New 'forgiving' fatigue model

[sub_optimal]

Originally Posted by sub_optimal

Second, what leads you to the conclusion that SpecificFactor is envoked when crews are moved from compartment to compartment?

... which is, of course, not what you said at all. I fail at reading.

[LGN1]

Just a small example to illustrate the formula (values taken from GWX):

[CREW_0] ;SEAMAN
MoraleMin=0.30
MoraleMax=0.60
FatigueMin=0
FatigueMax=0.80
CoefMorale=0.4
CoefFatigue=0.21
QualEffect=1

--> A rested seaman adds 0.76 to the efficiency in a compartment. When completely exhausted he adds 0.76*0.368 = 0.27968.

[CREW_3] ;PETTYOFFICER
MoraleMin=0.40
MoraleMax=0.70
FatigueMin=0
FatigueMax=0.80
CoefMorale=0.3
CoefFatigue=0.3
QualEffect=2

--> A rested petty officer adds 0.79 to the efficiency in a compartment. When completely exhausted he adds 0.79*0.44 = 0.3476. If he has the correct qualification for the compartment he adds 1.58 and 0.6952, respectively.

[CREW_6] ;SUBLIEUTENANT
MoraleMin=0.50
MoraleMax=0.80
FatigueMin=0
FatigueMax=0.80
CoefMorale=0.2
CoefFatigue=0.4
QualEffect=1.5

--> A rested sub lieutenant adds 0.84 to the efficiency in a compartment. When completely exhausted he adds 0.84*0.52 = 0.4368.

If the officer is in a compartment with a sailor (both rested), the total efficiency is (0.84+0.76)*1.5 = 2.4 (assuming the officer has the correct qualification). Clearly, having an officer with the correct qualification is very important.

Cheers, LGN1

[LGN1]

Originally Posted by sub_optimal

Wait... maybe I'm going crazy, but I thought that you had to do the addition and subtraction first, and then the multiplication and division.

Originally Posted by sub_optimal

... which is, of course, not what you said at all. I fail at reading.

Hi sub_optimal,

multiplication and division come first:

a*b+c = (a*b)+c
a*(b+c)=a*b+a*c=(a*b)+(a*c)

Concerning the specific and regular factors. I am pretty sure that (as you wrote) specific is for surfaced and submerged (my first post was wrong. I corrected it.). In the thread I linked it says that regular0 is a one-time hit when someone is moved to a compartment. I think I tested this once and confirmed it for myself. Maybe increase the value in basic.cfg and see whether you observe a change.

Cheers, LGN1

[sub_optimal]

Originally Posted by LGN1

multiplication and division come first

I looked it up; you're right. It's scary what I've forgotten from elementary school! Either that or I've hopped from one universe to another. :-) Yep, I'm sticking with that story.

Originally Posted by LGN1

I think I tested this once and confirmed it for myself. Maybe increase the value in basic.cfg and see whether you observe a change.

Yep, I tested this too, and it's correct. Still don't know what Regular 1 does, though. Might it be a hit when the sub takes damage or something?

[sub_optimal]

Earlier threads claim that the time compression above which fatigue will not increase (or decrease, depending on your perspective) is set by the "3d Render=" option, which defaults to "3d Render=32".

Does anybody know if this is true?

Also, how exactly do the fatigue coeficients work, in practice, and how do they interact with FatigueStep? Does anybody know the formula for this?

[sub_optimal]

Some more discoveries:

The fatigue system is divided into distinct 'rounds'
Each round is ten minutes long, and begins on the "0's" (1:10, 1:20, 1:30, etc)

[NB: I just double checked this, and it seems that fatigue is also sometimes recovered on the "5's"; clearly I haven't worked this out entirely yet.]

The fatigue penalty or recovery is assigned at the beginning of each ten minute round. If you put a guy into a compartment at 12:09 he will be just as fatigued at 12:10 as he would have been if he had been assigned at 12:01.

The above does not apply to the RegularFactor penalty, which is assigned as soon as a sailor is put into the compartment.

I think that the fatigue penalty defined by SpecificFactor is initiated at the beginning of each round by subtracting the SpecificFactor penalty from the man's fatigue score, where 0 is completely fatigued and 1 is completely fresh.

FatigueMax appears to define the point at which the ! shows up and the man is exhausted. However, this operates the opposite to way which one might expect given the way the other numbers work. It seems as if a sailor with a FatigueMax of .2 becomes exhausted fast (after 20% of his fatigue points are gone) whereas a sailor with a .8 score can tolerate 80% of his fatigue points disappearing before becoming exhausted.

