Here is my take on this issue, I will appreciate all insights.
It is a lengthy read, but I guess it was need to explain my subjective reasoning.
After commencing a small investigation with one of my friends, reading various subjective explanations of this issue and thinking of ways game actually works, I today believe that particular problem (tyre/fuel issue) is interconnected with sub-engine-mechanic of PP calculations and (maybe) RAM Overhead.
First thing I "discovered" is how my own personal experience is not drastic as many reported ones. I never had drastic changes in tyre-wear nor drastic changes in fuel depletion in races I've used for my testing (Willow Springs, Suzuka Wet and Silverstone endurances in S class) as other reports. However, I discovered why: I always run no-ABS with custom brake-bias and I always take into consideration the sub-mechanic that depletes the fuel faster in higher-revs and during splistream (as it is since GT5).
I observed some replays of my friends and discovered they always run the race in almost 100% of grip/cornering abilities/braking/revs of the car. Also, they were all doing the same thing after the first pit-entry (in the second stint): they would speed up their pace drastically, go into overtakes aggressively, etc, in order to sub-consciously (I presume) compensate for time lost in the pits.
I did most of my testings on Suzuka Wet race, using the same car all the time - NSX GT500 LM Base Model. Since wet-sequence was always the same (rain starts falling in lap 2, ends in lap 8) I could run very consistent laps throughout my testings (which led to similar conclusions on Silverstone race). After the first 5 laps - before entering the pits - I had tyre-wear around 7(F)-6(R) and something below half tank depleted. After the race end (lap 10) I had 6(F)-5(R) tyres for the second stint with fuel around the same 10% difference more-less. So, I had to agree that difference is there and it exists, but it was not drastic as many were reporting and I would probably never realize it even exist if it was not reported.
So, I start to observe why is my experience a bit different and what could be the reasons for obvious difference in wear/depletion (and laptimes, which I will tackle soon).
First I discovered the aforementioned difference in tyre-wear when using ABS-off. I had pretty much balanced wear throughout the race, unlike other reports. I realized how the ABS is the main culprit for that, because of the default brake-balance that game have for ABS and which noone probably even bothers to adjust. As we are all aware, game uses default BB of 5-5 for all cars and it actually becomes a significant issue when running an race with tyre wear with ABS ON because of two problems:
1) Default 5-5 balance is not the accurate balance of mass-shifting that corresponds with suspensions and individual characteristics of any vehicle - as I preach since GT5, the BB for every car should be adjusted accordingly and individually, either you use ABS or not (but it becomes a prerogative if ABS is not used), and
2) default 5-5 balance with ABS-on does not "override" the proper balance values and becomes a menace once the tire-wear comes in the picture because it actually wears the tires - especially the rear ones - much more than it should
With my NSX I used 6-1 balance (the one I usually use for Race compound for MR/FR vehicles below 1400kg of curb weight) and I had the above results in wear (7-6 after the first pit, 6-5 when the race ended).
*** for those unfamiliar with my personal "non-ABS crusade" for GT games it is worth noticing how "ABS" in Gran Turismo games IS NOT SIMULATING real-life ABS effect nor it has any connection with it. ABS in Gran Turismo is just a braking assist that not only prevents the brakes from locking (in a totally artificial way compared to real-life ABS), but it also influences on traction and stability properties, despite player have other assists deactivated. But nevermind that now.
Second, I noticed how almost noone (of other replays I watched and reports I read on GTP) takes into account the very ancient property of tyres in the GT series (since GT3), where fresh set of tires is actually cold immediately when you get out of the pits. The actual first lap on fresh tyres becomes a crucial for their later "development" during the race, because damaging the tyres in cold-state have serious repercussions on their wear in later stages. Because they are cold, fresh tires don't have a strong lateral grip as heated tires and you have to be careful not to take much lateral forcing into red-zone because they will wear unevenly and much faster later. The key "issue" here is how all races in GT start from the rolling-start and with tires already "heated" for the race - so many players are used to drive on 100% of potential from the start, despite of the stint. However, it is very damaging to run cold tyres in the same way as pre-heated and I see a second reason for greater wear of tires in the second stint because of that. But that issue couldn't affect the wear so significantly (either in my or others experiences), so there have to be some other thing on top of everything.
And third, as I said before, I discovered how the difference in my fuel depletion is also not drastic as other reports, but how it definitely exist. As aforementioned, by observing other people replays I discovered I actually run races much more calmly and conservatively, without going into the red areas of revs and with early shifting. I have to notion another race-related "issue" that appeared, and that is how after the pitting I run into much more traffic than in the first stint because of the position of the AI cars and their pit-sequences and I was getting into the slipstream more often, which also results with much higher revs and faster depletion.
So, I highlighted three reasons: brake balance/ABS, tyre heat state usage and actual usage of the power (engine) related to fuel consumption. All of those could partially answer the reasoning for the change of depletion/wear among stints, but even together they couldn't make the difference so drastic.
But, there was another serious thing besides actual depletion/wear: the lap-times in second stint were always slower then in the first stint. No matter how hard you try, you simply can't drive times in the second stint close to those in the first. All factors above also couldn't influence that by themselves by no logic. But there had to be a logic for all that. So, what is the fourth factor - and IMO most important issue?
It is actual PP sub-calculation engine or, as I call it, "inability of game to determine the starting PP values for the cars in pre-pit and after-pit state". So, what the hell does that even mean?
