C4 somewhere in the wilds of Minnesota, and happily providing that owner with peace-of-mind. I think it is a reasonable design, all tolled. A reasonable set of compromises.

Carlos, I honestly have no interest in sanctioning these for racing because I strongly feel that to race is to cage!

kurt M

(edit) posted this before I saw you last post with the finished pix. Too true I too don't like any of the bolt in bars but think that they might be better than nothing. IMHO I think you might have too much bracing in the bar considering atachment streingth and energy imput. a chain is only a strong as the weak link.

As to some of the previous posts and just some ramblings.
Roll bar and cage are two very different animals. All a roll bar can do is resist roof intrusion and to a much lesser degree side to side failure of the upper drivers compartment. A bar also gives a good hard point for harness attachment. A cage on the other hand should resist driver’s compartment deformation in all 3 axes and be able to deflect and distribute energy around the driver’s compartment. Weight is the limiting factor in the drivers compartment intrusion protection in ether case. Not talking about any added tube framing just the drivers compartment.

The key is to redistribute point source energy around and away from the driver.

The data I have been following is shows that the best thing to have is the driver’s compartment as strong as possible and driver well restrained. You do not want the cage to be an energy absorption crumple zone. You want the auto body outside of the driver’s compartment to absorb energy. If you tie the cage to suspension or body points that are remote from the driver’s compartment then you need to think about weakening them via wrinkle zones, using smaller tubing or other means of isolating the driver’s compartment structure from impact energy.
A 911 front impact as an example. A sausage cut and then partial reweld to the horizontal fore to aft front trunk tube will cause overload type impact energy to make the tube fail at the cut and divert excess impact energy away from the internal cage. The tube will act as a stiffening agent until the energy input exceeds the reweld point strengths and then will fail at the planed points. The body of the car takes the hit and only after it compresses dramatically will the cage have to become involved in notable energy absorption. We are talking about a real hard hit here. Look at NASCAR tube frames. The driver’s compartment area is built like a tank and when the cars are involved in high energy events they look all rounded off but the drivers compartments do not deform notably. Same with F1. You want a strong driver’s compartment cage that resists deformation but let the rest of the car crush at a prescribed rate of energy absorption. Absorb energy but crush the driver is not good. It is a balance of deceleration to intrusion. Too much of ether will “negatively impact” the driver.

mpaton

Hmm, got to reply to this bit.

Kurt, I've heard this kind of reasoning about cages before, and I think part of it may be misguided. I'm with you on using the body structure as a crumple zone while keeping a strong driver's compartment in general. Your example of a frontal impact deals with this well.

Where I think it falls down (or where there is new reasoning since I last looked at it) is in impacts where there is no substantial crumple zone built into the car. That is impacts vertically down onto the roof, and laterally applied at roof level, in other words those impact forces that arise from a rolling, somersaulting tumbling car. This is where I believe there can be too much strength (even though I will concede that it is hard to know what the the magnitude of the impact will be.

To protect the driver in big impacts (and the body structure in smaller ones), roll bars/cages especially those not tied to the structure at roof level should (IMO) be made to fit the interior as closely as possible, so that when they do deform, and absorb energy, they still leave an acceptable size of driver compartment.

If you disagree, then tell me why you would want to be in a tumbling car that has completely elastic impacts when it hits on the roof.

Michael Paton

smokey

There are two types of crashes that are not survivable due to simple geometry: a direct impact on the roof, i.e. a "pancake" landing on the roof from a height, and a t-bone impact on the driver's side. In either case, there is no way to provide enough crumple zone to reduce driver deceleration to acceptable levels without physical intrusion. Either the acceleration or intrusion will be terminal. In a roll-over, the chances of survival increase as the tumbling dissipates energy, as long as the centripetal acceleration experienced by the driver stays at survivable levels and there is no intrusion. Given these realtities, a strong driver compartment - and a smooth exterior that encourages tumbling - seem to make sense. As in everything esle, it's better to be lucky than good, although being both is obviously optimal. I have seen many cages where the struts to the shock towers are clearly in the crumple zone, and would act as a spear if they detach, or at the very least would reduce the effectiveness of the crumple zone and buckle the side longitudinals of the cage. Given that the powers that be don't understand the forces involved in a harness guide bar, my hope for any imporvement in this area is minimal.

Z-man

I've read this thread through, and it is quite interesting. Some non-technical comments/observations/questions:

As John and others have said, it's all about compromise. Will my bolt in rollbar (Redline, BTW) work in keeping my roof from occuping the same space as my head in a roll-over? I am counting on it. Are there other options? Sure!

