Thermoelectric Generator Statistical Analyzer

[kremlin]

I've been playing around with trying to get absurd power outputs from the TEG via hellburns and I have a pretty good idea of exactly what's going on and what makes them work. I have an idea that would both help folks figure it out and also assist seasoned generator operators in fine-tuning their builds as to truly push the limits on what the TEG can do, without giving anything away too easily.

It consists of three entities:

• Terminal in reactor control room for information display, linked to:
  
• A floor mounted sensor in the combustion chamber tracking temperature, pressure, and gas composition (in relative amounts).
  
• A similar sensor that tracks the same variables as above but in a pipe.
  

There'd be one floor sensor in the center of the combustion chamber, and four pipe sensors places at the input and output sides of the hot and cold loops. Both sensors would be indestructible.

The computer terminal would display information collected from these sensors and present them in (perhaps) a few ways, depending on what y'all think is reasonable/are willing to implement/etc.

The first (and simplest) interface would be a simple, realtime monitor of the three aforementioned gas characteristics (temp, pressure, gas comp.). The gas composition data could be as simple as just the relative concentrations of each gas in terms of percentages, and the numbers displayed upon interacting with the terminal would blink and change with each atmosphere tick similar to how most of the other generator computers work, with discrete data points being continuously updated. While simple, this would only really save us the trouble of running around the loops & coaxing staff assistants to poke their head in the combustion chamber to get a reading. What'd I'd like to see goes beyond this.

Secondly, extrapolating on the information from before, we could provide more nuanced data on the gas in the loops:

• As loop pipes have a fixed volume, we could deduce and display the molar gas content instead of just relative amounts (this gets tricky because the atmospheric system relies on arbitrarily sized bignums to describe gas content, which are infinitely divisible, unlike discrete molecules/atoms. We'd have to pick an arbitrary number that coincides bignums with the concept of molarity. Or think of a better idea).
  
  
  
• Pipe sensors placed across critical junctures (like the input/output sides of the TEG, the input/output sides of the combustion chamber, etc) could derive the pressure differential, and along with the above information could deduce the drift velocity of the gas through the pipes.
  
  
• Taking the above information and factoring in the heat capacity of the measured gasses, we could deduce a derived amount describing the amount of energy passing through each juncture. This is one step behind what the TEG itself does, which is display the difference between such amounts in the hot/cold loops: the energy generated!)
  

FINALLY, the most complicated interface, but by far the most helpful one, would be to display the data points listed so far in an integrated fashion, as opposed to a single, transient discrete data point per tick. Things get a little complicated here so I will draft up some equations to more concretely describe what I'm talking about.

Once you know what you're doing, it's easy (more or less) to start a hellburn that doesn't burn out or out-gas so long as long as you don't exhaust (or detonate) your burn gasses. There is a critical threshold you cross that separates a doomed, convergent hellburn that will eventually die from a proper one that will diverge into a preposterous, neutron-star-level amount of generated energy. This last interface would directly convey that information in a way that someone who is really trying to figure this shit out will understand, and not in a way that just gives it up too easily. It'll also assist seasoned generator ops to fine-tune their configurations, further optimizing them to suit the relatively short SS13 round times.

Instead of providing blinking numbers that describe transient, discrete data points, we could integrate these points as a function of time. Oh boy calculus.

There are several non-derived unit-similar categories of variables to track. They are grouped and listed below with asterisks denoting how important I think each one is, along with symbols describing them. Red denotes absolute quantities while blue represents differential quantities:

• Temperature of combustion chamber interior (*****) - Tci
  
• Temperature across hot loop segment passing through combustion chamber (****) - Tcc
  
• Temperature across hot/cold loop pipe segments (***) - Tc, Th
  • This changes when the pipes burst, along with pressure, etc.
    
• Temperature across hot/cold loop generator in/out nodes (*) - Tgh, Tgc
  

• Pressure of combustion chamber interior (*****) - Pci
  
• Pressure across hot loop segment passing through combustion chamber (****) Pcc
  
• Pressure across hot/cold loop pipe segments (***) - Pc, Ph
  
• Pressure across hot/cold loop generator in/out nodes (*) - Pgh, Pgc
  

• Gas content of combustion chamber interior (*****) - Mo, Mp, Mc
  • Molarity of Oxygen, Plasma, CO2 respectively
    
  • or whatever we choose, instead of moles, like dimensionless coefficients
    

We take the derivative of these quantities with respect to time, yielding the instantaneous rate of change for P/T/M. You can more or less eyeball this yourself watching extent computers, but having it calculated would help as well as be critically important for something later on. I'm only going to do ones for temperature and you can assume the pressure/molarity expressions are similar, just swap T for V or whatever:

Assuming F(Tci) is the function describing the temperature on a tile in the combustion chamber...
(The atmospheric engine's concept of time is tick based, be aware)



The first derivative, again, tells us something we pretty much see already. HOWEVER -- the second derivative is where the magic is. A deceivingly functional configuration that is doomed to fail or a mediocre configuration that will take 50 minutes to reach 10 MW would look much the same to someone only looking at the first equation. If we can show the rate at which change is accelerating, then we can have a pretty clear idea from intial setup how the reactor will play out, and with more or less instant feedback here, we'd be able to hone in on the parameters to make a really powerful reactor.



phew

I realize I've written a huge wall of text and that this isn't a trivial feature and unlikely to be just picked up by one of the devs, which is why I am going to work on a rudimentary implementation (using the goonstation 2016 leak). I am pretty certain I am capable of getting this idea most of the way there, but I'd like some input. There are some more things I wanted to cover but this is enough for now.