19 replies to this topic

[Amaris the Usurper]

Edit (November 7th): See my discussion about the post-patch behavior of double heat sinks here: http://mwomercs.com/...-effectiveness/.

Edit (November 2nd): My main purpose in starting this thread was to expose some (probable) bugs through experimental testing. The guess that the ItemStats.xml "heat" values were being used as rates of heat generation (instead of total heat generated) for small, medium, and large pulse lasers, as well as small lasers, was correct. Also, after the November 6th patch, double heat sinks (including those in the engine) will dissipate 0.14 heat/second, not 0.2 (as would be consistent with the TT game). See http://mwomercs.com/...heat-sinks-dhs/. Finally, weapon heat generation, beam duration, and cooldown time will become more transparent in the future; see http://mwomercs.com/...e-improvements/. Keep this in mind when reading. I will produce an updated guide to heat behavior once things have settled down.

TL:DR/Summary

Single and double heat sinks should probably dissipate 0.1 and 0.2 heat/second, respectively (i.e., they are probably intended to be coded this way). [Edit: This has been confirmed by the devs.] Medium and large pulse and "beam" lasers produce roughly the same amount of heat per shot. The values in ItemStats.xml, used in Ohmwrecker's tables, may not be used by the game at all, or they may be used in a complicated way. As noted by the devs, double heat sinks in the engine aren't fully effective. This is also the case when using single heat sinks with an engine that has zero or more heat sink slots, and no additional heat sinks are fitted (new finding). So, for example, if you have at some point switched from 11 to 10 heat sinks on your Catapult's 260 standard engine and been very disappointed with its ability to dissipate heat, you are probably not alone.

Long Version

[Disclaimer: Rudimentary knowledge of calculus required to fully "get" it. Brace yourselves.]

In what follows, I attempt to shed some light on these questions:

• How (in general) does the heat mechanic work?
  
• How much heat do the different weapons produce?
  
• How effective are heat sinks?
  
• Is there anything "bugged" about engine heat sinks beyond the already-noted double heat sink issue?

Theory of the Heat Mechanic

I hypothesize that the heat level changes at a rate

dH/dt = RHG-RHD

down to a minimum of zero, where:

H = absolute (not %) heat level of mech (number of heat points currently carried);
RHG = rate of heat generation by weapons;
RHD = rate of heat dissipation by heat sinks;
t = time (in seconds).

While this has never (to my knowledge) been stated by the devs, I think it summarizes how most of us (especially those familiar with the TT game) intuit that the heat system should work. Note that I have not yet said anything about how RHD depends on the number of heat sinks (or other factors).

Integrating the expression for dH/dt between t = 0 and t = T, we get

0 = THG-THD => THG = THD,

where THG and THD are the total number of heat points generated and dissipated, respectively. This expresses the obvious fact that, if H begins and ends at zero, the number of heat points generated must equal the number of heat points sinked.

Now, based on my observations in MWO (and consistent with the TT game), it appears that we should have

RHD = 0.1*NHS

for single heat sinks, and

RHD = 0.2*NHS

for double heat sinks, where NHS is the number of heat sinks carried. Thus,

THD = 0.1*NHS*T,

or

THD = 0.2*NHS*T,

depending on the type of heat sinks, and provided the heat scale stays above zero.

Edit: According to http://mwomercs.com/forums/topic/66075-heat-and-double-heat-sinks-dhs, the values of 0.1 and 0.2 for SHS and DHS are correct. Thanks, Culler.

Obviously, all of the above assumes continuous behavior of what must, in reality, be a time-discrete process. Let's just assume that the time increment is small enough that the continuous model is accurate.

Now for some justification. First, this is how things work in the TT game: one heat sink dissipates one heat point over ten seconds. Second, one can do some data mining and find an XML file (ItemStats.xml) containing weapon stats (including heat generated, beam duration, recycle time, etc.), which have been incorporated into popular tables by Ohmwrecker. The results I have found are not in agreement with a naive interpretation of the values contained therein (see below). The major point is that we do not know how or even if the game uses these stats. Because of this, we shouldn't reject ideas about how the heat system works solely because they are inconsistent with some other idea of what the numbers in ItemStats.xml mean. The important thing is the experimental data.

Mech Heat Capacity and the MechLab "Heat Efficiency" Statistic

As previously noted by others, the heat "capacity" (HC) of a mech (i.e., the amount of heat points that will cause shutdown) appears to be

HC = 30+NHS

for single heat sinks, and it probably should be

HC = 30+2*NHS

for double heat sinks.

Edit: Zyllos has done a test (in an Awesome) that appears to contradict the above formula for HC with single heat sinks. However, I later repeated his test (in a Catapult) and found results consistent with the formula, although I initially screwed up the analysis and got a wrong conclusion. That has been fixed. I have not repeated Zyllos's test, as I don't own an Awesome. If someone who does could try to repeat it, that would be great. See posts 8 and 11.

The "Heat Efficiency" statistic in the MechLab is also of interest. Its behavior has been decoded and is discussed here: http://www.reddit.co...ing_worked_on/. Based on experimental evidence, the number displayed in the MechLab does not appear to give a useful indication of the heat sinks' ability to dissipate the heat generated by weapons fire. At present, it should probably be ignored.

Methodology, Results, and Discussion

I have performed in-game tests, using a stopwatch along with the above equation for single heat sinks, to determine THG for various energy weapons. So far, I have only tested mechs with single heat sinks and standard engines. The results are somewhat consistent with a certain interpretation of the values contained in the XML file, but this is not conclusive (yet; see below).

