|
Post by shurugal on Dec 8, 2016 21:46:20 GMT
I've just had a thought about thermocouples...
The biggest limit i run into with making them more efficient is that as you increase the heat gradient across them, you increase the thermal expansion stress, requiring a larger thermocouple to absorb the stress, which in turn reduces the power output by spreading out the heat gradient as a side-effect.
What if we designed the thermocouples to only function at the desired temperature? If we spaced out the hot and cold side components so that they had to heat to the target temperature before they came into contact, this would eliminate the expansion-related stresses, allowing for much smaller thermocouples and much larger heat gradients.
To give a real-life example, the SR-71 Blackbird underwent so much heat expansion in its frame at operational speed, that all of its fluid passageways had to be engineered to fit snug at flight temperature, which meant that on the ground, it leaked like a sieve.
Is there any reason thermocouples could not be designed to work the same way?
|
|
|
Post by newageofpower on Dec 9, 2016 0:16:32 GMT
This is actually pretty intelligent. It may lead to some failure modes we haven't thought of before, but overall, inefficiencies in low power states are more than worth it for increased inefficiencies in a high power state.
Plus, we run our reactors within 5k of meltdown anyhow. Don't see what's so different here ; )
|
|
|
Post by lawson on Dec 9, 2016 5:11:19 GMT
Modeling a more sophisticated thermocouple geometry would help as well. Adding angled radial slots wouldn't add much resistance but would allow the inner surface to expand a lot more without over stressing the outer surface of the thermocouple disks.
|
|
|
Post by n2maniac on Dec 9, 2016 8:25:52 GMT
If the hot side anneals in use, this would automatically happen actually. Operating the material near its melting point is likely to do this. Doing this means the material will creep over time, though. Also, gonna suck if the reactor output ever drops faster than the annealing rate:
"Laser blasted off, power draw has dropped, dropping reactor power... oh no, the thermoelectrics cracked, hot liquid sodium is going everywhere!"
|
|
|
Post by shurugal on Dec 9, 2016 12:42:08 GMT
If the hot side anneals in use, this would automatically happen actually. Operating the material near its melting point is likely to do this. Doing this means the material will creep over time, though. Also, gonna suck if the reactor output ever drops faster than the annealing rate: "Laser blasted off, power draw has dropped, dropping reactor power... oh no, the thermoelectrics cracked, hot liquid sodium is going everywhere!" divert power to resistojet until new load is equalized?
|
|
|
Post by coaxjack on Dec 9, 2016 23:49:23 GMT
If the hot side anneals in use, this would automatically happen actually. Operating the material near its melting point is likely to do this. Doing this means the material will creep over time, though. Also, gonna suck if the reactor output ever drops faster than the annealing rate: "Laser blasted off, power draw has dropped, dropping reactor power... oh no, the thermoelectrics cracked, hot liquid sodium is going everywhere!" divert power to resistojet until new load is equalized? Things start going wrong, so you just blast away in a panic. We call this the 'Squid Ink Maneuver'.
|
|