2014 GTP 24 HEURES DU MANS [13:00 UTC 12 JULY]

  • Thread starter AJ
  • 756 comments
  • 45,835 views
I vote let it rain. We have plenty of time to practice in the rain and if needed we could throw out a yellow couldn't we?
 
Warning: TL;DR ahead.

Incredibly curious of exactly how the weather system works, I decided to systematically go through and analyze the weather system and patterns and here's what my research currently suggests.

There are 3 clocks running at all times in the game: Real Time (RT), Track Time (TT), Weather Time (WT). The two common times are obvious, the RT keeping track of your laps, and the TT keeping track time of day and can be changed by using Time Progression (TP)... nothing new there. However, what I hadn't fully had a grasp on was WT (mostly because I hadn't really played with it before, I don't usually host races.)

From the two descriptions of Weather (W%) and Weather Changeability (WC), you can set the percentage of "Weather" at the start of the race from 0% (no weather) to 100% (monsoon), and it will stay there for "a while" before beginning to change "at random." The "at random" depends on WC, which is vaguely described as the randomness with a higher number meaning more crazy weather changes... but that description is not entirely accurate... what it is, I've discovered, is the TP setting for WT. (Some of you may already know this, I did not, that's not entirely shocking but the implications of it are quite handy.)

If TP and WC are set to the same number, the rate of weather change will always be the same according to TT. For example, If you start a race with the following settings:
W%: 100
WC: 1
Surface Water (SW): 0
TP: 1
...the course will obviously start with a torrential downpour, and will gradually increase SW to reach 100% after around 30m, with a slightly random variation of not more than a minute. However, if you change both the TP and WC to 60, it will reach 100% in RT:30 seconds... which happens to be TT:30 minutes. This is true across the entire 1-60 span of both options, as long as they are the same, it will take around TT:30m for the SW to reach 100%.

You will note, that every time you double TP, you halve RT compared to TT. This results in a non-linear graph with the largest differences between RT and TT being at the smallest TP settings. TT:24h takes RT:24h at TP:1, only RT:12h at TP:2, and RT:6h at TP:4, a difference of RT:18h in the span of TP:4... but the entire range of TP:30-60 exists between RT:48m-24m. Everyone knows this, but this exact same non-linear graph is how it works for WC as well. SW:100% only takes RT:15m at WC:2, RT:7m30s at WC:4... etc... This is why changing the WC even slightly results in huge changes in the weather system... just changing WC to 2 doubles WT.

The other factor in the changing weather is the randomness factor. Again, from the description, initial weather will stay the same for "a while" before beginning to change at random. I do not know what exactly how long "a while" is... as it's likely a random WT number and I have only spent 1 night researching this so far, it would require a much larger sample set of experiments than I currently have. However, I do think that this amount of Initial WT is large compared to the rate of change after that time has passed. After this point has passed, it will begin to change based on a certain (or random) amount of WT. The amount of change, I believe, is based on a probability bell curve.

Anyone familiar with dice based games will know that if you roll two standard 6-sided dice, the most likely number you will roll will be 7. 2 and 12 are the least likely numbers to roll (this isn't exactly a bell curve, but it is if you add more dice... not really the point... I'm digressing and trying to use an example...). To overly simplify the algorithm that GT is likely using, let's just say that after the Initial WT is over, every random WT sequence GT rolls a pair of dice. If the number rolled is 7, the weather stays the same. If the number is higher than 7, weather increases; lower than 7, decreases. The amount of increase or decrease is based on how far away from 7 the actual number rolled is, with 2 and 12 being the most extreme changes. (Again, this is an over simplification and not really how it works... it just works the same way in principle...)