These are not the only discrepancies, however. It appears that, for engines that have zero or more heat sink slots, the engine heat sinks operate at reduced effectiveness until at least one additional heat sink is added and placed outside the engine. This probably should not be happening. (The problem may also be fixed by placing heat sinks inside the engine, when possible, but I have not tested this.) Consistent with the above, I have not observed any discrepancies for engines that require at least one heat sink to be placed externally in order to connect to a match (e.g., the 245 standard in the Jenner).

I have used the hypothesis above about how the heat mechanic works to calculate the heat produced per weapon. Basically,

THG = THD,

or, equivalently,

[number of salvoes]*[number of weapons]*[heat per weapon] = 0.1*NHS*T,

so that

[heat per weapon] = 0.1*NHS*T/([number of salvoes]*[number of weapons]),

where T was determined by recording the time the heat scale took to go from zero to zero with a stopwatch. Care was taken to ensure that the heat scale remained positive at all times, so that the heat sinks were always functioning and THD = 0.1*NHS*T was valid. All tests were performed out of the water on Forest Colony, Frozen City, or River City and with the throttle set to zero. There exists a general consensus that these three maps have the same heat effects.

The experimental results follow, although there are not yet enough to draw strong conclusions. I regret that they are not as complete or systematic as they could be, but I have limited time.

First comes an assortment of tests using mechs with heat sinks placed outside the engine. These don't really form part of my argument and are just here for reference/use in future discussion. The numbers are the times taken for the heat scale to "zero" (i.e., T) for the different tests. They were averaged and then used in the formula above.

CPLT-C1(F)
standard engine 260
25 single heat sinks
12 salvoes of 2 large pulse lasers
River City
69.0. 68.8, 69.1, 68.9 => 69.0 average => 7.18 heat/laser

CPLT-C1(F)
standard engine 260
31 single heat sinks
20 salvoes of 2 large lasers
River City
99.9 => 7.24 heat/laser

AWS-8T
standard engine 240
23 single heat sinks
10 salvoes of 2 large lasers
63.0, 63.2, 62.8, 63.2 => 63.1 average => 7.25 heat/laser

CN9-AL
standard engine 200
16 single heat sinks
20 salvoes of 2 medium lasers
Forest Colony
103.3 => 4.13 heat/laser

CN9-AL
standard engine 200
16 single heat sinks
10 salvoes of 2 medium lasers
River City
52.0, 52.0, 52.1 => 52.0 average => 4.16 heat/laser

Now for the important part. These are tests of medium pulse lasers using the Catapult. Around 4 heat appears to be generated for all numbers of heat sinks except 10 (no heat sinks outside the engine), where the number is instead around 5. However, in that case, assuming that only 8 of the stated 10 engine heat sinks are actually working, we again find around 4 heat generated. Given that the lasers should produce the same amount of heat regardless of the number of heat sinks carried, this should not be happening.

CPLT-C1(F)
standard engine 260
10 single heat sinks
2 salvoes of 4 medium pulse lasers
River City
38.4, 39.9 => 39.2 average => 4.89 heat/laser
but if we assume 8 engine heat sinks => 3.92 heat/laser

CPLT-C1(F)
standard engine 260
10 single heat sinks
2 salvoes of 4 medium pulse lasers
River City
40.7, 39.9, 38.5, 40.0, 40.0, 39.9 => 39.8 average => 4.98 heat/laser
but if we assume 8 engine heat sinks => 3.98 heat/laser

CPLT-C1(F)
standard engine 260
11 single heat sinks
2 salvoes of 4 medium pulse lasers
River City
29.6, 29.3, 29.4, 29.6 => 29.5 average => 4.05 heat/laser

CPLT-C1(F)
standard engine 260
12 single heat sinks
2 salvoes of 4 medium pulse lasers
Frozen City
27.0, 26.9, 27.2, 27.0, 27.0, 27.0, 27.1, 26.9 => 27.0 average => 4.05 heat/laser

CPLT-C1(F)
standard engine 260
31 single heat sinks
14 salvoes of 4 medium pulse lasers
River City
72.4, 72.3, 72.2 => 72.3 average => 4.00 heat/laser

Here are similar results, except for the Atlas and this time using medium lasers. We again find 5 heat generated for 10 single heat sinks (no heat sinks outside the engine) but 4 for 11 heat sinks (one outside the engine). Again, assuming that only 8 heat sinks are effective corrects the discrepancy. I have not performed more tests, because I installed double heat sinks on my Atlas last night, and I don't want to pay to remove/reinstall them. You are welcome to perform more and post the results here. This would also be a good opportunity to see if adding a heat sink to the engine also fixes the discrepancy. My guess is that the same thing is happening as with the Catapult using medium pulse lasers, however. Also, notice how comparatively "glitchy" the numbers are for 10 heat sinks compared to 11. I'm not sure whether this is repeatable or just a client/server communication issue.

AS7-D(F)
standard engine 300
10 single heat sinks
2 salvoes of 4 medium lasers
River City
38.9, 40.2, 42.5, 39.1, 46.7, 39.4, 40.4, 38.7, 43.8, 40.8 => 41.1 average => 5.13 heat/laser
but if we assume 8 engine heat sinks => 4.10 heat/laser

AS7-D(F)
standard engine 300
10 single heat sinks
2 salvoes of 4 medium lasers
Forest Colony
37.4, 39.8, 39.3, 39.0, 41.6, 38.3, 38.6 => 39.1 average => 4.89 heat/laser
but if we assume 8 engine heat sinks => 3.92 heat/laser

AS7-D(F)
standard engine 300
11 single heat sinks
2 salvoes of 4 medium lasers
River City
30.5, 30.7, 30.5, 30.7, 30.6, 30.7 => 30.6 average => 4.21 heat/laser

In the following three tests, I again used a Catapult, but instead tested small lasers. They appear to produce around 1.6 heat. The discrepancy with no engine heat sinks is again observed and can be corrected as before.

