High Temperature Thermo Sensors

[impyre]

Since the rate of heat exchange is fixed and linear, you can easily create a high-temperature sensor by "sampling" the temperature rather than reading it directly. This does decrease accuracy marginally, but it still works fairly well for anything you might be doing at high temperatures (where accuracy to the degree often doesn't matter as much.)

The idea is that you send in a gas (hydrogen works best of course) at a fixed, known temperature. This requires tight temperature regulation and good insulated ventilation. Split off a very small packet (1 gram in my case), and send it through a heat exchanger (like the two pieces of radiant gas pipe below). Since it's so small, it doesn't really affect what it's measuring very much, and it changes temperature *very* quickly. The last step is to combine this small packet of gas with another packet that's still at the original temperature. I used 10 grams of hydrogen at 99.8C in total; split it into a 1g packet and a 9g packet, then recombined them.

Because of how thermal energy works, the resulting temperature is basically a mass-weighted average. In this case, every 10 degrees of change in the sample packet causes about 1 degree of change in the final packet once it's re-added. Also, because the small packet gets to the same temperature as whatever I'm measuring, the change is directly proportional to the temperature of the thing I'm measuring.

571.6C in the sample becomes 147C on the sensor, and 486.9C becomes 138.6.

84.1C change in the sample equates to 8.4C change in the readout.

I currently have my sensor set to 138.8, which turns out to be about 489C. I chose this number because I'm using magma as a heat source, and it gets hot pretty fast. It's hard to keep the room controlled within a tight range because of how fast it heats up. It does stay between 571 and 489C, but I plan on using this room to heat up another room where I process oil into petroleum. However, if you drop the thermo sensor down to around 130C it tends to hover nicely in the 400C-500C range that's good for processing oil.

Here's the setup I used:
 

Spoiler

The calculations aren't too difficult, and if anyone wants to know more I'd be happy to help. Also, the math makes way more sense in kelvin.

[Saturnus]

Usually this type of temperature sensor uses chlorine instead of hydrogen since you don't have to have two separate pipes that are merged together. And you can have a more precise direct measurement without affecting the temperature of what is measured at all. See here for example.

 

[Icywolf]

If you use something like chlorine, you should be able to escape the merging etc, which makes this a pretty complicated setup.

[impyre]

@Icywolf That's a good point. I chose hydrogen because it has a *high* conductivity. I was trying for accuracy, consistency, and speed more than anything else. 1 gram of hydrogen carries 2.4 DTU per degree C, this means that to increase my sample from 100C to the measured temperature (around 500C max) only requires 960 DTUs. In comparison, a tile filled with 20kg of hydrogen at 500C has about 37 million DTUs. This means the sample extracts .0026% of the thermal energy in a single tile. Even a 2kg tile only bumps that to .026%, not even a tenth of one percent. In contrast, chlorine has a heat capacity that is 20% that of hydrogen. It takes a fifth the energy to heat up... however, the difference between .026% and .005% is truly negligible.

If you examine thermal conductivity, hydrogen's is .168 and chlorine's is .008. Hydrogen absorbs heat energy 21 times faster than chlorine. I *want* my sample to heat up, and the quicker it does so, the more compact the design can be and the more quickly it can react to changes. Also, you can compare chlorine and hydrogen while considering *both* heat capacity *and* conductivity, this is basically "thermal responsiveness". How easily it heats and cools. You divide the thermal conductivity by the specific heat to get an idea of how easily it heats and cools. This reflects the fact that the most responsive material is one that has high conductivity and low specific heat, while the least responsive material is one that has low conductivity and high specific heat. For hydrogen we get .07 and for chlorine we get .0167. Hydrogen is a little more than 4 times as responsive as chlorine.

I was going for something that can be used on a wide range of temperatures with little or no modification, and get consistent results. By ensuring that the sample reaches the same temperature as what you're actually measuring, you get an accurate reading every time... and it *always* scales by how much you merge the packet with. 10x scaling? 1/10. 2x scaling? 1/2. Simple. It's configurable in the field for any reference temperature with no math or work needed... it just works out of the box. You only need a single test point, and you can easily calculate whatever temperature you need.

Using time/space/mass limited heat transfer to measure a temperature works relatively well but it has several limitations. 

A) The results are highly dependent on the difference between the temperature of the reference gas and what's being measured.

B) The results are non-linear: using a 100C packet to measure something at 120C will get very different results than measuring something at 300C.

C) The results only work for a specific thermal equilibrium.

D) It's fiddly, it takes quite a bit of experimentation to get right for each implementation.

@Saturnus Any time you take heat up one thing by exposing it to another, some heat transfer occurs... it's inevitable. "More precise and direct" can mean a lot of things, but in this case I disagree. In any case, even if you use chlorine and get it tweaked properly, in order to get the same advantages my system offers, you'd still need to split it up so you can accurately reference the measured temp to a baseline. The two systems may look very similar but they operate on different principles.

There are basically two ways of extending our sensor's range... temperature averaging and temperature sampling. Temperature averaging is reliable and consistent, but slow to respond to sudden changes. There are a lot of ways you can do it, so it's pretty versatile. Temperature sampling involves exposing a substance at a known temperature to an unknown temp for a fixed amount of time and measuring the change. Sampling is very responsive to sudden changes, but is less consistent at temperature extremes (very different or very similar temps). This is because it hinges on the fact that conductivity is based (in part) on the difference between the two temperatures. Big delta-t means more heat is exchanged; however, there are other factors that also affect conductivity of the system as a whole... including what your test medium is.

