PSA: Door heat injectors delete heat

[Zarquan]

If you do high temperature things (e.g. petroleum boiling or geothermal steam turbines), then you have likely used doors to inject heat in to your system from something hotter.  You likely have something like this:

Imagine the steam bar is replaced with something like a magma biome or a volcano.

When I ran this system, I was getting numbers on my petroleum boiler that didn't make sense.  The energy used from the steam bar was too high, sometimes to an absurd degree over a 5 cycle average.

It turns out that when a door closes, there is a heat-change deletion bug, where the door sets itself to the average temperature of the two solid tiles of the door at the time the door is ordered closed.  That means that if your door is changing temperature quickly, it could delete a significant amount of heat.

In my trials, I found that I was losing between 2% and 25% of the heat spent by this petroleum boiler, depending on the reasonable actions I would take not knowing about the bug.

Major takeaway:  Power your heat injection doors!  The less time they spend in the closing animation, the less heat they can delete.

But if you want to avoid heat deletion altogether, there are options.

1. Liquid Shutoff Heat Injector
   1. This works by using a liquid shutoff and a friendly liquid to inject heat via a loop.  This slowed 0% heat loss.  However, for a lot of the temperatures we want to see, we generally need a liquid like uranium (with it's amazing temperature range and high SHC).  The material cost is also extremely low for your rare materials like uranium:  This setup only uses 70 kg, but you could easily get away with less.  You can also use a conveyor rail with a highly thermal capacity solid like refined carbon or genetic ooze (e.g. non-perishable food).
      
      
      
      I tried using steam, but there are severe limitations due to the low mass of a steam packet.  This also means that lead is also a bad choice for this, as it has a low enough SHC that it can't run it.  One benefit is that you can run both a liquid and a gas if you want to, so perhaps a combination of steam and lead could work.
       
2. Use physical liquid.
          This works by creating a temporary thermal connection between the hot side and the cold side.  It can transfer heat quite efficiently given a good amount of thermal connection.  I have two examples:
   1. Vertical:  A liquid bead falling.  Here, I use a classic 5 kg/s naphtha driven liquid pump to send lead up to an EZ-Bead pump, where the bead is broken by an airflow tile before it hits the sensitive machines (all made of steel or copper).  I set it up so that the liquid pump only runs if there is no lead detected by the liquid pipe element sensor.  This build requires no DLC to run, as it runs off lead and naphtha, but if you have uranium, it is way better in all ways because uranium is awesome.  EDIT:  This approach seems to delete heat, not sure where.  
   2. Horizontal:  Similar, but without the EZ-Bead Pump and all that stuff.  Here, I use 80 kg of lead as the liquid thermal conductor, which is more than enough to run 10 turbines from the magma biome, meaning this can be made without any DLC.  Of course, as in all things, uranium is far better.  The more lead you have, the more energy appears to transfer, but there is a limit at 99.99 kg of lead, as you do not want it to flow.  You could also try aluminum, but it has a higher melting point, limiting how much energy can be extracted.  EDIT:  This approach also deletes heat.  I think there might be a heat deletion bug relating to pumping. 

[tyraelpl]

Thank you! 

[JRup]

3 hours ago, Zarquan said:

EDIT:  This approach seems to delete heat, not sure where.

I would make sure to monitor any liquid volumes that are subjected to falling, either from a liquid vent or an escher waterfall (any kind of falling) as this is in the realm of liquid disappearing.

I use super coolant gas with a meter valve to provide heat to a flaking boiler, even steam will provide sufficient thermal mass to heat up a regular petroleum boiler if it's decently efficient.

Speaking of flaking. How has the update affected the steam turbine with flaking boiler? (Asking because some right to left flows are presenting a bit of an issue. Still monitoring.)

[Zarquan]

24 minutes ago, JRup said:

I would make sure to monitor any liquid volumes that are subjected to falling, either from a liquid vent or an escher waterfall (any kind of falling) as this is in the realm of liquid disappearing.

I don't think any of the mass is disappearing in that one and it is also happening with the other liquid blob approach, where the liquid is pumped away to break the connection.  Liquid pumps aren't meant to be used like this.  I have a theory that some of the energy might be vanishing in to the pipes, so I'm running it again without pumping the naphtha.

 

[Charletrom]

I accidentally discovered this bug making a niobium tamer. The doors were powered but the temperature change was very fast due to the high molten niobium temps. I had 5x geotuned the volcano as well so things were extra spicy.

[Zarquan]

5 hours ago, JRup said:

Speaking of flaking. How has the update affected the steam turbine with flaking boiler? (Asking because some right to left flows are presenting a bit of an issue. Still monitoring.)

Flaking seems fine as far as I can tell.  Liquid to liquid is still a little broken unless the tile under the flaking is an airflow tile and water to steam is still producing 25 kg/s.

[JRup]

2 minutes ago, Zarquan said:

Flaking seems fine as far as I can tell.

Thanks, I also checked my build thoroughly to find some areas that had some heat accumulation that shouldn't be there; tl;dr.:edge cases... Re-tiled, fixed.

[Zarquan]

6 hours ago, JRup said:

I would make sure to monitor any liquid volumes that are subjected to falling, either from a liquid vent or an escher waterfall (any kind of falling) as this is in the realm of liquid disappearing.

This pump-system not working is bizarre to me.  Take a look at this:

This is a system that is completely closed and doesn't have anywhere natural to dump its heat, but, running a trial for 5 cycles, I measured a heat usage of 323580 MJ, when the theoretical and measured energy usage (using a shutoff loop) should be closer to 112790 MJ.  That heat isn't going anywhere I can see, so it must have been destroyed, which is absurd and far worse than doors.