Most rollbars do NOT help reinforce the weakest point in 99% of the cars out there: the area between the A-pillars. Except for cars like the VW New Beetle, the A-pillars are typically the thinnest posts on the car, and thus, often the weakest. It is also the weakest since it is farther away from the rest of the pillars. (The B - C pillars are much closer together on most cars). In the 944, the B-pillar is much bigger, and I suspect much stronger than the A-pillar. Just take a look at some 944 rollover shots, and you'll see that most often, it is the roof section between the A-pillars that collapses vs. the section behind the driver. In order to reinforce the A-pillar section, a full cage is really necessary. While that may be ideal for a track-only car, most DE cars see some street time. A full cage without a helmet opens a whole new can of worms... So, why do I bother with a rollbar, if it is not reinforcing the weakest area of my roof? It's all about compromise and peace of mind, IMHO.

Question about bolt in vs. welded roll bars: Many folks argue that the sheer strength of a bolt in rollbar is weak. But welded in rollbars are welded/bolted to plates that are mounted to the floor/sills of a car, ie sheet metal. I doubt welding to sheet metal is that much stronger than using the stock seat belt anchor points. Like I said, I am non-technical in regards to the engineering perspective of rollbars and cages. But I have a hard time believing that something welded to sheet metal is stronger than something bolted to a beefy grade 8 anchor point. Can anyone elaborate on this?

Now , a question about seat braces. If part of the purpose of a roll-bar and cage is to absorb the impact and keep the crash energy away from the occupants, why attach the seats to the rollbar or cage? Doesn't that essentially negate much of the impact absorbtion, especially if the cage is directly tied to the frame? If one is concerned about the seats collapsing, wouldn't a better seat design be a more effective solution?

Thanks, and sorry for hijacking this thread.

John - when/if I ever get a 996 (it is on my wishlist...) you're rollbar will be my first modifiation! I have your 944 rollbar, and believe in its design and its effectiveness.
-Z-man.

RedlineMan

Some good further points, Gents;

My current thinking is that you must above all else maintain a "cacoon" around the driver that will not significantly deform, and will survive intrusion. Within that cacoon you need proper padding, belting, and seating. SFI only, 6-point definitely, ridgidity positively, in that order!

In other words, you are trying to restrain the driver and limit their movement as much as possible. This means SFI high density padding in impacts areas, polyester 6-point belts, left & right side nets, and a stiff seat that is fully attached to the chassis, both top & bottom. The old thinking had this transfering too much energy to the driver, but recent testing has shown the driver can take huge forces and is actually more vulnerable to the "wet noodle whip" of loose or floppy equipment than any direct impact.

Let me offer some more insight into my thinking on bolt in roll bars. There are indeed many things you are trying to accomplish, and it is tough to cover all those bases.

First - I do feel that a welded bar is best, given proper design. That means fairly large weld plates located at chassis junction points of heavier original construction. You choose the junction of many beam structures in the unibody and essentially create an endoskeleton, mimicking the existing chassis and utilizing its pre prescribed stress points for mounting.

IF the weld plates are located at these structural areas, and they are of sufficient surface area, they will indeed dissipate load over a wider area than a bolted mount in single shear, as in one run into a seat belt hole. That is obvious. If you design a piece to require no mods, then you mount to the belt holes by default and accept the compromizes inherent in that setup.

Second - Expanding on the above, if we can assume that the single shear bolt mount is less strong than a weld plate install, how can we mitigate possible mount failures? I have always attempted to create my bars as a sub-structure within the cars interior. A structure that can withstand loads from many angles, on its own, without even being mounted.

If you take a bar hoop, stand it up, and load it from above, it will withstand X amount of force before collapsing. If you add a belt bar to the hoop, you have created a box. This belt bar will effectively "shorten" each leg's exposure to leverage, and add a tension strap that will keep the side legs from spreading as easily. It will now withstand more load from above and also from the side. Call it X+ a Modest Amount.

However, a box is the second weakest of the basic geometric figures (circle is worst). If you then take that hoop with belt bar and add some angled tubes in between, you are SIGNIFICANLY increasing the strength of the structure because you are creating one or more of the strongest basic geometric shape: The Triangle. Call this X+ a Significant Amount.

If you look at them, the main section of all of my bars can be divided into 5 geometric shapes - sub-structures if you will - each of them having their own inherent amount of strength. This effectively decreases the work load of any one section, and transfers any localized force throughout the rest of the sections, unlike a relatively unsupported section of a roll hoop which, with a lack of support or bracing, would simply bend.

