Engine heat
2
(49 posts, started )
Would I be right in thinking, that at the moment everything has a strength, which determins how much it bends under an impact. Shouldn't it just get weaker the more it is damaged? It might do already, I haven't really tested, but I would have thought that would be all need.
Quote from sinbad :Totally agree. Random failures do not exist in real life. If a part breaks when another "identical part" does not, when used in the exactly the same way, it's because the parts were NOT identical.

Outside of computer games, there is no such thing as an identical part.

With parts that see fatigue (uh, the ones that break and end your race) the state of science doesn't come close to allowing someone to define the life of a part with 100% accuracy. Since race cars are by definition built on the edge, it is very realistic to base failures on a statistical model. Sure, the input of the driver needs to be taken into account, as that obviously influences things.

This doesn't change the fact that in real life, someone will lose a race because a part failed in a mode the designers either:
-didnt anticipate at all, and may have never seen in many hours of testing.
-knew may happen eventually, but the chance was determined to be small enough to be an acceptable risk.

Quote :
If something happens to the car they can see what caused it, they don't just go ahhh well it's random and we were unlucky.

As stated, there are plenty of failure mechanisms that aren't thoroughly understood. The cause and subsequent solution might not be forthcoming. Even the very best designs sometimes fail unexpectedly.
Quote :
Same with punctures. No puncture is random, you get a puncture because you drove over something sharp.

I don't think you fully comprehend the concept of "random". Something that is "random" can be described by the science of probability and statistics. Assuming no user error (driving off the racing line) suffering a puncture is about as random as an event can get. You do get the puncture because you drive over something, but that something being where it was at that time can be described as random.
Quote :
If you want to enjoy the forced dnfs of engine blowups and the like now, in the same way as they'll be done if it's random, you don't need all that stuff coding in. Just put all your names into a hat before the race, and the names that are pulled out will not finish. It's that simple, and that silly, and adds nothing to the experience.

Well thats obviously wrong. There is a huge difference between a race in which a competitor doesn't start and a race in which a competitor retires on lap 32. This difference might very well effect who else finishes the race and the order they finish in.
It would seem to me that the factors that come in to play with mechanical problems (Ones not caused by accidents etc.) are mostly influenced by things happening outside the track, ie. how much time and money the team spends on testing and hiring engineers, and how they balance reliability vs. performance. To simulate this properly you would propably need some sort of race team manager game integrated into LFS where you can make the decisions on who to hire and what to aim for

Because of that I propably wouldn't like a "random" failure system that I cannot influence in any way (Although in most real teams the drivers propably cannot influence it that much either but still), but maybe something like a slider in the setup menu that allows me to choose between max reliability and max performance, that could give me a few extra HP's but with the risk of a mechanical failure...
A important variable is, as Kegetys suggested, the setup. I go a little farther and say that all the variables in the setup should have a point beyond wich is dangerous to go (to muchi bracke pressure aplyed at full bracke = possibility of the bracke to jam... etc)

safe unsafe stupidly unsafe (4 fun)
[-----------------=--|---------|-----]

I allso like the idea with "real mechanics", but there should be different modes: full mechanic team wint statistics (all the mechanics and telemetri ppl), fluu mechanic tema (the telemetry controlled by the mechanics), one mechanic (that controlls the pit stop and telemetry) and self mechanic (self explanatory )
It isn't a long way from reliability slider to engine tweaking, you just set the power output/reliability and hope it will last through the race. Like in the other end you have Renault F1 and in the other one McLaren F1. Other is fast but unreliable and the other one not so fast but reliable.

As to the whole thing about mechanical failures and its randomness, almost all mechanical gadgets, for example pistons, crankshafts etc. are engineered by using some probability functions and graphs. The variables are: frequency (if moving part), forces affecting to the part, part material characteristics and the form of the part itself. Different materials can handle varying loads very differently.

There are some engineering methods like: safe life (part won't break down in certain time in certain conditions), fail safe (the system is allowed to break at some points but the system is still operating) and damage tolerance (certain parts are changed at times to prevent any damage coming critical in the system). Why am I speaking about engineering methods...read on->

So what kind of engineering methods are used in building the lfs cars? Do we want to inspect the car at times to ensure it is OK, or do we just change everything between the races? Real race cars usually get put into pieces between the races to check and to change some of the parts. And some of the parts may have some errors in it that we can't really see and be sure. Here we use some probability graphs and equations to check if the part probably will last the next event. So they are using probability functions! So actually the mechanical probs are random to some extend.