Now, if you only roll the dice a handful of time, there's a pretty good chance you'll never roll a 2 or 12 and have some sort of extreme weather change. At TP:1, you only roll the dice a handful of times. However, over a given amount of RT, you will double the amount of rolls every time you double the WC factor. Given a large enough number of rolls, it becomes increasingly likely that you will experience a few 2 or 12's... This is not to say that with TP:1 you would never roll a 2 or 12, it's just not likely. If you do roll a 2 or 12, it's also not likely that you'll roll another, and even less likely that you'll roll the opposite. This is why a race could start with a low W% and TP:1, and end up having TT:17h+ of rain, because perhaps a 12 was rolled so the weather spiked up... but then every roll after that just hovered around 7, and so the W% never decreased.

Also, because the W% can't go any higher than 100% or lower than 0%, if it ever reaches either of those extremes, it becomes exponentially more likely to stay there. If you have 100% weather, every roll of 7 or higher means the weather stays at 100%... if you have 0% weather, every roll of 7 or lower means the weather stays 0%.

As for testing, if you want to have an idea of what the 24 hour race will be like with TP:1, just set the W% to what you want, and set the TP and WC to 60. This will reflect the same number of "dice rolls" as it would with TP:1 over 24 hours. If you set the same W% and do it again, you will probably get a different weather pattern, because the changes are random, but roll frequency will still be the same. However, because you can't raise WC any higher than 60, to represent TT:24h at TP:2 in the least amount of RT, you need to decrease the TP to 30. With TP:30 and WC:60, it will take RT:48m to accurately recreate what the weather would be like over TT:24h, TP:1, WC:2.

Now, as for my personal opinion, I think the best thing to run is a low percentage W% (like 10-20%) with TP:2. This makes it unlikely that a torrential downpour will happen... and quite likely that the weather would stay fine the entire time... but it still leaves the chance for rain. And, yes, there's even a chance that it will begin raining and never stop... but at least at TP:2, it takes RT:15m to go from SW:0-100% in the event that it does suddenly become a monsoon. Cars become tough to drive on slicks around 20% (~3m) and next to impossible to drive at 50% (~7.5m), which is plenty of time to get around the track and back even if you were unlucky enough to literally have just passed the pit lane when it opened up... as long as you know not to slam down the gas coming out of every corner...

I encourage people to test this out or prove me wrong if you'd like... This just seems to be the most logical explanation based on the evidence that I've acquired thus far... I plan on eventually running some more tests... but now that I understand how it works I'm less curious and motivated. :sly:
 
Warning: TL;DR ahead.

Incredibly curious of exactly how the weather system works, I decided to systematically go through and analyze the weather system and patterns and here's what my research currently suggests.

There are 3 clocks running at all times in the game: Real Time (RT), Track Time (TT), Weather Time (WT). The two common times are obvious, the RT keeping track of your laps, and the TT keeping track time of day and can be changed by using Time Progression (TP)... nothing new there. However, what I hadn't fully had a grasp on was WT (mostly because I hadn't really played with it before, I don't usually host races.)

From the two descriptions of Weather (W%) and Weather Changeability (WC), you can set the percentage of "Weather" at the start of the race from 0% (no weather) to 100% (monsoon), and it will stay there for "a while" before beginning to change "at random." The "at random" depends on WC, which is vaguely described as the randomness with a higher number meaning more crazy weather changes... but that description is not entirely accurate... what it is, I've discovered, is the TP setting for WT. (Some of you may already know this, I did not, that's not entirely shocking but the implications of it are quite handy.)

If TP and WC are set to the same number, the rate of weather change will always be the same according to TT. For example, If you start a race with the following settings:
W%: 100
WC: 1
Surface Water (SW): 0
TP: 1
...the course will obviously start with a torrential downpour, and will gradually increase SW to reach 100% after around 30m, with a slightly random variation of not more than a minute. However, if you change both the TP and WC to 60, it will reach 100% in RT:30 seconds... which happens to be TT:30 minutes. This is true across the entire 1-60 span of both options, as long as they are the same, it will take around TT:30m for the SW to reach 100%.