CPLT-C1(F)
standard engine 260
10 single heat sinks
6 salvoes of 4 small lasers
River City
51.8, 50.6, 49.4 => 50.6 average => 2.11 heat/laser
but if we assume 8 engine heat sinks => 1.69 heat/laser

CPLT-C1(F)
standard engine 260
11 single heat sinks
6 salvoes of 4 small lasers
River City
35.2, 35.3, 35.6, 35.5, 35.2 => 35.4 average => 1.62 heat/laser

CPLT-C1(F)
standard engine 260
16 single heat sinks
10 salvoes of 4 small lasers
Forest Colony
40.3, 40.4, 40.1, 40.1 => 40.2 average => 1.61 heat/laser

Additional Discussion: Interpretation of the Values in ItemStats.xml

Averaging across the various tests (except for those with only engine heat sinks) for the different weapons tested and rounding to the nearest tenth, we find the following results for total heat generated per weapon:

small laser: 1.6
medium laser: 4.2
large laser: 7.2

small pulse laser: n/a
medium pulse laser: 4.0
large pulse laser: 7.2

The XML file gives different values in the "heat" field. It has been suggested that this field really indicates the rate of heat generation, so that the total heat generated would be given by the number in the "heat" field times the number in the "duration" field. Thus we would have:

small laser: 2 "heat", 0.75 "duration" => 1.5 total heat;
medium laser: 4 "heat", 1 "duration" => 4 total heat;
large laser: 7 "heat", 1 "duration" => 7 total heat;

small pulse laser: 3 "heat", 0.5 "duration" => 1.5 total heat;
medium pulse laser: 5 "heat", 0.75 "duration" => 3.75 total heat;
large pulse laser: 9 "heat", 0.75 "duration" => 6.75 total heat.

These numbers are much closer to the computed results than are those in the "heat" field. However, the fit is still not all that great. I am tempted to explain this away as error introduced by client/server communication issues, but I deliberately took measurements over long time periods (> 25 s) to minimize this problem. More data are needed to draw any strong conclusions.

Future Work

All of the above pertains to "normal" maps (i.e., not Caustic Valley) and "normal" conditions (i.e., zero throttle, not jumping, and not standing in water). It is possible that each of these conditions incurs a proportional reduction in the number of effective heat sinks (i.e., the total is reduced by a given percentage). Once the bugs are ironed out and we have a good idea of how much heat each weapon actually produces, these would be good things to look into.

Credits

There was a thread on the beta forums a few days ago that contained some nice discussion on issues similar to the ones addressed here. Some information on it can be found here: http://www.reddit.co...ing_worked_on/. The main contributors were (I believe) Mar Helmer, Galen Cox, and CptPlanet, who had been running experiments similar to mine. Also, in one of the followups, someone (you know who you are) posted the idea that the "heat" values in the XML file are actually rates of heat production. I am not sure who first came up with the relation for total mech heat capacity. Also, I apologize to all the people irritated by my standing around and doing nothing while shooting walls.

Edited by Amaris the Usurper, 07 November 2012 - 03:20 PM.

[betocorp]

Good work, this explains a lot of things, like I overheating early while building to have exacly 15s heat capacity

And for DH? they work? I didnt used because 1M + to upgrade to DH is a lot of $ this week.

[Amaris the Usurper]

Apparently, DHS located outside the engine function normally, but those located inside the engine function like SHS. I haven't done any testing on this myself, but that is what I hear. For example, a mech with 20 DHS, 10 of which are in the engine, would dissipate heat like a mech with 10+2*(20-10) = 30 SHS. The DHS bug is in addition to the one that I noticed and discussed above for engine SHS.

Edited by Amaris the Usurper, 30 October 2012 - 01:22 PM.

[MustrumRidcully]

So your theory is now that beam duration and the itemstats.xml value for heat work together to generate the "total " heat of the weapon?

But why this special mechanic for lasers - it's not like a PPC generates no heat at all? Is that really all function as they intended it to do?

I remember that they said something about changing how heat is applied - is this an unintended side effect of that, or by design?

I mean, in the end, the overall heat values all create severe balance issues in the game, but these findings add another level of weirdness to the heat system.
I am not convinced that PGI themselves really knows how it all interacts.

I would actually consider filing a ticket on this and ask if this behaviour is intended...


Also, kudos to the effort of practical testing. That's an area I am too bored by, I prefer getting values from others (those that datamine or experiment like you) and then just analyse the values. I would be useless without people like you!


BTW, the heat efficiency system in MW:O has a fundamental flaw - it basically calculates the ratio between heat generated and heat dissipated. But that has nothing to do how fast a mech actually overheats.

A mech that produces 2 heat per second and dissipates 1 heat per second, and a mech that produces 4 heat per second and dissipates 2 heat per second would have the same ratios, but they would still not overheat in the same time frame.

The first mech produces 1 waste heat per second and has a total capacity of heat of 40 - that means he'd last 40 seconds.
The second mech produces 2 waste heat per second and as a total capacity of 50 - so he'd overheat in 25 seconds. Quite a difference!

I take this just as another data point that PGI doesn't really have a firm grip on the implications of the heat system for the gameplay. Or at least didn'T have the time to actually implement something of more practical value.