If we don't want to completely model the physics related to part mechanical failures we must use some kind of probability equations. The variables in these functions cover the estimated part life (how much stress the part was engineered to handle, i.e. the resistance to break down), forces affecting to the part (rpm, load, heat...) and a memory that keeps track of the usage of the part (part condition in general).

So the "mechanical failure system" makes all the time some adjustments to the part's condition and the part will break sooner or later. A good system that does all this has to handle a lot data as there are many many parts in a single engine that can break.

The guys developing the "Racing Legends" simulator threatened to make a sim where all damage is physically modelled. The problem is that we can't make 100% sure damage modelling even in the best commercial "physics simulators" (like Femlab and such). So the only way to model the mechanical failures is to use probability functions where there are some factors we can have some effect (like clutch usage and driving line (over the big curbs or not...).

If some one even reads all this

EDIT. I just red this and it seems that I forgot the point there .

So what I was implying was that if we want to have as realistic damage modelling as possible it will be random to some extend. The randomness derives from the parts and their little deviations (cracks, material properties etc...) so each part is little different and acts little differently. Also there are human errors as some one said earlier which are impossible to predict if we had them in lfs. Btw. should some cars be more reliable than others? What about damage that couldn't be repaired? Should those human errors be in lfs? (even if they were totally random, bad luck, as you can have no influence to these)

Yeah, point made
I think the reliability of internet connections & computer reliability covers for any "random" or "no-fault" failures in LFS. Your connection drops out during that big race? Maybe the throttle cable broke - or you hit a Kangaroo!

Failure due to driver abuse should be(and some already is) modelled. Overheated tyres, engine damage from over-revving, Suspension damage from violent kerb-hopping etc...
Quote from Sk0rp :According to that I would like to have technical failures like engine-blowups, gear change problems and something like that. The frequenzy of the failures should be ajusttable from the server site.

Nope. Failures should only happen when bad driving is taking place, or cars are run hard for long period of time
The question is, how do you decide on defective components? There are no other ways than randomly and at set times or intervals. I agree that failures should happen from bad driving, but they should also happen from defective components. And that is where the "random" element pretty much needs to come in. It can be rare, most car components do not break, but it would be more realistic (not necessarily fun) to have component failures through no fault of the driver.
Disagree with you, James, well it had to happen some time

I think thisd is one aspect, when on-line racing, where fun should take the priority over realism. Bad things should happen only as a a penalty for careless or poor driving.
Couldn't we end up with an option?

Everyone here probably knows how much I'd like to a 'Hardcore Simulation' mode. In this mode, perhaps have 'random' failures, so that in every 1000km of driving you will probably get a failure of some part. Maybe none, maybe two. It might not even be a failure that ends your race, but might slow you down, or require a change of driving style. This would be in addition to other factors, such as tyre blistering, suspension/track rod 'wear', engine life, electronics failure through vibration etc etc

Then we can have a less hardcore mode, made more fun by the lack of random failures (which most people would agree aren't fun, even if they are a part of real racing). I myself would use the less hardcore mode now and again for fun racing.

This way we get the hardcore simulation, but also an option to make the learning curve, and frustration ratio to a more acceptable level.

Everyone's happy, and LFS still retains it's crown as the most realistic simulator (assuming a competitor doesn't come and blow LFS away!)
I agree with Tristan, I think there should be a setting. I actually ENJOY having random failures. Yes they are frustrating, but that's motorsport for ya. There's nothing like crusing around in first (my case, last) position and then suddenly you find you had a defective piston head and your engine just sort of fizzled out. BUT, it would be quite unfair and no fun for some other people (and for me after a while ). So a server-side option might be a good idea.

Quote from Al :Disagree with you, James, well it had to happen some time

Yeah, doesn't happen often - does it?
I dont think random part failures are the way forward at all.
Driver induced failures i'm all for but 'part failure' is going to be nearly impossible to programme. Scawen would have to seperatly program the physics for each component of the car and work out how wear n' tear affects each part aswell as how wear n' tear on the part attached to it affect its stength and so on.... my god LFS would NEVER be finished.

Anyway that all mostly happen because of human errors, we can trace most mechanical failures back to a human error of some form. Does that mean we need to programme a virtual human brain to decide when he should make a mistake in a part that might fail somewhere down the line in a long race?

But simple driver induced failures, they are what we really need. If your whacking curbs out of it then your more liely to get a puncture of suspension failure, or debrit on the track will cause blow outs and so on.
Anyway 'random' component failure is EXREMELY rare in motorsport, infact i would like for someone find when a 'random' mechanical failure happened in a motorsport event that didn't happen because of a human mistake.

I have raced for 13 years now and never once have i had a 'random' mechanical failure. Every failure has been down to a human error. Nearly all teams will replace parts that are prone to breakdown (maybe because they are a weak part *again a human manufacturing mistake*) regularly to save from mid-race failure.