You will note, that every time you double TP, you halve RT compared to TT. This results in a non-linear graph with the largest differences between RT and TT being at the smallest TP settings. TT:24h takes RT:24h at TP:1, only RT:12h at TP:2, and RT:6h at TP:4, a difference of RT:18h in the span of TP:4... but the entire range of TP:30-60 exists between RT:48m-24m. Everyone knows this, but this exact same non-linear graph is how it works for WC as well. SW:100% only takes RT:15m at WC:2, RT:7m30s at WC:4... etc... This is why changing the WC even slightly results in huge changes in the weather system... just changing WC to 2 doubles WT.

The other factor in the changing weather is the randomness factor. Again, from the description, initial weather will stay the same for "a while" before beginning to change at random. I do not know what exactly how long "a while" is... as it's likely a random WT number and I have only spent 1 night researching this so far, it would require a much larger sample set of experiments than I currently have. However, I do think that this amount of Initial WT is large compared to the rate of change after that time has passed. After this point has passed, it will begin to change based on a certain (or random) amount of WT. The amount of change, I believe, is based on a probability bell curve.

Anyone familiar with dice based games will know that if you roll two standard 6-sided dice, the most likely number you will roll will be 7. 2 and 12 are the least likely numbers to roll (this isn't exactly a bell curve, but it is if you add more dice... not really the point... I'm digressing and trying to use an example...). To overly simplify the algorithm that GT is likely using, let's just say that after the Initial WT is over, every random WT sequence GT rolls a pair of dice. If the number rolled is 7, the weather stays the same. If the number is higher than 7, weather increases; lower than 7, decreases. The amount of increase or decrease is based on how far away from 7 the actual number rolled is, with 2 and 12 being the most extreme changes. (Again, this is an over simplification and not really how it works... it just works the same way in principle...)

Now, if you only roll the dice a handful of time, there's a pretty good chance you'll never roll a 2 or 12 and have some sort of extreme weather change. At TP:1, you only roll the dice a handful of times. However, over a given amount of RT, you will double the amount of rolls every time you double the WC factor. Given a large enough number of rolls, it becomes increasingly likely that you will experience a few 2 or 12's... This is not to say that with TP:1 you would never roll a 2 or 12, it's just not likely. If you do roll a 2 or 12, it's also not likely that you'll roll another, and even less likely that you'll roll the opposite. This is why a race could start with a low W% and TP:1, and end up having TT:17h+ of rain, because perhaps a 12 was rolled so the weather spiked up... but then every roll after that just hovered around 7, and so the W% never decreased.

Also, because the W% can't go any higher than 100% or lower than 0%, if it ever reaches either of those extremes, it becomes exponentially more likely to stay there. If you have 100% weather, every roll of 7 or higher means the weather stays at 100%... if you have 0% weather, every roll of 7 or lower means the weather stays 0%.

As for testing, if you want to have an idea of what the 24 hour race will be like with TP:1, just set the W% to what you want, and set the TP and WC to 60. This will reflect the same number of "dice rolls" as it would with TP:1 over 24 hours. If you set the same W% and do it again, you will probably get a different weather pattern, because the changes are random, but roll frequency will still be the same. However, because you can't raise WC any higher than 60, to represent TT:24h at TP:2 in the least amount of RT, you need to decrease the TP to 30. With TP:30 and WC:60, it will take RT:48m to accurately recreate what the weather would be like over TT:24h, TP:1, WC:2.

Now, as for my personal opinion, I think the best thing to run is a low percentage W% (like 10-20%) with TP:2. This makes it unlikely that a torrential downpour will happen... and quite likely that the weather would stay fine the entire time... but it still leaves the chance for rain. And, yes, there's even a chance that it will begin raining and never stop... but at least at TP:2, it takes RT:15m to go from SW:0-100% in the event that it does suddenly become a monsoon. Cars become tough to drive on slicks around 20% (~3m) and next to impossible to drive at 50% (~7.5m), which is plenty of time to get around the track and back even if you were unlucky enough to literally have just passed the pit lane when it opened up... as long as you know not to slam down the gas coming out of every corner...