Edited by MustrumRidcully, 30 October 2012 - 01:20 PM.

[Amaris the Usurper]

Quote

So your theory is now that beam duration and the itemstats.xml value for heat work together to generate the "total " heat of the weapon?

Exactly. This is my best current idea for what the ItemStats.xml numbers mean.

[total heat] = ["heat" number from ItemStats.xml]*["duration" number from ItemStats.xml]

However, as I said, while the predicted numbers match the experimental data more closely if we assume this, the matching still leaves something to be desired. Mostly, it bothers me that both large and large pulse lasers are measured to produce roughly 7.2 heat, when the above theory predicts 7 and 6.75. Why would they be the same?

Quote

But why this special mechanic for lasers - it's not like a PPC generates no heat at all? Is that really all function as they intended it to do?

I'm not sure. For weapons (like PPCs) that have zero duration, it appears that the total heat is given in the "heat" field and that the rate interpretation is only valid for continuous-beam weapons. It is also possible that the total heat values were at one point given in ItemStats.xml, but stopped being used later. It may be just a coincidence that multiplying the "heat" and "duration" fields gives fairly good numbers.

We appear to have the same understanding of how long it would take a mech to overheat.

Edited by Amaris the Usurper, 30 October 2012 - 04:54 PM.

[Derek Icelord]

Just a thought on your heat theory for lasers:

Your estimations give pulse lasers a lower heat generation than the standard laser when they are supposed to be higher. It's totally possible PGI mucked that bit up, but it does make me stop and think a bit.

Perhaps something along the lines of:

• Heat in the ItemStats.xml file is the amount of heat generated once the weapon is fired.
  
• For lasers, the heat begins to dissipate even as the beam is in its "on" duration, thus a medium laser has 1 second to "burn" while its heat is being dissipated.
  
• An additional disconnect might be that since the heat bar slides up instead of jumping, the animation masks some of the heat output since heat is dissipated in the short time from the weapon firing to the heat meter fully displaying heat.

[EtherDragon]

Another thing that is confusing is that each heat-sink increases your max threashold by "1", so the over-all percentages change based on your max heat threashold.

[Zyllos]

Derek Icelord, on 30 October 2012 - 04:27 PM, said:

Just a thought on your heat theory for lasers:

Your estimations give pulse lasers a lower heat generation than the standard laser when they are supposed to be higher. It's totally possible PGI mucked that bit up, but it does make me stop and think a bit.

Perhaps something along the lines of:

• Heat in the ItemStats.xml file is the amount of heat generated once the weapon is fired.
  
• For lasers, the heat begins to dissipate even as the beam is in its "on" duration, thus a medium laser has 1 second to "burn" while its heat is being dissipated.
  
• An additional disconnect might be that since the heat bar slides up instead of jumping, the animation masks some of the heat output since heat is dissipated in the short time from the weapon firing to the heat meter fully displaying heat.

I am *ALMOST* certain this is how it works. When the weapon is fired, the heat is generated as the weapon is continuing it's fire. So ACs, PPCs, LRMs, ect all generate their heat as per the XML file. As this heat is generated, heat is continuously being reduced. So for instant duration weapons, they basically will have no way of reducing the heat so it is all front loaded. Lasers on the other hand have their heat dissipated as they are fired. This can easily be seen with a mech that has a lot of heatsinks and firing a SBL. The heat bar never moves.

But I was messing around with my ER-PPC on my 28 SHS Awesome and was getting some odd numbers:

Assuming the maximum heat of a mech is 30 + HS (I think DHS actually only increase the this value by 1, as it is 1 heatsink), this Awesome should have a maximum value of 58 before shutting down.

16 shots of the ER-PPC caused the mech to shutdown.
16 * 3s = 48s of continuous fire (and heat dissipation)
2.8RHD * 48s = 134.4 heat dissipated
16 * 13 heat = 208 heat generated
208 - 134.4 = 73.6 heat generated
73.6 / 58 = 1.27 (127% on the bar)

This is an issue because to get 127% over the bar, that means I needed to fire at ~105%, which is past the shutdown line. 13 / 58 = 0.22 (22%), which seems correct as every shot ended up showing about a 21% increase, which I assume the 1% missing is due to the heat being dissipated as the bar is moving up to be displayed (it does not automatically move up/down with each frame but instead slowly makes it way to the current heat during each update of the frame to "smooth" the bar fluctuation).

Either way, more work is needed to draw more concrete conclusions.

[Amaris the Usurper]

Derek Icelord, on 30 October 2012 - 04:27 PM, said:

Just a thought on your heat theory for lasers:

Your estimations give pulse lasers a lower heat generation than the standard laser when they are supposed to be higher. It's totally possible PGI mucked that bit up, but it does make me stop and think a bit.

Perhaps something along the lines of:

• Heat in the ItemStats.xml file is the amount of heat generated once the weapon is fired.
  
• For lasers, the heat begins to dissipate even as the beam is in its "on" duration, thus a medium laser has 1 second to "burn" while its heat is being dissipated.
  
• An additional disconnect might be that since the heat bar slides up instead of jumping, the animation masks some of the heat output since heat is dissipated in the short time from the weapon firing to the heat meter fully displaying heat.

The measurements show the same total heat generation (about 7.2) for large and large pulse lasers, but the theoretical predictions (based on my best guess of what the ItemData.xml numbers mean) show nearly the same values (7 and 6.75, respectively). These numbers are fairly close (within a few percent); the differences may be due to measurement error, although I'm not convinced of this.

