Do pipes exchange heat when within wall tile?

[SchlauFuchs]

5 hours ago, wachunga said:

Edit: Also it looks like your sheet is using H2 for everything instead of H3 or H4 where appropriate.

Good catch, with the H2 issue fixed the curves look better:

5 hours ago, wachunga said:

The problem with your math is that averaging, log or otherwise, isn't a thing. Despite it being the common wisdom on these forums it doesn't exist. Or perhaps it exists in some specific circumstances, but I have yet to find evidence of it.

I tested this quite extensively in the past and some brief retesting doesn't show anything has changed. But you shouldn't believe me more than anyone else. It's trivially easy to test in debug and I encourage you to do so. I'd be happy to know if I missed a change.

Okay, we need to find out if that averaging has been applied or not, because I remember having seen that in one of the Developer logs around the cosmic Upgrade, with lots of people moaning about their colony suddenly dying the heat death.

When exactly did you test it? 

5 hours ago, wachunga said:

Cell/Cell transfers are a lowest conductivity relationship.

Cell/Building transfers are a multiplicative relationship.

Pipe/Contents transfers are a multiplicative relationship.

What is a multiplicative relationship, how is the math here?

[wachunga]

The first post I linked was shortly after the update that supposedly changed things, so I tested it before and after. Most of the whining was a result of people seeing things which didn't actually exist. Again you don't have to believe me, it's easy to test for yourself. The two posts I linked give examples of tests you can do.

I don't know the exact math and I don't care, it's not important. What's important is that the conductivities of the two materials are multiplied together. Tungsten interacting with something transfers 4x as much heat as wolframite interacting with that same thing. If it was an average of conductivities, this would not be the case. I provided a test you can easily do in the second link, try it.

If I sound curt it's because I've had this conversation numerous times with people who refuse to test things for themselves and instead vomit dogma in my face when I've spent quite a bit of time trying to help them and this community.

It's entirely possible I screwed something up, but I've yet to see evidence of that. No offense meant, it's just super irritating.

[wachunga]

In an effort to explain why the devs said the thing about averaging, I've revisited the issue. After more testing it appears that averaging may be occurring during the pipe/contents transfer. As before, I can't find evidence of it in cell/cell or cell/building transfers. I can't recall the exact materials and temperatures I used when I tested pipe/contents so it's possible I overlooked it due to the difficulties in testing that. Or maybe it changed since then.

Due to the poor range of conductivities to test with I am not super confident but it does make sense in retrospect. The pipe/contents transfer mechanism is what changed in that patch, so it's logical that the patch notes were about that mechanism. However this appears to be an arithmetic average and not a log average.

A 20C tungsten pipe segment with 499.9C petroleum inside heats up by 55.5C after one tick. A 20C wolframite pipe segment with 499.9C petroleum inside heats up by 15.2C after one tick. This is a 3.65x difference. The arithmetic average would be a 3.65x difference. The log average would be a 2.65x difference (check my math). Multiplying the conductivities would be a 4x difference (this is what happens in cell/building transfers). This suggests an arithmetic average, but again testing pipe/contents is somewhat difficult.  The arithmetic average and conductivities multiplied together is often indistinguishable, given the conductivities we can use. I could very well be wrong about this. If anyone has testable evidence to the contrary please share it.

 

In conclusion:

Cell/Cell: Governed by the lowest conductivity

Cell/Building: Governed by the conductivities multiplied together

Pipe/Contents: Governed by the conductivities' arithmetic average ((a+b)/2) 

 

[martosss]

5 hours ago, wachunga said:

In an effort to explain why the devs said the thing about averaging, I've revisited the issue. After more testing it appears that averaging may be occurring during the pipe/contents transfer. As before, I can't find evidence of it in cell/cell or cell/building transfers. I can't recall the exact materials I used when I tested pipe/contents so it's possible I overlooked it due to the difficulties in testing that. Or maybe it changed since then.

Due to the poor range of conductivities to test with I am not super confident but it does make sense in retrospect. The pipe/contents transfer mechanism is what changed in that patch, so it's logical that the patch notes were about that mechanism. However this appears to be an arithmetic average and not a log average.

A 20C tungsten pipe segment with 499.9C petroleum inside heats up by 55.5C after one tick. A 20C wolframite pipe segment with 499.9C petroleum inside heats up by 15.2C after one tick. This is a 3.65x difference. The arithmetic average would be a 3.65x difference. The log average would be a 2.65x difference (check my math). Multiplying the conductivities would be a 4x difference (this is what happens in cell/building transfers). This suggests a simple average, but testing pipe/contents is somewhat difficult as the arithmetic average and conductivities multiplied together is often indistinguishable, given the conductivities we can use. I could very well be wrong about this. If anyone has testable evidence to the contrary please share it.

In conclusion:

Cell/Cell: Governed by the lowest conductivity

Cell/Building: Governed by the conductivities multiplied together

Pipe/Contents: Governed by the conductivities' arithmetic average ((a+b)/2)

OK, I tried to redo your test. However, I didn't use ticks, I used full seconds - I split a pipe in 2 sections and let  one 10kg packet of petrolium travel and heat 1 section of pipe. Then when it's empty(they alternate, so there's 1 empty pipe segment between packets) I look at the temperature. Pipes are in vacuum so no outside heat is transfered. The results after 1sec:

Tungsten heats by 183.4°, Wolframite heats by 90.1°. That means the ratio is 193.4 / 90.1 = 2.759

Now log average - tungsten-petrolium  and wolframite-petrolium yield 17.0528 and 6.452, respectively. The ratio is 2.643.

Now averages - (31 and 8.5) is 62/17 = 31/8.5 = 3.64.

Now I'd say the log-avg(2.643) is reaaaly close to the actual ratio(2.76), much closer than the pure average(3.64), so I believe the log average is used in the pipe-content transfer.

That way of testing is really smart - different materials with the same heat capacity, but different thermal conductivity. So that means building content is calculated by the same formula, i.e. the one that is used for pipe-content? That would mean it matters a lot from which material you build farm tiles if you plan to supply them with hot water.