Time-Dependent Natural Convection in a Square Cavity

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Time-Dependent Natural Convection in a Square Cavity

eschuldt
FAETool V1.15.2
Matlab 2018b

I have followed the "Natural Convection in a Square Cavity" tutorial and I'm able to produce the same expected output as described.

I now want to repeat this exercise, but for a Time-Dependent solution. Using the same scenario the only things I change are:

Solver Type: Stationary --> Time-Dependent
Time Stepping Scheme --> Backward Euler (As Crank-Nicolson would result in "High-order time stepping schemes not supported with integral constraints")

When I try to solve, it errors and says "Matrix Dimensions Must Agree"

Any insight would be appreciated.
Reply | Threaded
Open this post in threaded view
|

Re: Time-Dependent Natural Convection in a Square Cavity

Precise Simulation
Administrator
This post was updated on .
Thank for reporting the issue. Fluid flow problems without outflows, like your example, require the pressure to be unique (to converge). In order to do this an integral constraint setting the mean pressure to zero is introduced per default. This modifies the system matrix which seems to conflict with the time dependent solver (this will be addressed in a future update).

Alternatively, you can remove the integral constraint (clear "Equation" menu > "Subdomain Integral Constraints..." > p), and instead fix the pressure in one point ("Boundary" menu > "Point Constraints..." > p), this should work for all time stepping schemes.

EDIT: Update 1.15.4 is now available with a fix supporting integral constraints and time-stepping schemes.