My equation for the rate of change of heat (dH/dt) explicitly takes into account the fact that heat dissipates while the lasers are firing, as does the integrated form which states that the total heat generated equals the total heat dissipated (THG = THD). This is what I used to compute the experimental weapon heat values. The results are totally independent of the movements of the heat scale (I did not even record the peak percentage). The only things that matter are the number of weapon firings, the number of heat sinks, and the time taken for the heat scale to return to zero.

I have seen other people report the peak percentage reached after firing a laser, sometimes inferring that (for a given number of heat sinks) this will be proportional to the total heat generated. However, since, as you said, heat dissipates while the laser is firing, the peak percentage will be somewhat lower than would be expected if all the heat were generated instantaneously (like with a PPC, missile launcher, or autocannon), so this is not true. Also, I have observed the peak percentage to be rather flaky--it seems to vary up and down quite a bit, depending on connection quality.

In summary, we are in complete agreement on your last two points. That is what I see happen on the heat scale when I fire lasers, and I have already taken it into account.

EtherDragon, on 30 October 2012 - 04:47 PM, said:

Another thing that is confusing is that each heat-sink increases your max threashold by "1", so the over-all percentages change based on your max heat threashold.

This is also true and should be kept in mind. According to my theory, the peak percentage after firing a number of lasers (of the same type) should be

[peak percentage] = 100*(RHG-0.1*NHS)*[laser duration]/(30+NHS)

for SHS and (probably)

[peak percentage] = 100*(RHG-0.2*NHS)*[laser duration]/(30+2*NHS)

for DHS, after they are de-borked. All terminology is the same as in the first post. You would either find RHG by [total heat generated per laser]*[number of lasers fired]/[laser duration] and predict [peak percentage], or you would measure [peak percentage] and solve the equation for RHG.

Edit 1: Maybe we will have HC = 30+NHS for DHS, as well. My basis for the other formula was that the TT heat scale goes to 30 before automatic shutdown, but only the excess heat ([total heat generated]-NHS for SHS or [total heat generated]-2*NHS for DHS) actually makes it onto the scale at the end of the turn. Thus, in order to have the same "alpha strike" ability in MWO when using insta-heat weapons (PPCs, etc.) as in TT for a given variant, we would need to have HC = 30+2*NHS when using DHS.

Edit 2: Based on work by Zyllos, the formula above for SHS appears to be wrong (and I would rather not talk about the second one until DHS are fixed and it can actually be tested). However, I repeated his tests (though in a Catapult instead of an Awesome) and found results consistent with the formula for SHS. See posts 8 and 11.

Edited by Amaris the Usurper, 31 October 2012 - 06:27 AM.

[Amaris the Usurper]

Here are some results for the remaining energy weapons (small pulse laser, ER large laser, PPC, and ER PPC; nobody cares about flamers). A CPLT-C1(F) with a 260 standard engine and varying numbers of SHS was used. The methodology was the same as before.

Frozen City
16 SHS
20 salvoes of 4 small pulse lasers
77.7, 77.8 => 77.8 average => 1.56 heat/laser

Frozen City
29 SHS
10 salvoes of 2 ER large lasers
71.3, 71.2 => 71.25 average => 10.3 heat/laser

River City
29 SHS
10 salvoes of 2 ER large lasers
71.7, 71.3 => 71.5 average => 10.4 heat/laser

Frozen City
25 SHS
10 salvoes of 2 PPCs
72.0, 71.7 => 71.9 average => 8.98 heat/PPC

Frozen City
25 SHS
5 salvoes of 2 ER PPCs
51.8, 51.7 => 51.8 average => 12.9 heat/PPC

Below are the collected averages from the individual sessions on each different map. The first expected value assumes that "heat" in ItemStats.xml refers to total heat generated, while the second assumes that the [total heat generated] = ["heat" value from XML file]*["dissipation" value from XML file] theory is true. If only one value is given, the predictions are the same.

small laser: 1.62, 1.61 (expected 2 or 1.5)
medium laser: 4.13, 4.16, 4.21 (expected 4)
large laser: 7.24, 7.25 (expected 7)
ER large laser: 10.3, 10.4 (expected 10)

small pulse laser: 1.56 (expected 3 or 1.5)
medium pulse laser: 4.05, 4.05, 4.00 (expected 5 or 3.75)
large pulse laser: 7.18 (expected 9 or 6.75)

PPC: 8.98 (expected 9)
ER PPC: 12.9 (expected 13)

Edited by Amaris the Usurper, 31 October 2012 - 08:22 AM.

[Amaris the Usurper]

Zyllos, on 30 October 2012 - 05:12 PM, said:

I am *ALMOST* certain this is how it works. When the weapon is fired, the heat is generated as the weapon is continuing it's fire. So ACs, PPCs, LRMs, ect all generate their heat as per the XML file. As this heat is generated, heat is continuously being reduced. So for instant duration weapons, they basically will have no way of reducing the heat so it is all front loaded. Lasers on the other hand have their heat dissipated as they are fired. This can easily be seen with a mech that has a lot of heatsinks and firing a SBL. The heat bar never moves.

But I was messing around with my ER-PPC on my 28 SHS Awesome and was getting some odd numbers:

Assuming the maximum heat of a mech is 30 + HS (I think DHS actually only increase the this value by 1, as it is 1 heatsink), this Awesome should have a maximum value of 58 before shutting down.