So lets just stick to the simple stuff, collision, debrit and other driver induced failures. Like someone said above we have computer component and connection failures to make up for the other things.
Well, what about a simple solution? In server-list will be displayed if the server has enabled "random" faliures(i will call them random, because it is impossible to simulate work of mechanics adn teanm in any videogame today). If you wanna to risk a faliure and get yourself some more fun, you will pick this server. If you wanna to have sure finishing, you will have to pick server with disabled random faliures.

Humans MAKE mistakes, and there is NO difference if it is caused by a real human or by random generator. And because CPU is not human, in LFS it will be by a random generator.

And when we are talking about random faliures, did you see Kimi Raikonnen? His engine came down in 4 races and once his suspension broke down in last lap. Are these faliures caused by some human mistakes?(ok, maybe the suspension is)
Some would argue that the engine failures were too (and may have been).

But human are not responsible for every failure. Sometimes there are conditions, situations or whatever that CANNOT be designed for, and defects in materials that cannot be seen using reasonable (even F1) methods.

Therefore SOME failures have to be modelled as random to be realistic.
I agree with those who say "random" (as in simulating manufacturing flaws or badly designed parts) should be optional or not included in LFS at all (mandatory only in the hardcore mode described by tristan if it becomes reality). IMO this type of random failures would detract from the fun (if they happend too often racing becomse frustrating and if they are rare it might become pointless because we already have failures from lost connections etc.). Sure it would make LFS more realistic but do we really need it when we can have driver induced failures?

Driver induced failures should definately be in because they have only positive aspects (more realistic and challenging without the "why did my car randomly stop!?!, this game sucks" type of effects). Also random (manufacturing flaws or human error) part failures probably won't be missed because so many failures can be driver induced.(off the top of my head: flat tires, tires coming off the rim if you hit kerbs too hard, engine damage, gearbox damage, clutch damage, suspension damage, exhuast damage from heavy landings in rallycross, brake failures, bodywork parts falling off etc.)
And who says that you will have a "random" faliure every race? When I played Grand Prix 4 where are faliures implemented quite good, it wasn't frustrating. I had some defects, but it olny added some realism to game. and some fun too. That feeling of possible problem kept me still aware of everything what I am doing.

Yes, 2 defects in 5 races WILL be frustrating, but one "random" faliure in 40 races won't hurt. It can happen, it'll be real, LFS WANTS TO BE REAL!

BTW: Again, who says that the faliure will everytimes stop you? Overheating engine only may force you to slow down...
Quote from nikimere :
I have raced for 13 years now and never once have i had a 'random' mechanical failure. Every failure has been down to a human error. Nearly all teams will replace parts that are prone to breakdown (maybe because they are a weak part *again a human manufacturing mistake*) regularly to save from mid-race failure.

What do you race?
In the end everything can be explained as human error. Shifting error, too high rpm, manufacturing error, designing error... Maybe it would be best imho if we had the random failures minimal whatsoever. Because people hate them, and there is no way how to evade them (manufacturing errors, designing error...). If there are any in LFS . Of course I'd like to have it, but then we face the situation where other car is more reliable than other...

But for how long is the F08's engine meant to last? One 2 hour race? 24h endu race? There are a lot variables to think. But I would make it that the probability of damage would rise as the time goes by in a race. Of course not time related but
...in engineering people think the parts life as cycles (hertz as a unit). Farming machinery are engineered to little amount of cycles and car engines to infinite (infinite as 10^9 iirc) cycles. So a car's engine has to be able to take 10^9 rotations before braking is probable. Race engines are in the high levels (F1 for example) very precisely engineered to last the minimum while getting the max power.

So make the XRR's engine blow up more often because of it's size and turbo. And the FXR's awd might loose some drive shafts. And let's have some electrical errors too! Lights go off, flashing, engine stalling, gauges gone crazy... Tyre probs: manufacture fault in 1/10000 tyre: will blow up any time! Shouldn't pitting also have some random failures: Fueling pump broken, broken bolts etc...Maybe a 1/1000000000000 chance of having a little dip of sand in the fuel
Yeah I see what you're getting at Hyperactive. If we were racing a specific thing (such as F1 cars in Grands Prix, then 'random' failures are easier to program because they can be based on real life, and the races are of a similar nature.

But with LFS some people might want to do 24hr FO8 races, and if the engines were lifed then everyone would retire by 3 hours. What to do for 21 hours?

There is still a big part of me that would like some form of slightly random failures, but they'd have to be carefully implemented (and ONLY in a hardcore mode if there was one) so as not to stop certain forms of racing.