[JRup]

20 minutes ago, Zarquan said:

hat heat isn't going anywhere I can see, so it must have been destroyed

I take it you've already accounted for the 1.5°C rebound that happens on phase change.

I'm also trying to wrap my head around this particular setup.

[Zarquan]

47 minutes ago, JRup said:

I take it you've already accounted for the 1.5°C rebound that happens on phase change.

I'm also trying to wrap my head around this particular setup.

The idea is that I use a small quantity of molten lead to send the heat up in to the boiler by standard conduction.  It is significantly thermally connected to the lower region via diamond temp shift plates and weakly connected to the upper area with a less conductive obsidian temp shift plate, chosen for it's low thermal conductivity and low SHC.  EDIT:  Sorry, referring to an older version.  The upper temp shift plate is aluminum, as I needed more thermal conductivity when I went to the lower amount of lead. (I was originally using 30 kgs and a big pump.  Note, that system lost less heat.)

When the thermosensor in the heat battery under the boiler hits 406 C, it orders the liquid vent to dump the lead (which is 3999.9 g for some reason, I think I did that on purpose but I don't remember why) on to the ground.  When the temp is measured above 406 C, the mini liquid pump runs for 7 seconds, more than enough time to clear out the lead.  It is being tricked in to thinking it is submerged by the nearby naphtha because of a mystery about pump range.  Once all the lead is pumped, the boiler is disconnected from the heat source.

The heat source is the bar of steam that the lead sits on, which is preheated 1000 kg of steam at 1000 K.

Somewhere in this process, heat is destroyed and I don't understand where.

I also did take in to account the 1.5 C in my theoretical value.

[Yalp]

Back to the Door-Topic:

Edit TL/DR:

aCtUalLy: We can copy any temperatures indefinetly and completely trivialize heat management with a few doors and a bit of automation

EditEnd

 

Doors only save the temperature of their Root tile (bottom tile) when they are opened.
This ignores the fact, that a closed door consits of two tiles, ignoring the temperature of the top tile and effectivly copying the Root Tile Temperature when closing the door!

 

This takes effect immediately after the open command was issued by automation, enabeling us to copy temperatures at a significant rate without actually opening the doors.
(I tested this 5 years ago: 1.2 s closed/0.1 open had the best result.
In theroy 1 tick of heat transfer would be optimal, but faster timings had the doors open/close completely on occation - depending on simulation strain i guess)

Lining up a number of doors can amplify this until there is no temperature change in the initial temperature source (usually 3-5 doors iron / steel).

Because of the door orientation it is directional (copying upwards and from left to right)

 

I'm actually  using this in my current run, somewhat as a demonstration since I am streaming it... (usually I avoid *advanced mechanics* this strong)
Cooling 30 kg/s of SG with a 0% uptime thermo regulator.
(no debug/Baator map mod)
the bottom tile was cooled once many cycles ago and has been sitting at this temp ever since...

Spoiler

 

[Zarquan]

4 hours ago, Yalp said:

Back to the Door-Topic:

Edit TL/DR:

aCtUalLy: We can copy any temperatures indefinetly and completely trivialize heat management with a few doors and a bit of automation

EditEnd

 

Doors only save the temperature of their Root tile (bottom tile) when they are opened.
This ignores the fact, that a closed door consits of two tiles, ignoring the temperature of the top tile and effectivly copying the Root Tile Temperature when closing the door!

 

This takes effect immediately after the open command was issued by automation, enabeling us to copy temperatures at a significant rate without actually opening the doors.
(I tested this 5 years ago: 1.2 s closed/0.1 open had the best result.
In theroy 1 tick of heat transfer would be optimal, but faster timings had the doors open/close completely on occation - depending on simulation strain i guess)

Lining up a number of doors can amplify this until there is no temperature change in the initial temperature source (usually 3-5 doors iron / steel).

Because of the door orientation it is directional (copying upwards and from left to right)

 

I'm actually  using this in my current run, somewhat as a demonstration since I am streaming it... (usually I avoid *advanced mechanics* this strong)
Cooling 30 kg/s of SG with a 0% uptime thermo regulator.
(no debug/Baator map mod)
the bottom tile was cooled once many cycles ago and has been sitting at this temp ever since...

  Reveal hidden contents

 

That isn't the behavior I've been seeing.  When I open a door, it takes the average of the two door tile from the moment the door is ordered to be opened, not setting it to the lower tile's temp. 

Is this because you don't fully open the door?  That could make sense because the door's building temp is the temp of the lower tile and both tiles are set to the building temp when closed.  And if the door never opens fully, it might not average.

EDIT: But this I think is also a different bug from the one I'm seeing.  Heat is lost when the door is fully opened even if heat is being added exclusive to the bottom tile if the door opens all the way.

[Yalp]

1 hour ago, Zarquan said:

Is this because you don't fully open the door?

apparently, kinda, but not exactly TIL:

 

Unpowered:

Dupe opens/closes -> average

Debug open/close -> root tile

Quick automation (0.2s green) -> root tile

Signal switch (open fully) -> average

 

Powered:

Dupe -> average

Signal switch (open fully + waiting time) -> root .... oh wait, also average

 

I only did the other tests once and the last one twice ... i did not intend to do any testing... enough for now .

 

Weirdly enough, i have a way higher delta_t between the two tiles of the door this time ~1 K as opposed to 0,03 K. Same doors, reloaded after 5 h or so