16 shots of the ER-PPC caused the mech to shutdown.
16 * 3s = 48s of continuous fire (and heat dissipation)
2.8RHD * 48s = 134.4 heat dissipated
16 * 13 heat = 208 heat generated
208 - 134.4 = 73.6 heat generated
73.6 / 58 = 1.27 (127% on the bar)

This is an issue because to get 127% over the bar, that means I needed to fire at ~105%, which is past the shutdown line. 13 / 58 = 0.22 (22%), which seems correct as every shot ended up showing about a 21% increase, which I assume the 1% missing is due to the heat being dissipated as the bar is moving up to be displayed (it does not automatically move up/down with each frame but instead slowly makes it way to the current heat during each update of the frame to "smooth" the bar fluctuation).

Either way, more work is needed to draw more concrete conclusions.

That's an interesting result. The formula for total heat capacity is the weakest part of what I originally posted. I have done no research on it myself and was just going by what others had said. I don't own an Awesome, but I did attempt to reproduce your results in my CPLT-C1(F), fitted with 28 SHS and an ER PPC. Firing the ER PPC continuously on Forest Colony resulted in a shutdown on the 13th shot. After powering back up, I still had heat on the scale. It took exactly one minute from the start of firing for the heat scale to return to zero. That's 169 heat over 60.0 seconds = 2.82 heat dissipated per second, compared with a theoretical prediction of 2.8, which is comforting. On the other hand, by my calculations, if n is the number of times the ER PPC has been fired, we should have (assuming 13 heat per shot and a 3 second delay between shots)

H = 13*n-2.8*3*(n-1) = 13*n-8.4*(n-1) = 4.6*n+8.4

for the absolute heat level immediately after then nth shot. Since my mech shut down on the 13th shot, the heat capacity would have been between 63.6 and 68.2 (corresponding to n = 12 and n = 13 in the above). Obviously, this is bigger than the predicted 58.

One possible explanation appeared to be that the the delay between shots was a little longer than recorded in ItemStats.xml, so that more cooling was actually going on between shots. However, I repeated the test on Forest Colony and found that it takes 38.7 seconds to fire 13 shots, which is in agreement with a 3 second firing delay. It looks like the heat capacity formulas are wrong and that, in actuality, larger mechs may have an inherent heat capacity advantage (since you took 16 shots to reach shutdown).

Like you said, we need many more tests across different chassis (and possibly with different engine types?) before we can draw any firm conclusions.

Edit: I need to put away my jump to conclusions mat (or at least not do math after midnight). Since I started timing with the first shot, 38.7 seconds is actually the time it takes to fire 12 shots, not 13. That makes for 38.7 seconds / 12 shots = 3.23 seconds per shot. Thus the real recycle time is probably 3.25 seconds, not 3, as in ItemStats.xml. Reworking my formula above, we would have

H = 13*n-2.8*3.25*(n-1) = 3.9*n+9.1

for the absolute heat level after the nth shot. Thus, the heat level would be 55.9 after the 12th shot and 59.8 after the 13th shot, which is in complete agreement with the theoretical prediction, assuming a heat capacity of 58. With that said, I am not sure what was happening in your Awesome test.

Edited by Amaris the Usurper, 31 October 2012 - 11:16 AM.

[dtgamemaster]

Just want to post some constructive opinions here.

1) In layman terms. Basically you are timing the duration from firing the weapon a number of times(the more the better) till all the heat dissipates away. Total Heat that is generated by the weapons = Total Heat that goes into the system from start to end = Total Heat that is dissipated by the Heatsinks(when heat is in the system).

Find the total heat that is dissipated by the heatsinks and you can get heat generated by weapons.

2) This is the problem. A very big assumption was made.
"RHD = 0.1*NHS"
1 single heat sink dissipate 1 heat in 10 seconds.
But does this apply for 10, 16, 25, 31..etc.. heatsinks?
It has been experimentally proven that the number of heat sinks does not scale linearly with heat dissipation. Either that or there is some problem with the heat dissipation model.

And this is the problem we face.
How much heat does the heat sink in numbers actually dissipate?????

3) Someone pointed out that values derived from the game client should not be wrong. Another has pointed out that the game depend on server values (which is hidden from us and can be tweaked???), otherwise we can edit our own XML files to cheat at the game.

I am neutral on this issue and hope the developer can give us a definitive answer if the heat values presented in Ohmwrecker's chart is accurate or not.

4) Finally thank you for presenting this new methodology of testing, It has given us a lot of food for thought and to work out the actual mechanic of the heat system.

[Amaris the Usurper]

dtgamemaster, on 30 October 2012 - 10:35 PM, said:

Just want to post some constructive opinions here.

1) In layman terms. Basically you are timing the duration from firing the weapon a number of times(the more the better) till all the heat dissipates away. Total Heat that is generated by the weapons = Total Heat that goes into the system from start to end = Total Heat that is dissipated by the Heatsinks(when heat is in the system).

Find the total heat that is dissipated by the heatsinks and you can get heat generated by weapons.

2) This is the problem. A very big assumption was made.
"RHD = 0.1*NHS"
1 single heat sink dissipate 1 heat in 10 seconds.
But does this apply for 10, 16, 25, 31..etc.. heatsinks?
It has been experimentally proven that the number of heat sinks does not scale linearly with heat dissipation. Either that or there is some problem with the heat dissipation model.

And this is the problem we face.
How much heat does the heat sink in numbers actually dissipate?????

3) Someone pointed out that values derived from the game client should not be wrong. Another has pointed out that the game depend on server values (which is hidden from us and can be tweaked???), otherwise we can edit our own XML files to cheat at the game.

I am neutral on this issue and hope the developer can give us a definitive answer if the heat values presented in Ohmwrecker's chart is accurate or not.

4) Finally thank you for presenting this new methodology of testing, It has given us a lot of food for thought and to work out the actual mechanic of the heat system.