FatigueStep seems to be the number added to the fatigue score for every round the sailor begins in a Crew Quarters compartment.

The SpecificFactor score for the Quarters compartments (which is negative) does not seem to be added to their fatigue score. This is very mysterious, because this is different from what others describe. However, I assigned a '-1' score to this compartment, expecting that it would bump the player up to full health immediately, and yet their fatigue seemed to dissipate at the usual rate.

I would be very grateful if anybody could shed light on this further.

[Hitman]

Earlier threads claim that the time compression above which fatigue will not increase (or decrease, depending on your perspective) is set by the "3d Render=" option, which defaults to "3d Render=32".

Does anybody know if this is true?

Yes it is. Fatigue will operate only under the TC set for 3D render.

Some time ago Kaa created a fatigue model I liked much, but that was dropped for unknown reasons. By setting 3D render to 1024 (Or whatever maximum TC you had), and giving all crew small fatigue values, they became exhausted over a period of months, hence making a good representation of overall fatigue in patrol. You didn't need to move crewmen around any more, they just tired always the same -and a bit faster during storms if surfaced-. I think this was a very good representation of one aspect involved, i.e. overall fatigue. Other models liek the one you are exploring represent better the normal watch shifts during patrol, but when a crewman is rested, he is as rested as if he had not been already one month in patrol. I prefer the other method, and as with all these simulations goes, it's always a trade-off; rarely will you be able to say "this or that is more realistic" ... it's a matter of choosing what aspect of reality you prefer to simulate in such a limited enviroment.

[sub_optimal]

Hmmm... in that case, do you know if it's possible to automatically load a 3d Render setting with a fatigue model in SH3 Commander?

[Hitman]

I don't know if it is possible to link those two parameters together (One fatigue model and TC), you will need to contact JSCones about that.

[sub_optimal]

I've posted a message to JScones. Let's see if he or she is interested in the idea.

[LGN1]

Originally Posted by Hitman

Some time ago Kaa created a fatigue model I liked much, but that was dropped for unknown reasons. By setting 3D render to 1024 (Or whatever maximum TC you had), and giving all crew small fatigue values, they became exhausted over a period of months, hence making a good representation of overall fatigue in patrol. You didn't need to move crewmen around any more, they just tired always the same -and a bit faster during storms if surfaced-. I think this was a very good representation of one aspect involved, i.e. overall fatigue. Other models liek the one you are exploring represent better the normal watch shifts during patrol, but when a crewman is rested, he is as rested as if he had not been already one month in patrol. I prefer the other method, and as with all these simulations goes, it's always a trade-off; rarely will you be able to say "this or that is more realistic" ... it's a matter of choosing what aspect of reality you prefer to simulate in such a limited enviroment.

I created a similar model for myself. Your crew starts fresh and is exhausted after 6 weeks (for the VIIB/C) (the efficiency of the exhausted crew stays above interval1). You don't have to move anyone around (except the lazy WO) as in the no-fatigue model, but in contrast you have some fatigue effect.

The NYGM team had the nice idea to use morale to get a long-term effect. IIRC, they did it by giving the morale step a negative sign. Unfortunately, this has two minor problems: 1) If the journey is too long or you choose a too large morale step, the morale can become negative. 2) Morale is only reduced in the rest quarters. So, you need to regularly move your crew around so that anyone gets a morale reduction (and you should set 3d Render to your max. TC rate). As a consequence, the long-term effect only works well with micromanagement. That's at least how I remember it. Stiebler, please correct me if I am wrong.

Cheers, LGN1

[LGN1]

Originally Posted by sub_optimal

FatigueMax appears to define the point at which the ! shows up and the man is exhausted. However, this operates the opposite to way which one might expect given the way the other numbers work. It seems as if a sailor with a FatigueMax of .2 becomes exhausted fast (after 20% of his fatigue points are gone) whereas a sailor with a .8 score can tolerate 80% of his fatigue points disappearing before becoming exhausted.

The SpecificFactor score for the Quarters compartments (which is negative) does not seem to be added to their fatigue score. This is very mysterious, because this is different from what others describe. However, I assigned a '-1' score to this compartment, expecting that it would bump the player up to full health immediately, and yet their fatigue seemed to dissipate at the usual rate.

IIRC, the ! does not show up when FatigueMax is reached, but 10% before it's reached.

You are right, the rest quarters have two variables: fatigueStep/moraleStep and the specific factors. I don't remember what the influence of the specific factor was, but I would not be surprised if it has no influence in this case.

The rest sounds quite similar to my observations.