I encourage people to test this out or prove me wrong if you'd like... This just seems to be the most logical explanation based on the evidence that I've acquired thus far... I plan on eventually running some more tests... but now that I understand how it works I'm less curious and motivated. :sly:
Some very interesting research and calculations. One downside, even on a track with time change on 1 or even on a track without the time change facility, I have still seen standing water go from 0-100-0% in a matter of minutes.
 
Some very interesting research and calculations. One downside, even on a track with time change on 1 or even on a track without the time change facility, I have still seen standing water go from 0-100-0% in a matter of minutes.

Right, if you have TP:1 but still WC:60 it'll still do 0-100% in 30 seconds because they're separate clocks. But, if you run them the exact same, then it will perform the exact same in Track Time as TP:1, WC:1.
 
So effectively what you are saying is that if weather change is on anything other than 1, it will increase the rate of SW gain / loss in RT?

Also, I might be wrong here, but I was under the impression that the WC number controlled the amount of times the weather changes occur during the race, or number of dice rolls using your calculations. Based on the fact that with WC on 1, from my experience there is only 1 change in weather during races. And 60, being 60 changes.

So what happens when WC is set to 1, and TP is set to say 14?
 
So effectively what you are saying is that if weather change is on anything other than 1, it will increase the rate of SW gain / loss in RT?

Also, I might be wrong here, but I was under the impression that the WC number controlled the amount of times the weather changes occur during the race, or number of dice rolls using your calculations. Based on the fact that with WC on 1, from my experience there is only 1 change in weather during races. And 60, being 60 changes.

So what happens when WC is set to 1, and TP is set to say 14?

Originally, I thought that it affected the amount of times as well, but that's only partially correct. It affects the entire "Weather Clock" which includes weather progression as well as surface water increase. So, higher the WC, the quicker the surface water increases by a predictable amount.

WC affects that weather change with the exact same equation as TP... so if they're the same, they'll work the same as compared to RT... Since we'll only experience lap times and such in RT, we'll always compare our experience of weather to RT. Compared to RT, at 100% Weather,
WC:1 will go 0%-100% in 30 minutes
WC:2 = 15m
WC:4 = 7m30s
WC:8 = 3m45s
WC:15 = 2m
WC:30 = 1m
WC:60 = 30s

That's RT, or as we will view it driving around the track... WC and TP are independent of each other, but because they work on the same equation, we can use them to test things like 24h of Track Time in 24m of Real Time.
 
WC:1 will go 0%-100% in 30 minutes
WC:2 = 15m
WC:4 = 7m30s
WC:8 = 3m45s
WC:15 = 2m
WC:30 = 1m
WC:60 = 30s

I haven't been keeping up with the chat on weather, but based on these numbers would you agree that we should be looking at a weather changeability of <6?

Great work @Severn !
 
I haven't been keeping up with the chat on weather, but based on these numbers would you agree that we should be looking at a weather changeability of <6?

Great work @Severn !

I would say that you wouldn't want to do a weather changeability above 2 or 3... because even though it takes 7m30s to reach 100% at WC:4, once you're past say 20% on racing slicks you're going to be slowing down significantly, and once you're past 50% you're going to be crawling... 50% would happen in less than 4 minutes.

Edit: It should also be noted that a few times during testing changes to the weather were not immediately reflected on course (the course did not fully reset)... so even though I would change the WC, the rate at which the SW would increase would stay the same... this was a glitch. Starting the race (or otherwise resetting the track by exiting, changing course, whatever), will reset the Weather to the race settings and progress as predicted.
 