Thanks for your input!

1) Exactly.

2) More generally, I am saying that

RHD = [constant]*NHS

for SHS. I chose [constant] = 0.1 because of broad consistency with the ItemStats.xml values, except for all pulse lasers and the small laser (all other energy weapons generate the amount of heat indicated). However, assuming that "heat" in ItemStats.xml actually indicates the rate of heat generation (and not the total heat generated) largely removes the inconsistency. Also, I just found experimentally for the ER PPC that the recharge time is 3.25 and not 3 as stated, so it is possible that some of the beam duration times are also not accurate. Using the correct times (from experiment) might remove the remaining (small) errors.

Regarding the formula, I do mean exactly what I say: 10 heat sinks would give RHD = 1, 11 heat sinks would give RHD = 1.1, 12 heat sinks would give RHD = 1.2, and so on. So I do believe that heat dissipation scales linearly with the number of heat sinks (in contradiction to what has been stated by some others).

Let's look at some of the consequences of this idea. Even if the constant of proportionality is not 0.1, we would still have

RHD = [constant]*NHS

for SHS. Consequently, we would also have for total heat generated (THG)

THG = [constant]*NHS*T.

This implies that

THG/(NHS*T) = [constant]

for all chassis, numbers of heat sinks (NHS), and measured dissipation times (T). The weak link appears to be THG, since the numbers from ItemStats.xml appear to be flaky in some way. Without solid data about heat generation, we can't determine the constant.

However, there is a way around this. I don't think anyone would disagree that a given weapon should always produce the same amount of heat, regardless of the chassis, number of heat sinks, etc. Thus, we should have

THG = [number of shots]*[heat per shot]

for a given weapon. Now, let's consider two different heat tests. Plugging in the expression given for THG above, we should have

[number of shots]₁*[heat per shot]/(NHS₁*T₁) = [constant],

and

[number of shots]₂*[heat per shot]*(NHS₂*T₂) = [constant],

where the subscripts indicate the values for Test 1 and Test 2. Therefore (after we divide these two equations),

[number of shots]₁/(NHS₁*T₁) = [number of shots]₂/(NHS₂*T₂)

for any two tests involving multiple firings of a given weapon, assuming linear scaling, regardless of the constant of proportionality. Also, if this is true, then the the heat sink scaling law is linear.

Let's compute the [number of shots]/(NHS*T) figure for various tests and verify that the result is always the same when using a given weapon. Starting with medium lasers:

CN9-AL w/ 16 SHS took 103.3 s to dissipate the heat of 40 shots.
Result: 40/(16*103.3) = 0.02420

CN9-AL w/ 16 SHS took 52.0 s to dissipate the heat of 20 shots.
Result: 20/(16*52.0) = 0.0240

AS7-D(F) w/ 11 SHS took 30.6 s to dissipate the heat of 8 shots.
Result: 8/(11*30.6) = 0.0238

Medium pulse lasers:

CPLT-C1(F) w/ 11 SHS took 29.5 s to dissipate the heat of 8 shots.
Result: 8/(11*29.5) = 0.0246

CPLT-C1(F) w/ 12 SHS took 27.0 s to dissipate the heat of 8 shots.
Result: 8/(12*27.0) = 0.0247

CPLT-C1(F) w/ 31 SHS took 72.3 s to dissipate the heat of 56 shots.
Result: 56/(31*72.3) = 0.0250

Large lasers:

CPLT-C1(F) w/ 31 SHS took 99.9 s to dissipate the heat of 40 shots.
Result: 40/(31*99.9) = 0.0129

AWS-8T w/ 23 SHS took 63.1 s to dissipate the heat of 20 shots.
Result: 20/(23*63.1) = 0.0138

The numbers are very consistent; if there is any nonlinearity in the scaling of heat sink effectiveness with the number of heat sinks, it appears to be negligible.

3) Your guess concerning what the ItemStats.xml values mean is as good as mine. Input from the devs would be great.

4) You're welcome!

Edited by Amaris the Usurper, 31 October 2012 - 11:19 AM.

[Culler]

There's a known bug with heat dissipation, I believe to the effect that each heat sink past 20 working significantly less effectively than say the 12th heat sink does. This is in addition to the bug where the engine DHS work as SHS.

There's a post about it in the command chair. I'd run these tests again once those are fixed if I were you.

[Amaris the Usurper]

Culler, on 31 October 2012 - 08:14 AM, said:

There's a known bug with heat dissipation, I believe to the effect that each heat sink past 20 working significantly less effectively than say the 12th heat sink does. This is in addition to the bug where the engine DHS work as SHS.

There's a post about it in the command chair. I'd run these tests again once those are fixed if I were you.

The post in Command Chair just confirms the 0.1 and 0.2 numbers for SHS and DHS and that engine DHS are currently functioning as SHS. Since I have been testing with SHS only, this is not relevant. Also, in repeated testing, my formulas give consistent answers for the heat generated by a given weapon, regardless of the number of heat sinks. I have not found any discrepancies for large (> 20) numbers of heat sinks.

The only exception is that when using engine SHS only (no additional heat sinks fitted), the engine HS appear to operate at less than 0.1 heat/second efficiency, or (looked at another way) there are fewer than 10 engine heat sinks, contrary to what is stated in the MechLab.

Edited by Amaris the Usurper, 31 October 2012 - 09:07 AM.

[Amaris the Usurper]

Here are some results for LRMs, using the CPLT-C1(F) with the same methodology as before.