My thinking is that this division of labor - so to speak - assures that my bar will take more load from more angles than a simple hoop, or hoop with belt bar... or even same with a diagonal thrown in.

In the event of a mount failure, I still have three left to keep the bar in position. So too does the bar remain a fully enclosed and braced structure within the unibody. I.E. it does not rely on the chassis for its strength. This, I feel, is critical.

I see it as very unlikely that two mounts would fail. The greatest chance of this would seem to be the two lower mounts shearing if the car landed on its roof. In that instance, the sides of the curved roll hoop jamb into the frame rails, and the 10 degree interior braces take up the vertical load.

Well, got to get back to the shop, so that's all for now. Just some of my thinking and how I arrived at the design that I did.

karlooz

hey thx for the info. i'm basically looking for a roll bar for TTs, no "racing" for me in this car. your roll bar looks like it will meet the guidelines my pca region TT guidelines.

kurt M

I never though that there should be any reduction is build strength in the driver compartment. As I said I think the driver’s compartment should be as strong as possible and I would not expect to have intentionally induced weaknesses in the driver’s compartment cage. Having a cage that is tied to two small points and built like a tank is not a good design. It will only move away from the intrusion. If it was made tight to all sides yes it would move and press against the far side but what would direct it to do that or stay there? The mounting system would fail from the movement.

Again, I think a bolt in roll bar like the one being discused herei s an improvement from no bar and even more so if used with a helmet, seat back brace and I think, absolutly required, a helmet. Adding a bar to protect from a very unlikely roof crush and then hitting you head on it in a common rear ender is no safety improvement at all.

Very true there are impact events that can't be protected from regardless of any reasonable build. You can just have too much energy and in the wrong place. Luckily they are the exception and not the norm. You need to build for the 99% not the 1%. You could mount a 20 foot dia reinforced steel ball on wheels and install the driver in the center packed in foam peanuts with a CCTV to drive with. That would have fantastic impact energy absorption cap but would make for an odd car that would be about as slow as mine.

I think you are far more likely to hit something inside of a tumbling car or one that sustains a roof top impact than have the compartment intrude in and hit you. A potato in a trash can type of thing. The can keeps its shape but you are mush anyway. That is the fallacy in the “no harness without a roll bar” argument. You can’t “duck” away from a crushing roof or be gently pushed away, you will be slammed into it by the same energy input that is crushing it from the other side.

Hold the driver in place and keep the area around the driver intact is the key. Let the rest of the car die a valorous death.

mpaton

Well perhaps we will have to just disagree. I think you don't understand the huge amounts of energy absorbed by steel deforming in past the yield point in the stress/strain curve. Lots of energy is absorbed with small deformations. I'm certain I don't want to have a superstrong roll hoop that keeps the car tumbling and can be pulled from what used to be a car and reused for the next accident. That's what I've seen in these dirt oval cars on TV.

And I know that's not what John said about his design; it's just a question I always have when someone tells me only that his rollover protection is good because it's strong. There's more to it than that.

The point about discussing a roll bar or hoop as opposed to a cage is that it doesn't do anything for you in frontal impacts, and may not do much in side door impacts. So that leaves the main thing it protects against as rollovers, impacts on the roof. The stronger such a bar is, the stronger its floor mounts (or wherever it feeds downward loads into the chassis) need to be, or else the bar will be driven through the floorpan (assuming a unibody car).

On the weld in versus bolt in issue, most of the installations I've seen (and that's quite a lot) don't load most of the bolts in shear. Even if they do bolt in, the bolts go through the bar endplates, then the unibody, and then another plate of similar thickness to the bar endplate. A downward load (perhaps I should be saying floorward load) will try to tear this plate through the floor, which is why some plates are L-shaped and also welded to the sidemembers/frame/sills/rocker panels.

On where a roll bar should attach to, I'd agree that seat belt anchor points are better than just selecting a random location on a floor pan. However I'd note that seat belt anchor points are intended to take those considerable loads in a direction usually different, and possibly opposite to the direction that a roll bar will load them. So that location may not be as strong as expected, and may negate some of the strength qualities that people like in their roll bars.

For what it's worth, in the early 1970s, Safety Devices sold 3 lines of roll bars and cages. The one for Club Racing and rallying used seam welded tubing, and FIA compliant cages used cold drawn seamless tubing. I had (and used) one of these and also had, and tested the 3rd line, an FIA compliant aluminum cage for about a 40lb (I think) weight savings. They were very particular about materials, and saw a LOT of results of accidents, and I never heard of the seamless tubing causing any problems.

Michael Paton

cooz

got one for a 993?