Cheers, LGN1

[Stiebler]

LGN1 said:

The NYGM team had the nice idea to use morale to get a long-term effect. IIRC, they did it by giving the morale step a negative sign. Unfortunately, this has two minor problems: 1) If the journey is too long or you choose a too large morale step, the morale can become negative. 2) Morale is only reduced in the rest quarters. So, you need to regularly move your crew around so that anyone gets a morale reduction (and you should set 3d Render to your max. TC rate). As a consequence, the long-term effect only works well with micromanagement.

Yes, that is substantially correct, and the reason why subsequently I reduced the morale effect for key personnel (especially officers) in NYGM. As a result, it has been a very long time since I last had any trouble with crew management on NYGM U-boats. (I don't think morale is improved in the rest quarters, in Observer's NYGM implementation; it was so long ago that I've forgotten now).

The big problem was persistent poor morale if you don't sink ships. Sail a couple of months without sinking anything, and morale fell too heavily. But if you command a IXd2 U-boat in 1944, you're not likely even to encounter any targets as you patrol from Bordeaux to Cape Town or the Indian Ocean (attacking convoys with such a large boat is suicide in 1944). Similarly, very long schnorchel patrols in 1945 from Norway to the English Channel area, without ever seeing any targets unless one encounters a chance convoy en route, ruined the crew's morale.

I never thought of changing the 3D-render effect, always left it at tc=32.

Stiebler.

[Volk2]

Is someone using RUb? It looks like something similar to "overall fatigue in patrol" (or "combat stress", or whatever):

Fatigue: the new fatigue is based on a combat stress model. It is no longer fatigue. The feature is meant to reward careful crew assignments while removing the burden of micromanagement that the original game placed on the player. In this mod the torpedo compartments and the repair area are the most stress-inducing areas of the boat, so try to keep crewmen out of these areas until they are required to work there. Also, try to limit placing watch crews on duty during bad weather. It's best to submerge when the weather is bad. Always make sure not to spend too much time at under 32x time compression.

Combat stress and crew management: When crewmen become fully stressed they will work at lower capacity. In the first couple of missions (until you train specialists) this can make it difficult to run the boat. In the short-term, place NCOs and officers in compartments which are having trouble. Between patrols, give the necessary specialist badges to NCOs and officers so that they can work the worst-affected compartments more efficiently.

Sleeping areas and fatigue: Fatigue (now combat stress) will no longer heal during time spent in the sleeping areas, but these areas will prevent crewmen from becoming more stressed. Fill these compartments with all non-essential personnel in order to keep them rested for duty.

Fatigue Marker: the fatigue marker's shape and colour has been changed. This is to remind players that high fatigue is not a 'red alert' situation. Crewmen can and will do their job even when highly stressed, but their efficiency will drop considerably. A stressed torpedo crew often takes 24 minutes to load a tube, while a fresh crew takes only 12 minutes.

[sub_optimal]

Ladies and gents,


I apologise for neglecting this thread for so long. Unfortunately, my important modding activities were rudely interrupted by this thing everybody keeps calling 'actual work'. Some people just don't seem to have a sense of priority.


So I've mocked up a spreadsheet, based on the assumptions and deductions mentioned in this thread, which should allow us to create fatigue models that behave in whatever way we want them to, and I post this first effort by way of example. This text is SH3 Commander-ready, so paste it in to the fatigue models file, under a new number, and it should Just Work.


Here is a description of my first attempt:


Name: combat-oriented, surface attack, 16h hold-down, forgiving – Draft 1


The model is...


Combat-oriented: The idea is to simulate fatigue in combat conditions, so 3d Render should be set to the same or greater than the max time compression during combat. Because of the 'forgiving' nature of the model (see below) this combat-oriented model is intended to be a bit like having 'no fatigue' during normal operations, while requiring fatigue management during combat.


Surface attack: The model is designed to be used during submarines that attack on the surface at night. In normal circumstances you should get about 8 hours out of your men on the surface, 4 hours in a storm.


16 hour hold-down: When a diesel-electric submarine is attacked by surface vessels, they will try to 'hold it down' in order to exhaust the crew. It is typical, then, for escaping submarines to spend long periods submerged. To simulate this, the model gives much longer fatigue times for a submerged submarine, averaging around 16 hours.


Forgiving: If, like me, you have a tendency to accidentally exhaust the crew during the course of normal operations, you can rest them relatively quickly, so that you don't have to spend ages and ages at 32 acceleration while your crew sleeps. All crew are set to completely recover in two hours game time. This is intended to make it difficult to do this in the middle of an attack, but easier if you're patrolling, as resting a whole watch will take less than four real-life minutes. Naturally, this means your crew might not be exhausted by the end of the patrol. In this respect, it's a bit like having 'no fatigue' turned on.