11 SHS
Forest Colony
20 salvoes of 2 LRM 5s
72.7 => 2.00 heat/launcher

11 SHS
River City
8 salvoes of 2 LRM 10
58.2 => 4.00 heat/launcher

14 SHS
Forest Colony
8 salvoes of 2 LRM 15s
57.2 => 5.01 heat/launcher

14 SHS
Forest Colony
8 salvoes of 2 LRM 20s
68.6 => 6.00 heat/launcher

The numbers are in close agreement with those in ItemStats.xml.

Edit: Here are some results for SRMs.

11 SHS
Forest Colony
10 salvoes of 2 SRM 4s
54.5 => 3.00 heat/launcher

14 SHS
Forest Colony
10 salvoes of 2 SRM 6s
57.1 => 4.00 heat/launcher

Again, there is very close agreement. It's looking like there are only discrepancies with the lasers, although I haven't tested any ACs yet.

Edited by Amaris the Usurper, 31 October 2012 - 10:06 AM.

[MustrumRidcully]

Amaris the Usurper, on 31 October 2012 - 09:01 AM, said:

The post in Command Chair just confirms the 0.1 and 0.2 numbers for SHS and DHS and that engine DHS are currently functioning as SHS. Since I have been testing with SHS only, this is not relevant. Also, in repeated testing, my formulas give consistent answers for the heat generated by a given weapon, regardless of the number of heat sinks. I have not found any discrepancies for large (> 20) numbers of heat sinks.

The only exception is that when using engine SHS only (no additional heat sinks fitted), the engine HS appear to operate at less than 0.1 heat/second efficiency, or (looked at another way) there are fewer than 10 engine heat sinks, contrary to what is stated in the MechLab.

Hmm. Maybe this is the long standing bug. I noticed that light mechs (which often did never hav emore than engine heat sinks) work extremelyI bad on Caustic. Of course Caustic is harsh, but it seemed too harsh on Light Mechs. I figured this was just Caustic's general problem and an inherent imbalance of the mech. But if it was something else...

We'll see next week perhaps maybe?

[IamSeanConnery]

Hey guys, I've done a lot of work myself on this and came up with the following calculations:
Heat Dissipation rate= HEAT_SINKS * TYPE_OF_SINK * MAP_HEAT * COOLRUN_EFF - ENGINE_HEAT
where:
HEAT_SINKS = heat sinks under or equal to 12 + heat sinks over 12 * 0.65
TYPE_OF_SINK = (0.125 for single heat sink, 0.15 for double heat sink)
MAP_HEAT = (0.8 for Caustic, every other is 1
COOLRUN_EFF = (1.15 for Elite unlocked, 1.075 for Basic unlock, 1 for none)
ENGINE_HEAT = (0.3 for XL engines, and for 200STD 0.3 + 0.02 for every 10 engine rating over that)

Also the Heat/Second values are off for some of the weapons I tested:
Sm Laser should be 0.55555
Med Pulse Laser should be 1.03333
Sm Pulse Laser should be 0.606
Med Laser .989
etc...

Edit: I'm not saying all this data is correct, but it does predict results in the game fairly well. Usually within 1/10 of a heatsink. I'm open to updates based on test results.

Edited by Platinum Booger, 01 November 2012 - 10:34 AM.

[Amaris the Usurper]

Here are some results for autocannons. This time, I have both heat and refire delay ("cooldown" in ItemStats.xml, but I don't want to introduce confusion by using this term) data. The methodology for the heat tests is the same as before. I used the HBK-4G(F) with standard engine 200, unless otherwise noted. For the refire delay tests, I simply held down the fire button, began timing when the first shot fired, and stopped timing when the last shot fired. If 10 shots were fired, this time would correspond to 9 refire delays, and so on. The values from ItemStats.xml are show in parenthesis.

10 SHS
River City
15 salvoes of 3 AC/2s
45.1 => 1.00 (1) heat/autocannon
10.5 seconds to fire 15 shots => 0.746 (0.5) second refire delay

100 standard engine
10 SHS
River City
20 salvoes of 3 AC/5s
60.0 => 1.00 (1) heat/autocannon
36.6 seconds to fire 20 shots => 1.93 (1.7) second refire delay

100 standad engine
10 SHS
Forest Colony
20 shots from 1 AC/10
59.9 => 3.00 (3) heat
51.4 seconds to fire 20 shots => 2.71 (2.5) second refire delay

13 SHS
Forest Colony
14 shots from 1 AC/20
75.4 => 7.00 (7) heat
54.7 seconds to fire 12 shots => 4.97 (4) second refire delay

As you can see, all standard autocannons appear to be producing the correct amount of heat; the matching is too close to be coincidence. On the other hand, all of the refire delays appear to be too long.

Also, here are some refire delay data for missiles.

LRM 5
30.5 seconds to fire 10 shots => 3.39 (3.25) second refire delay

SRM 6
37.5 seconds to fire 10 shots => 4.17 (4) second refire delay

SRM 4
34.9 seconds to fire 10 shots => 3.88 (3.75) second refire delay

These appear to be consistent with ItemStats.xml.

Edited by Amaris the Usurper, 01 November 2012 - 01:37 PM.

[Arcturious]

I'm glad the devs have stated the fix will be in on the 6th. What is more interesting is that it will include a long standing bug resolved, and that both combined will make a significant difference to heat.

My own testing has shown that it is purely number of heat sinks that matter, not type or the heat efficiency rating. Same for engines, you actually get more heat dissipation when taking a small engine needing more heat sinks than the 10 built in.

Hopefully these upcoming changes leave heat finally making sense. Until they go in, all discussion is pretty moot as it will all change in just under a week.