I chose to go the averaging route, like wachunga did in his post.

However, instead of relying on thermal transfer between adjacent tiles to average the temperatures (which is slow), my design averages them instantly in the pipe. When two packets are combined their temperature is automatically averaged. So this has all the advantages of thermal averaging and the advantages of sampling too.

 

[Icywolf]

@impyre The thing about using lower SHC and conductivity is how much energy it will take you to bring it down to normal temperature (say 273.15K) after the measurement is taken. You don't care how fast it heats up - you have 1 second exposure (ideally) and you measure the delta T. You want sampling as you want to be able to react to spikes in temperature, and you want to be able to do it fast. Averaging in your case just adds complexity, and I fail to see the benefit of it. You are diluting your sample in order to be able to measure it, when you could be using something that does not heat up as much and measure that. It might be easier to calculate (as you said you dilute it 10* times it shows 1/10th temp) compared to multiplying the result by x, but once you know x multiplication becomes easy. 

The main benefits of using gas like chlorine is keeping the build simpler/smaller footprint and not using as much energy to normalize the sample after the measurement. 

[Saturnus]

17 minutes ago, impyre said:

 

@Saturnus Any time you take heat up one thing by exposing it to another, some heat transfer occurs... it's inevitable. 

You'd think so but not. That's why chlorine clamp work. If you have 1g of chlorine or less in a pipe heating or cooling it does not transfer any energy.

18 minutes ago, impyre said:

"More precise and direct" can mean a lot of things, but in this case I disagree. In any case, even if you use chlorine and get it tweaked properly, in order to get the same advantages my system offers, you'd still need to split it up so you can accurately reference the measured temp to a baseline.

That's why the chlorine system is superior. You do not need to split it up because you do not need a baseline as the temperature change is a direct relationship between the hot side and the cold side.

For example here.

Valve is set to 1g/s. Right chamber is petroleum at 0C. Left side is petroleum at 450C.
One of the insulated pipes through the hot petroleum is obsidian. The rest is ceramic.
Temp sensor stabilize at 270.1C. Ran it 10 cycles. No chance in temperature anywhere.

[impyre]

@Saturnus

The thermal energy in *one* tile of 740kg petroleum at 450C is 941,635,200 DTUs.... A single gram of chlorine at the same temperature has 347 DTUs... a single gram of hydrogen has 1105 DTUs.  A single gram of chlorine at 0C has 131 DTUs, and hydrogen has 655.

To change a single tile (let alone 4) of petroleum by a tenth of a degree you'd need to transfer 130,240 DTUs.

A single gram of chlorine running between 0 and 450 sides will (assuming complete transfer) 216 DTUs per second. Which means you'd see a tenth of a degree difference in one cycle for one tile. Four cycles for four tiles. If you allow half of the transfer it'd take twice the time.

The hydrogen does the same thing, but at a rate of 450 DTUs per second. I can assure you heat is transferred with both chlorine and hydrogen. I have tested it extensively. Hydrogen does transfer heat about twice as fast as chlorine in this setup, but again as I said earlier.... in the case of the thermo sensor setup we're taking a drop out of an ocean. It's not surprising you wouldn't notice it... but I promise you that if you run the test with smaller quantities of petroleum (like 2kg in one tile instead of 740kg in four tiles) you can measure the change easily. 

[Saturnus]

1 minute ago, impyre said:

I can assure you heat is transferred with both chlorine and hydrogen. 

Nope. Chlorine doesn't transfer heat because the rate of change per second is small enough to be ignored due to rounding errors.
 

[impyre]

4 minutes ago, Saturnus said:

Nope. Chlorine doesn't transfer heat because the rate of change per second is small enough to be ignored due to rounding errors.
 

I think I get what you're saying. So my test is flawed because smaller mass of petroleum encourages higher delta-t, but with a really small delta-t the game just makes it zero. So how do we know that doesn't work the same for hydrogen too?

Edit: I'm going to find out lol

[Saturnus]

Just now, impyre said:

I think I get what you're saying. So my test is flawed because smaller mass of petroleum encourages higher delta-t, but with a really small delta-t the game just makes it zero. So how do we know that doesn't work the same for hydrogen too?

You can try with hydrogen. It doesn't work. It works with chlorine. It might work with other gases with very low SHC, such as CO2, but  it's not something I've tried.

Btw, the set up I posted above have now run on for another 50 cycles as I was testing something else at the same time. No change in temperature anywhere in the system.

[fredhp]

I hate that 300C limitation :(. When automation update was developed, 300C was sufficient. But now, with volcanoes, high temperature materials,  and rockets with crazy heat generation i think Klei should change the limit to, at least, 4000C.

As i said in another thread, i dont do any kind of oil to petroleum cooker due to this temperature sensor limitation - i convert oil to sour gas an then to natural gas instead. If the sensors limit changes, i will invest time into a oil to petroleum cooker. Lots of "crazy things" will be possible with high temperature sensors!

[Lilalaunekuh]

6 hours ago, fredhp said:

I hate that 300C limitation

=> I am using melting resources to build "my" high temperature sensors.

(Melting gold is close to optimal for shutting down a aquatuner made of thermium)

But I really like the idea of splitting and merging gasses to measure temperature and I will do some testing with merging "superheated" liquids

(If your water packets are small enough you can heat them above 1000°C without bursting your pipes. Maybe "superheated" water would be able to achieve a better accuracy compared to a hydrogen-based system)