Simulate the Cool-down Process in Detail

savreline commented 7 years ago

@nagimov This will be done in the following parts (see image) Part 1 (18hr real time, 2 days machine time) - cool down of LN2 only. Helium is stationary and is only cooled due to conduction from LN2. This requires significant modification of the code as He must be stationary - so have to kick Helium out of h_solver, change data structure and write a new solver for He. As He is all gas at such high T, we can get away with using MCp*dT/dt=Q to calculate He temperature change. Cp is grabbed from Coolprop. Almost done with the new code. Part 2 (3hr real time, 8hr machine time) - cool down of Liquid Helium. Feed the solution data from part 1 as initial data. Use the code developed over the term without any major modifications, other than adding a new block written for feeding XTPage data as input enthalpy.

data

savreline commented 7 years ago

Modified code has been completed. Code runs slow (1 second time steps) due to very high K for copper. Ideas going forward

Use Euler's method instead of ode45
Use adaptive time steps (small time steps at the beginning, then larger time steps)

savreline commented 7 years ago

Got data from 2016 shutdown for the thermocouple in the vault. Lowest temperature is 240K, so data should be reliable. Will simulate overnight.

data1 data2

savreline commented 7 years ago

Run over the weekend - first 2 hours matched up

dec12run

savreline commented 7 years ago

After copper's thermal conductivity was fixed - new temperature distributions in cryopanel is as follows. HX_AU is limiting cool down of copper parts.

walls