The model is also age-aware, in that as crew get more experienced, their stamina will go down. A young pup at the OS level will last about half an hour longer than the 'average' sailor, a grizzled old chief will become exhausted about a quarter of an hour sooner than average. (in this case, 'average' is set to the LS levels for ratings, Lt for officers).


I would greatly appreciate comments on this. Also, if anybody has any ideas for a model they'd like to see, I have a spreadsheet that makes them really easily (which I can share if people want to tell me how), so I can very quickly whip one up on request.


[instertNumberOfModelHere]


Desc=combat-oriented, surface attack, 16h hold-down, forgiving – draft 1 by Sub_Optimal
CREW_0|FatigueMax=0.8
CREW_0|FatigueStep=0.0166666666666667
CREW_1|FatigueMax=0.785
CREW_1|FatigueStep=0.0179166666666667
CREW_2|FatigueMax=0.77
CREW_2|FatigueStep=0.0191666666666667
CREW_3|FatigueMax=0.76
CREW_3|FatigueStep=0.02
CREW_4|FatigueMax=0.75
CREW_4|FatigueStep=0.0208333333333333
CREW_5|FatigueMax=0.742
CREW_5|FatigueStep=0.0215
CREW_6|FatigueMax=0.8
CREW_6|FatigueStep=0.0166666666666667
CREW_7|FatigueMax=0.76
CREW_7|FatigueStep=0.02
CREW_8|FatigueMax=0.74
CREW_8|FatigueStep=0.0216666666666667
FATIGUE_COEF|RegularFactor00=0.001
FATIGUE_COEF|SpecificFactor00=0.0139583333333333
FATIGUE_COEF|BadWeather0=0.0139583333333333
FATIGUE_COEF|RegularFactor10=0.001
FATIGUE_COEF|RegularFactor11=0.001
FATIGUE_COEF|SpecificFactor10=0.00930555555555556
FATIGUE_COEF|SpecificFactor11=0.00697916666666667
FATIGUE_COEF|BadWeather1=0.00279166666666667
FATIGUE_COEF|RegularFactor20=0.001
FATIGUE_COEF|RegularFactor21=0.001
FATIGUE_COEF|SpecificFactor20=0.00930555555555556
FATIGUE_COEF|SpecificFactor21=0.00697916666666667
FATIGUE_COEF|BadWeather2=0.00279166666666667
FATIGUE_COEF|RegularFactor30=0.001
FATIGUE_COEF|RegularFactor31=0.001
FATIGUE_COEF|SpecificFactor30=0.0160416666666667
FATIGUE_COEF|SpecificFactor31=0.0003
FATIGUE_COEF|BadWeather3=0.00534722222222222
FATIGUE_COEF|RegularFactor40=0.001
FATIGUE_COEF|RegularFactor41=0.001
FATIGUE_COEF|SpecificFactor40=0.0003
FATIGUE_COEF|SpecificFactor41=0.00930555555555556
FATIGUE_COEF|BadWeather4=0
FATIGUE_COEF|RegularFactor50=0.001
FATIGUE_COEF|RegularFactor51=0.001
FATIGUE_COEF|SpecificFactor50=0.0160416666666667
FATIGUE_COEF|SpecificFactor51=0.00930555555555556
FATIGUE_COEF|BadWeather5=0.0160416666666667
FATIGUE_COEF|SpecificFactor60=-0.1
FATIGUE_COEF|SpecificFactor61=-0.1
FATIGUE_COEF|SpecificFactor70=-0.1
FATIGUE_COEF|SpecificFactor71=-0.1
FATIGUE_COEF|RegularFactor80=0.001
FATIGUE_COEF|RegularFactor81=0.001
FATIGUE_COEF|SpecificFactor80=0.0160416666666667
FATIGUE_COEF|SpecificFactor81=0.00930555555555556
FATIGUE_COEF|BadWeather8=0.0160416666666667
FATIGUE_COEF|RegularFactor90=0.001
FATIGUE_COEF|SpecificFactor90=0.0139583333333333
FATIGUE_COEF|BadWeather9=0.0139583333333333
FATIGUE_COEF|RegularFactor100=0.001
FATIGUE_COEF|SpecificFactor100=0.0139583333333333
FATIGUE_COEF|BadWeather10=0.0139583333333333
FATIGUE_COEF|RegularFactor110=0.001
FATIGUE_COEF|RegularFactor111=0.001
FATIGUE_COEF|SpecificFactor110=0.00930555555555556
FATIGUE_COEF|SpecificFactor111=0.00697916666666667
FATIGUE_COEF|BadWeather11=0.00279166666666667