Last edited:
The best example I've found for watching weather changes was to set W%:60, WC:60, TP:60. All times are presented in Track Time, as this would be the same as though I had set WC:1, TP:1 and watched for 24 hours Real Time except I could watch/drive for just 24 minutes.
It started with a light rain (as 60% is a light rain at Sarthe) that did not accumulate. After a little over an hour the SW began to increase at about 1% every 2 or so minutes until it reach the mid 20's%. I leveled there for a few more minutes before drying up rapidly (several % per minute) back to 0%. It would stay at 0% for the rest of the race, but there were a few more light showers overnight. Then, when the sun came up in the morning, the sky was clear and sunny. Around noon, it started to became hazy and overcast again but never reached a point of raining before completing at 1600.

I did a couple more at that W% and a few others, but that one stuck out to me as a nice example of the weather changing over 24 hours.
 
^From that would you suggest actual settings at the start W%:50, WC:1, TP:1?

That would depend on everyone else... at 50% you do run as close to a 50/50 shot of have good weather, bad weather, a little of both... or whatever... obviously it's pretty much entirely unpredictable as far as what the weather will actually do. If we'd like it to be less likely to go to or stay on the side of bad weather then we'd want to use a lower W%. The only thing that will be guaranteed is that for an unknown amount of time (I haven't seen a noticeable change in anything less than an hour yet) the weather will be at whatever the W% was originally set to... anything after that is up to chance and probability.
 
I can confirm everything @Severn said is accurate; seemingly at WC:1, 100% weather means surface water increases at approximately 3% per minute meaning that the track goes from completely dry to completely wet in 32 minutes.

In practical terms this means that you can do 4 complete laps at WC:1 before Racing Softs are simply not worth the risk (~35-45%). However, this shrinks to 1.5 complete laps at WC:2 due to the not quite linear reduction in grip level as the surface water goes up.

It is important to remember that this is at 100% weather, which in reality is a sudden torrential downpour and, as previously discussed by Severn, is unlikely to occur without lighter rain preceding it, which would increase the amount of time teams had to react. This combined with the Severn's logical hypothesis that increasing the WC increases the number of potential changes I would suggest that WC:2/3 would provide the best combination of realism and dynamic weather, without making it undrivable.
 
Earlier this week I did some testing as well. WC and TC at 1, 0% surface water, 100% rain. It was raining for the whole time (25 min) and the surface water indicator never once passed 0%.:confused:
 
what you people need to be testing is does the weather actually channge??? .......start with 100% weather , wc1 or 2 , tc1.....because i had a lobby open for almost 5 hours and it never quit raining, once road surface hit 100% it just stayed there. you need to get the rain to stop and have the road surface decrease.
 
what you people need to be testing is does the weather actually channge??? .......start with 100% weather , wc1 or 2 , tc1.....because i had a lobby open for almost 5 hours and it never quit raining, once road surface hit 100% it just stayed there. you need to get the rain to stop and have the road surface decrease.

Yes, the weather will change. However, it may *not* change or it may get worse... it's random after a certain amount of time. Whatever you set the W% initially will stay that way for at least an hour (I haven't noticed a noticeable change in anything less, anyways.) No matter what we set the weather to, there will always be a chance that after a little while it will begin to rain and never stop... however, you can decrease the likeliness of this happening by setting a lower and lower initial W%. There is no guarantee...

In your case (if I remember correctly you had it set to 70% initially?) it sounds like the weather got worse, and then just hovered around bad... never decreasing enough to stop raining. As I stated, though, you can test what the weather at WC:1, TP:1 at any W% by setting WC:60, TP:60 and see the changes that would occur over 24 hours in just 24 minutes. If the weather changes and it starts to rain at 02:00 on the track timer, it would have started raining at 2am during the actual 24 hour race. However, again, the changes are random... so you can run that exact same test 100 times and never get the exact same weather over the course of the race. It's all about unpredictability. The best we can do is try to limit the severity of the Surface Water increase so no one ends up stuck at the end of the track if a sudden downpour does occur.

