Efficient Room Shape and Thermodynamics

[dkmoo]

Having played a lot of cold biomes, I wanted to figure out the best insulation methods. While many tests has already been done to conclude things like airlocks, double walls, etc.. I didn't see a whole lot test done on actual room shapes other than 1 post concluding using square rooms.  In real life, thermodynamics works by minimizing surface area/perimeter to reduce heat transfer, so in theory, circles work best.  In game, for the same room size, it makes sense that square rooms retain heat better than rectangular rooms, but taking that logic further, circular rooms should be even better. The problem is that making circles on a grid system is effectively a bunch of smaller horizontal and vertical sides that make up a curve, so if the game counts each of those edges, then circular rooms could have more "edges" than a square room, making it less heat efficient. I did my own little test. Results are rather interesting:

Short version: Circular rooms are better, but avoid 1x1 diagonals (ie, 45 degree) when making your curves.

Long version: See pic. Did three tests with 8 setups. each test has same room size. Outside temperature is -152C.
Test 1: Three rooms of 25 square size. Setup A retained most heat and B the least. C in the middle
Conclusion: It looks both the horizontal and vertical "edge" of each diagonal block is counted so heat is lost horizontally AND vertically. Making diagonal walls inefficient. So it looks like minimizing the "edges" is the way to go - A has least number of "edges" (20), followed by C (24), then B (28). Their temperatures follow the same pattern, making B the worst, even though it has the smallest "perimeter" in terms of wall blocks.

Test 2: 2 rooms of 61 square size. D retained more heat than E.
Conclusion: Even though both has the same number of "edges" (36). The circular room in this case is better because the "curves" are formed with 1x2 blocks instead of 1x1 block. This means that heat is only lost through the "long" edge, because the "short" edge is effectively a 2-layer wall, making heat loss from that direction minimal. In this set up, thermodynamics is working as nature intended. the circular set up has fewer "effective" edges so smaller heat loss.

Test 3: 2 rooms of 37 square size. G retained more heat than F
Conclusion: Confirmation of test 1 and 2 - Even though "circle" in theory is more efficient, using 1x1 (45 degree) diagonal walls will negate any efficiency gains from perimeter-minimization effects of a circular set up.

Therefore - make circles, but don't use 45 degree walls to make the curves.



[attachment deleted by admin due to age]

[Skinnay]

Interesting choice of shapes to test. I'd be curious to see how a plain square does in comparison, and also a tile count next to each design. It almost looks like E is a better design than D because it has more inside space. Hard to tell. Good work though. 

[dkmoo]

Interesting choice of shapes to test. I'd be curious to see how a plain square does in comparison, and also a tile count next to each design. It almost looks like E is a better design than D because it has more inside space. Hard to tell. Good work though.

Thanks - E and G are actually my substitutes for the "plain" square. The problem of using a true square is that it's very hard to make them have the same area as a circle that can be generated in the grid system, (without making them insanely large).  The little dingers you see in the bottom of E and G are just to achieve the same area as D and F, respectively, and they have very minimal impact to the temperature b/c they don't create extra "effective edges" noted in test 2. Having the same area is the control in these tests - ABC has the same area, DE has the same area, and FG has the same area. 

The closest one think i looked into was a 6 unit radius circle with 101 square unit area, vs a 10 x 10 square room, but even that isn't an exact comparison since its 101 vs 100.