AVUS Grosser Preis von Berlin 1991 [FINISHED, RESULT INSIDE]

[strop]

All the difficulty with the reliability would be easily fixed if we had the value displayed in the engine tab (which it used to be in previous builds. This will be amended in future builds, I believe). For now, though, this is a surprisingly difficult issue understood with certainty by no players (including myself). Aside from time constraints that's why I didn't enter the last round: while the 50 reliability figure is good, it's too contingent upon other factors to reflect what you really want. But since I was playing this a lot before, I can offer a bit more when it comes to how it worked:

What I can see without any calculation is that the current rule means that 1/2 * cooling * 50.0 reliability = 25.0

That means, if car reliability and engine reliability are both 50.0, the resulting reliability is 25.0

Erm... time for more confusion!

If the car reliability and engine reliability are both 50.0, the average reliability as displayed in the testing tab will be 50.

But if the component reliability is 50 and the engine ventilation is set to exactly half that which is required, then the engine reliability will be 0. Which will automatically set the car reliability to 0.

If the component reliability is 50 and the engine ventilation is set to exactly 75% of that which is required, then the engine reliability will be 25. And then the average reliability will be 37.5.

The best way to figure out the average reliability is to set the cooling to greater than minimum (just to make things harder, there may be in some cases a bonus to overall reliability if you use more than required cooling). Then, you can see the engine reliability and the total reliability is maximised. The component reliability, therefore, is:

  Code:

Engine reliability + 2*(Total reliability - engine reliability)

Or 2*Total reliability - Engine reliability.

The one thing I haven't verified is if component reliability has anything to do with Environmental Resistance. Env Resistance is most largely influenced by the material type and quality of the chassis and panels, which is why reliability is the most affected by changes to that.

/inorite

[BurningBridges]

Another way is to just calculate effective engine reliability which can be done easily, and simply ignore mechanical failures.

[strop]

It's simpler, to be sure. No tournament that I know has even bothered to look beyond the overall reliability (Der Bayer's doesn't count, it has access to secret code business). In this case the engine reliability is particularly pertinent due to the high engine speeds.

[BurningBridges]

Just quickly redid the MTBF calculations tor the 24h race

You see even with MTBF 100 you have a realistic chance to fail. That is simply because it's not a % chance, but a 100,000km mean failure rate.

mtbf_6225.jpg (98.34 KiB) Viewed 6266 times

[Boksi]

Oh hey, I didn't finish last place despite submitting a pickup truck! Nice.

[BurningBridges]

YES that was cheeky, and you absolutely got away with it. Well done
Also where is the 300 km/h truckari we ordered?

[BurningBridges]

I made a spreadsheet with some examples, it shows what happens if you have 50%, 75% or 100% cooling and how it would affect engine reliability. I calculated engine reliabilities from from 10 to 100, and only >50% required cooling, because 50% is the minimum.

And if Excel is too much hassle, here this is basically it:

Clipboard01.jpg (83.8 KiB) Viewed 6228 times

[BurningBridges]

All right, I think I have a simple way how to make this work.

In a 24h race all cars will have to make lots of stops anyway to refuel (between 5 and 25 stops).

Now once we have calculated the chance to fail (see above, in the example CTF varies between 6 and 70% with the most likely values clustered around 15%), a certain amount of time can simply be added to the pitstop time.

That means cars would be repaired during the regular refueling stops, and no extra pit stops occur.

What I like about it is that cars have no fixed reliability requirement but high Chance to finish is recommended, because otherwise you will lose a lot of time during pitstops.

P.S. The actual formula for pit stop time per CTF is not ready and there has to be a test race. But it can be a simple linear relationship because the MTBF calculation is already highly exponential at the extremes.

[BurningBridges]

I suggest we keep it very simple: For every point in Chance To Fail, a flat time of 1 minute would be added to the total pit time.

That means you can easily check whats going on by looking in the right column of the chance to fail calculation. The number for Chance To Fail is equal to the number of minutes that you lose for repairs.