And Nikimere - if you've never had any failures like that, I think you've been very lucky, or very skilled, in getting yourself very good engineers/mechanics/team members etc. I would be interested to know what you race as well, so that I can come and put file marks in your conrods
Quote from tristancliffe :
But with LFS some people might want to do 24hr FO8 races, and if the engines were lifed then everyone would retire by 3 hours. What to do for 21 hours?

So you havent to pish it to the limits Racing engines can "take" the heavy loadement during the race. Engines for 24h Le Mans are designed to work for all the race. But racers AREN'T pushing them ti the limits. So you will not rev your FO8 to 10k RPM, 7k will be enough, you will brake 50 meters earlier, you will not floor the throttle for all the race and be sure the car will survive. Long races aren't only about speed, they are about sensitive driving, aren't they? And only possible faliure will force you drive with sense. GIVE US FALIURES... plz...

AND A LITTLE PHYSICS LESSON:
Random does NOT exist, if you will be able to give attention on EVERYTHING, every particle in space, you WILL be always able to presume every "random" event with 99,5% percent accuracy. (if you will use superstring space theory for calculations) Because anybody is not able to do that, we call hardly presumable events RANDOMS....
Quote from MadCatX :So you havent to pish it to the limits Racing engines can "take" the heavy loadement during the race. Engines for 24h Le Mans are designed to work for all the race. But racers AREN'T pushing them ti the limits. So you will not rev your FO8 to 10k RPM, 7k will be enough, you will brake 50 meters earlier, you will not floor the throttle for all the race and be sure the car will survive. Long races aren't only about speed, they are about sensitive driving, aren't they? And only possible faliure will force you drive with sense. GIVE US FALIURES... plz...

That's not it. In Lemans the drivers won't brake 50m before they normally would, neither they use only 7k out 10k rpms. The engine and the car alltogether has that kind of parts in it so it can take the stress the 24h race puts on it. The drivers drive so that they can drive as fast as they can without too much risk of spinning etc. Also they do try to put the engine to as little stress as possible, but they still go as fast as they can. But we are talking a reductions about measured in hundreds of rpms, not 3k

To sum it up, the car has been made more reliable which means generally slower. But there is still a chance of breaking, otherwise the car won't be competitive.

Racing engines have cope with huge loads. The rpms are very high all the time, the parts are made as light as possible and the loads are quite big and the engine is under stress all the time. In normal car your rpms rise only when you're accelerating, racing is all about accelerating. Think about the stresses the gearbox, clutch and axles have to cope with. All the time accelerating, slowing down or speeding up. If there is a small crack in the cylinder block it will break down after some time. The probability depends how soon it is noticed and for how long the part is under stress. Also the type of stress counts, but we don't want to go deeper into this... The longer it is under the stress the greater the probability it will break.

It is all about a ratio which includes money, time, technology and use. Money stands fr resources, Ferrari has better possibilities to make a better engine than Jordan (for example). Time means the time available for the project. If you have lots of time you can make a perfect engine in 1000 years. When you cut time you leave something untested. Technology mean the facilities, BMW has a windtunnel inardi doesn't. Use means what kind of stresses the part goes through during its life. That's why a car than has been driven for 1,000,000 kilometers breaks done more easily than a car driven only for 1,000km. But if the latter engine has been in race use, it will probably break down earlier, because the stresses are many many times bigger...

Quote from MadCatX :...we call hardly presumable events RANDOMS....

Exactly!
Quote from MadCatX :
AND A LITTLE PHYSICS LESSON:
Random does NOT exist, if you will be able to give attention on EVERYTHING, every particle in space, you WILL be always able to presume every "random" event with 99,5% percent accuracy. (if you will use superstring space theory for calculations) Because anybody is not able to do that, we call hardly presumable events RANDOMS....

Did you just suggest we apply superstring theory to race engine parts modeled in a PC based simulator?

I'm not sure if thats funny, or just really silly.
Now you've mentioned I bet Scawen's nipped down to local library to get Superstring for Dummies and DirectX :P
Quote from skiingman :Did you just suggest we apply superstring theory to race engine parts modeled in a PC based simulator?

I didn't suggest to use this incomplete super-complex-hyper-hard-to-calculate theory in games. I only tried to explain WHY randoms does not exist. When you can presume them(and with correct calcultion algorythm or how I should call it you CAN), then they aren't randoms no more, because RANDOM = Unpresumable event.

(Yes, I know about Brown's uncertainty principle, but don't make it so complicated)
2

Engine heat
(49 posts, started )
FGED GREDG RDFGDR GSFDG