(I just noticed that you said you started with 100% initial W%, in that case it becomes a lot more likely that the weather will not change. Because it cannot go any higher than 100%, half of the options for GT to randomly pick (basically, anything that says the weather gets worse) suddenly becomes just "stay at 100%"... It is entirely possible that GT thought of this and the only thing it can do when it hits 100% is begin to decrease, but from what I could tell, there was little to no difference in SW accumulation between 90% and 100%... so even if the weather did decrease for you, there's still a good chance that it ended up hovering around between 80-100%, which will get up to SW:100% but at different rates... and maintain SW:100% if it already exists...)
 
Last edited:
Thanks for all your experimenting and explaining severn. I finally understand how weather works in GT. 👍
 
Thanks for all your experimenting and explaining severn. I finally understand how weather works in GT. 👍

Well, I'll do you one better. In an effort to visualize the "randomness" factor, I decided to throw a spreadsheet together that may help... maybe. I make no claims of knowing exactly what the timing or algorithms that GT actually uses, but I have developed something that should at least show that it's random, and regardless of what you select there's always a chance things could go a little nuts.

(Again, none of these numbers, calculations, or anything are likely correct. They just give an easy way of showing an example.)
Suppose that whatever you set for your initial weather lasts for at least 1 hour, and then after that the weather changes at a rate of 4 times per hour. Suppose also that the probability curve for the weather changes are most likely to stay the same or only change a few percent, and the least likely change is that of a full 30% change up or down in a single change. A 30% change is extremely unlikely, but still possible.

Using a gradual color change between Orange (most dry - 0-10%) to Dark Blue (most wet - 90-100%), each of these charts shows 10 unique possible races. With the initial weather set to 0%, most of the time you'll probably get a pretty dry race... but this doesn't completely rule out the chance of having one that goes Wet and stays there.
Weather0.png


At the same time, if you set the initial weather to 90%... most of the time you'll get a pretty wet race... but it doesn't mean that an unlikely dry race after a few hours is totally impossible.
Weather90.png


You can see my full spreadsheet here.
Every time you load or refresh the spreadsheet entirely new random numbers are generated for each possible weather change (so it can take a little bit to load), so you will always see a new, completely different, weather change pattern for each of the different charts.
 
I will leave one more bit of research before I officially abandon this project to get back to car research :lol:

I used my formula to run 100 races at each of the starting W%. Then averaged the W% overall, Time in the Dry, and Time in Wet. Regardless of how my formula differs from PD's, the difference over 100 races will be negligible because of that randomness factor. Also, just because these are the averages, there are certainly some extreme swings in what happens in an individual race... you can never be 100% certain how an individual race will go.

W%:0; Avg W%:32.76; Avg Dry Hrs:18.93; Avg Wet Hrs:5.07
W%:10; Avg W%:34.91; Avg Dry Hrs:18.70; Avg Wet Hrs:5.31
W%:20; Avg W%:39.49; Avg Dry Hrs:17.46; Avg Wet Hrs:6.55
W%:30; Avg W%:42.31; Avg Dry Hrs:16.19; Avg Wet Hrs:7.82
W%:40; Avg W%:45.52; Avg Dry Hrs:15.61; Avg Wet Hrs:8.40
W%:50; Avg W%:45.08; Avg Dry Hrs:15.97; Avg Wet Hrs:8.03
W%:60; Avg W%:48.80; Avg Dry Hrs:13.34; Avg Wet Hrs:10.66
W%:70; Avg W%:57.77; Avg Dry Hrs:10.22; Avg Wet Hrs:13.78
W%:80; Avg W%:59.98; Avg Dry Hrs:9.62; Avg Wet Hrs:14.38
W%:90; Avg W%:59.50; Avg Dry Hrs:9.56; Avg Wet Hrs:14.45
W%:100; Avg W%:61.41; Avg Dry Hrs:9.42; Avg Wet Hrs:14.59

Again, these numbers can be found in my spreadsheet. It will still generate random numbers on each load, so your results will vary (also, individual W% sheets are just numbers, no formatting because it slows it waaaay down.)
 
Last edited:
Back