Or in terms of what you can expect: A car with average reliabilty would lose ~20 minutes, a car with minimal reliability and ventilation ~70 minutes, and a car with maximum reliability and ventilation ~6 minutes.
(All over a 750 lap endurance race which is 6225km or ca 24h)

[BurningBridges]

I checked again and it wouldnt be necessary that the "Overview" panel works. This was initially a requirement as I needed the fuel consumption per 100km, but I can get this just as well from Detail Stats even when there is nothing in Overview.

This means very low reliability values can be allowed, even if this isnt the best idea because of repair time. The only snag is that people can now submit cars with the lowest reliability values and lose an hour or so in the pits But I still think I will go this route, as it adds a nice unpredictability, at least when you dont track the provided tables. Its a race afterall and many things can go wrong. Also it's a 24h race (and over 6000km distance) so reliability does really matter.

And if you do look at the tables you will be able to see the spot where you should be considering reliability and ventilation. Go either too high or too low and you pay, either in lap times or pit times.

[BurningBridges]

A first draft for the 1996 rules, all in one page (click on the image to see all)

Avus_24h_1996_1.jpg (200.36 KiB) Viewed 6158 times

C30 is a bit of a reroll of the 1990 rules ), with more room for the N/As. I reduced the cost limit a bit to make this more interesting and viable for amateur teams.

The turbo factor of 2.4 is more or less perfect, as was shown by last years result where a N/A could narrowly win a race. I rounded it up slightly to 2.4 to get even numbers. Keep in mind that this adds a further 1% to the N/As which is small but not insignificant.

Also keep in mind that you have no parts restrictions whatsoever and can go really low with reliability (although at an increasing cost in pit time).

I am just not sure about the tyres, I may use something from turbosquids formula to reduce this a bit (like +2 / +4)

Also 1L, 2L and 2.4 could very well be 4, 6 or 8 cylinder engines, but a V12 could be the best option for the 4.8L. If one considers 500cc per cylinder as the optimum, 4.8L would be somewhere in the middle between a V8 and a V12, (a V10 actually) so make of this what you want, but I would probably go to a (more expensive) V12.

The detailed descriptions of the fuel and repair calculations are missing, but will be provided once this is up and running. For now just let me say that refueling takes 2:00 minutes as in 1991, and every point in Chance to Fail adds another 1:00 minute to your total race time.

EDIT: I made some important changes.
1. Only front engined cars in the "amateur" class (C30)
2. Different repair and pit stops times.
The two classes are now quite different. You have less money in the C30 and reliability and fuel consumption have more of an impact. Also, we have a class that can no longer be dominated by rear engined cars.
So it is now longer a purely random choice if you join the smaller or the bigger class.

[BurningBridges]

And here an updated table that caps engine reliabilty at 80 (I have never seen higher values), and shows what happens when you run with low (25%) or extremely low (10%) ventilation.

A car without any ventilation will be disqualified.

More or less the same happens with ventilation < 10%. You would get a resulting chance to fail close to 100, which I consider impossible to compensate (100 minutes pit time --> 8 seconds per lap).

chance to fail table 2.jpg (178.87 KiB) Viewed 6180 times

[BurningBridges]

Pit time per % Chance to Fail could be increased to 2:00 minutes, so the seconds for pit time would double. There seems to have been a mistake in my calculations, 16 seconds per lap for the practically unventilated / unreliable engine is what I wanted, not 8.

Also requirement will be > 0.0 reliability for engine and cooling >0. That way I can make sure no one gets away with something outright crazy.

Otherwise it should work as intended.

EDIT: still open for discussion as such values are of course very important. Somewhere between 1 and 2 minutes should be right

chance to fail table 3.jpg (179.54 KiB) Viewed 6179 times

[squidhead]

You'll need to run a simulation of a race to figure which works which doesn't I think

[BurningBridges]

Both would work. It's important that all cars lose time during pit stops. Even a car with 70 reliability and 100% cars need 1.4s per lap repair time.
I just want to prevent anything funny going on at the lower extremes.

I only need to put the two columns side by side and see if there is something unusual. Or perhaps take the value in the middle like 1:30

A simple way to test this would be to have three different classes, with different values for repair time.
Lets say the amateur class gets 2:00 minutes, the intermediate 1:30 and the manufacturers 1:00