RedlineMan

Just to be clear in case you were speaking directly about my designs, THEY ARE NOT mounted in TWO places. They are mounted in FOUR. It is very easy to criticize a design if you do not know or realize the parameters of its execution. Coming up with a product that does not require ANY modifications to the car, and whose presence CANNOT be detected when it is removed, is a bit trickier than you realize, metinks. Again, IF you were speaking directly about my designs.

To mod your analogy, take an egg and put a driver inside. Crash = Smash to everything. Now, build a cage inside that egg, not even attached to the egg itself, and the driver is restrained to the cage. The egg will smash easily, but the cage will protect.

Silly analogies aside, my point is that a strong structure inside a unibody is not worthless even if it does not directly keep the unibody from deforming. The structure inside will impede that deformation at some point, and to some degree, and this is a good thing, no? Keeping the driver from hitting that structure is another matter.

Precisely why I do not do floor mounts. It takes a lot more than backing them with a plate to make them strong. The floor is the WEAKEST part of any unibody. Besides, the main forces are not going to pull up on a roll hoop in a rollover. They will push down. In any event, a floor is not structural regardless of how big the plate is.

I would not assume any time was spent by Porsche considering the directionality of their belt mounting points. I would assume that they simply made sure that the panel they mounted them in could sustain the required load.

Further, in regards to 911 and 944 rear outer belt holes, they are placed in the strongest portion of the entire unibody. If you have ever dissected one of these chassis, you will see that the steel is multiple layers thick, and the thickness of each piece exceeds normal 20-22ga. sheetmetal. This outer rear mount is far and away the strongest belt point of any of them. They are actually in thicker stock than the suspension mounts! FAR exceeding the strength of the front lap belt mounts as well.

Good stuff guys!

kurt M

I think folks are missing my points. I am not trying to pick on this bar and am only discussing bars in general. The discussion of this bar was per the builder’s original request for input.

1. I think a bolt in bar has to have compromises but is better than no bar.

2. I think this bar is better than no bar.

3. Bolting a bar to the floor pan is stupid and this bar is not.

4. The 1# safety addition in adding a bar is being able to add harness and race seats. That is the real world safety improvement from adding a bar. 99% -1% rule. Roof crush is rare, driver flopping and hitting something inside the car is not.

The main hoop has no attachment points in line with the hoop, all the bolt points are off the plane of downward or side force. Push hard on the side of the main hoop and the near side seat belt bolt will fail in single shear regardless of the sheet metal it is bolted to and the bar will deform and displace. The two rear points are far off the plane as well and will help in keeping the bar upright but are disadvantaged in a side load and will do little in crush loads other than force the bar forward if the bar feet are compressed outward. Side impact is not what the bar is made to prevent so it is fine in that regard.
The feet are angled. What it to the outsides of the angles? If it is not strong and can resist penetration then the bar will want to splay apart. You could resist this by having an added cross bar that was low and near the feet. Can you add a curved or otherwise designed plate to the feet to spread the load out and help prevent penetration and or splay?

All I was saying with regards to the added tubing is there is little reason for having a weight lifter stand on a milk crate. Build to improve the weak points. Over building a point that is not the weak point does not improve the unit overall it just adds weight and cost.

NASCAR and dirt track cars have very strong rigid driver’s compartments. #3 aside they walk away from hit, flipped, rolled, spun, all of the above and then some trashed cars all the time. They expect to have this happen sooner or later and build for it. Personally I would not want to be directly attached via harness to something that is being deformed or displaced. Then I am being displaced and deformed. That is the cars final duty, not mine.

RedlineMan

Hey Kurt;

I don't dissagree with a lot that you are saying, although a large percentage of it is pure surmize. The only way to know what actually happens is to crash it, and I personally hope no one ever tests anything I build. For these reasons, I tell people that they should look at these products as a really good way to mount a harness. Any other assumptions of worthiness are the property of the owner. I explain my thinking, and the rest is up to them.

Another thing I have learned is that the difference between commenting on something, and then pulling out some seats and having a go, are light years apart. When you actually start trying to figure out how to do it for real in the car of your choice, theorizing takes a definite back seat to reality.

Gutting a car and building a cage is the easiest of all tasks, and even then takes a fair amount of skill. Building something inside an exisiting interior compounds the problems faced MANY times over. Trying to do this without requiring modifications is truly a stout challenge.

For that reason, I tell people that if you take any aspect of one of the bars, you can rest assured there is a reason it is why it is. Whether you can figure out what that reason is or not, there is a reason.

Not sour grapes, just experience!