By observing the times on Suzuka Wet race I noticed the strange pattern: When my first lap time (on pre-heated tyres) and full tank was 2:04/2:05, my 4th lap (prior to pit entry) was 2:01. But after I went out of pits, my first fast lap (6th lap) was 2:09 and my last lap would actually be around 2:05, 2:04 in the best occurrence. And then all puzzles begun to fit.
My personal belief is game simply can't override two different states when entering/exiting the pits:
>>A) first pit entry "state", where car weight is its default PP weight (the one displayed in Dealership) + fuel weight and minus fuel used during the first stint, and
>>B) pit exit "state", AKA the moment when the cars start to accept additional fuel in the pits and continuing the race with new tires and "newly added" fuel
I believe that car in the "first state" have its Dealership weight with the added weight of the fuel already INCLUDED in the overall weight - and that game ALWAYS uses that weight, with or without fuel depletion enabled, in the races as cars starting weight (in all game modes). What I am basically saying is that cars in the game have additional weight of the fuel included all the time during the race as their TRUE START WEIGHT VALUE. So, during the first stint, the depletion sub-mechanic is simply lowering the weight of the car accordingly to consumption form that "TSWV". For the sake of the numbers, let's say that the car has a Dealership weight of 1200kg+200kg of fuel = TSWV of 1400 kg. After the first stint, I am entering the pits with 3/5 of the fuel consumed, so my car actually has approx. 1280 kg (>>A). Because of the lowered weight I managed to run faster laps as fuel depleted (and it was always lap #3 or sometimes #4 because of the combination of lower weight and fastest state of tyres and maybe some slipstreaming).
And then comes my "pit exit state" (>>B). I believe that game simply can't - does not have RAM on disposal for that for all cars in the race - to calculate the new weight accordingly to state of depletion (1280 kg in my example). I presume game simply "reads" the first lap starting value in full (aforementioned "True Start Weight Value" of 1400 kg) and lumps extra fuel on that number as addition = resulting with car that have more weight than on the very start of the race. In my example-calculation, when exiting the pits after the first stint, car have all added fuel added on TSWV weight, so if I decided to go with 50 litres (100 kilograms of weight for example), instead of having 1380 kg when leaving pits, car actually have 1500 kg - 100 kg more than it had on the very start of the race.
When I put that idea into aforementioned 3 reasons for occurrence of increased tire-wear/fuel depletion that are - in my opinion - simply result of way people actually drive the race, it became only logical explanation of the overall issue.
It can explain why is the actual fastest lap in the 2nd stint - the last lap of the race - almost the same as the first lap of the race = only in that moment car have the same weight (using the additional fuel added on TSWV value in the pits brings the weight down to initial TSWV value, but never close to weight car have when entering the pits for the first time)
It can explain why are my tyres more worn in the 2nd stint = during the second stint car is almost 10% heavier than in the 1st stint - due to fuel being added on the "TSWV" value - which results with more mass being pushed on suspensions and increases the wear (more weight-more wear)
So, to conclude, I think there are two groups of factors: individual (where the combination of ABS and the actual way the car is driven by the player influences difference in wear between stints) and the problem with calculation explained above, which is purely issue of the game.
All above is my subjective and personal belief.
I guess it could be corrected if PD would find a way to allocate RAM to keep the real-time weight-data (Dealership weight + remaining fuel) prior the fuel has been added in the pits, but I have no idea how actually RAM-taxing could that be.
Why do I personally think it is connected with RAM?
There has to be a reason why the issue is happening, if my reasoning is correct - and "error" is not an explanation, because if it was simply an "error" in real-time formula, those 3 lines of code would get corrected in the first update. But they are not. And I tried to understand why.
All I can think of is RAM allocation, because engine itself can do a real-time calculations for everything else in the background: tyre wear for all cars, fuel depletion, AI routines, pit routines, weather routines, skidmarks..
But this actual thing - determination of actual weight on the pit-entry, followed by weight change during refueling and weight after refueling - is the only real-time variable in the whole "race procedure" that gets changed in the mid-time with actual 2 variables that needs to be inter-dependant and that have no FINITE value (the word "finite" is the key to understand my point):
A) Weight of the car at the moment of the pit-entry (Dealership weight + weight of remaining fuel) and,
B) The weight of the car after the fuel is added (Dealership weight + weight of the remaining fuel + weight of the added fuel)
Both variables are "infinite" - they can be whatever value possible, simply because you can use as much fuel as you like prior to entry to the pits and you can refill the tank with any value you like.
On the contrary, when changing the tires, there is actually only 1 "infinite" value:
A) The level of wear when the car gets to the pits
Once the tires are changed, they are back to "finite" value, as "Brand new" - and game "knows" where to "start" - B is FINITE (same as repairing the damage for instance, where the repaired damage is always a finite value).
With fuel, both A and B are "infinite" value and game has to "determine" where to "start" once the B variable (refueling on top of the existing weight and remaining fuel) "begins". My presumption is how the game actually simply uses the only FINITE value it can have, and that is the weight value for the very start of the race - which I call "TRUE START WEIGHT VALUE" in my first post.
And that is why I think it is RAM issue, because there is not other explanation. In order to get proper weight-change for all 16 cars in the field once they pit, the game would have to calculate 2 "infinite" values all the time for all 16 cars and track them accordingly and act accordingly once the new fuel is added. And somehow I think it is not really that simple to execute with all other real-time things going on in the race.
I somehow think that one "wild" variable - such as the fuel one I've described - becomes RAM problematic because it is the only of all above that have two "infinite" states during its "revolving".
However, I might be absolutely wrong.
maybe devs comes from micro$oft where a bug is called a feature 
)
