Hi Exa team,
I have a case where I inject solids particles with a side EB, inputs.txt (64.8 KB)
I assign a volflow rate corresponding to a mass flowrate of 150g/hr
bc.sideinlet.biomass.volflow = 9.018759018759019e-08 #150g/hr and rho=462kg/m3, hr to seconds
bc.sideinlet.biomass.volfrac = 0.1
But as I check the monitor of the total mass inside the domain, especially the first 0.5 s where particles are not being elutriated yet, the injected solids mass flowrate is much larger than the prescribed value. monitor_lagrangian_tot-mass.csv (56.8 KB)
This must use a different geometry file than the previous thread. What I have doesn’t run.
You need to strip out all of the unnecessary settings for your setup – no species, energy or density advection and all the chemistry. Make sure it runs and reproduces the issue you observe, then provide all of the files (inputs, csg geometry, and size distribution).
I was able to identify a bug in how excess solids volume is managed from step-to-step, and there is a pending merge request with the proposed fix.
MFIX-Exa computes the number of particles to add each time step by sampling the distribution and summing over all bins. The conversion from likely observations (real numbers) to actual particles (integers) can result in not adding enough particles (underflow) or adding too many (overflow). The remainder is tracked and used to adjust the solids volumetric flowrate up or down on the next time step. In your case, the long tail of the distribution causes a substantial overshoot which requires two or more timesteps to pass before more particles should be added. This exposed a bug in the logic where the remainder was zeroed at the start of the subsequent step following an overshoot, therefore the remaining overflow was not accounted for in the next time-step leading to too many particles being added.
The figure below shows the change in solids volume with time. The lightest line is approximately the instantaneous change in solids volume (every five steps) and the darkest line uses a sampling interval 16 times longer showing that the inflow is approaching the targeted flow rate ~9.0e-8 m3/sec.
