Ok, it makes a bit more sense now!
That's a pretty setup, very extensive, and it's a valid implementation of a heat die, and should give somewhat consistent results.
but if you look a little closer...
Because you don't know the actual coefficient of performance, you're having to calculate it, which introduces an unknown error plus, you have to take no less than five temperature measurements, each at +/- 1.0 deg C, in order to figure out the actual power applied. Then you have an unknown error, from the placement of all these temp probes (the side die probe can be off by as much as 3 deg C).
You could tighten it up, with +/- 0.1 deg C temp probes, but you still have a couple of unknown variables.
Also, as I discussed in e-mail with your partner (as you know), you'd have an issue with repeatability, because you can't maintain the same clamping pressure on that pelt. As I suggested, you could torque those nuts, lock them with another nut, and use a thread locker.
Then, as you use it, you'd have to leave the free plate alone, because if you mount anything on it, you're varying the compression. So it has to be the cold side that's got the clamping screws running freely through it.
With all of the formulaes that are used, you then have to calculate the amount of variation to the end result, from the accuracy of each temp probe.
The formulae also takes into account the current applied to the TEC, so you'd have to measure that, with its own margin of error, which again, has to be factored in.
So if you take a quick look at a heater cartridge, where the power applied is directly measured, with an error from the volt and amp measurement, it's a lot more straightforward, with a smaller margin of error. In this case, There's *STILL* a need for some expensive measuring instruments!
Then there are the secondary losses, applicable to any setup... and the list goes on and on...