Independence check for PIC simulation

Hi,

As we know, for PIC simulation, parcel is used to take the place of real particle to reduce the computational amount. As a result, an empirical parameter particle weight is introduced. So, do we need to check the independence of this parameter for PIC simulation, especially when it is related to the conversion of reactants in reactor simulation? This question is from our observation that for fluidized bed reactor simulation using PIC method, the conversion of gas reactant is reduced proportionally when particle weight is doubled, which deviates from the experimental value.

A further questions is that if the parameter particle weight is set empirically, is it necessary to check the grid independence? We searched a amount of research paper including those using CPFD, two kinds of check are not mentioned. Does it mean that for PIC simulation, particle weight and grid independence check is not needed?

If you can afford it, it is always a good idea to check results sensitivity to parameters and grid spacing. Below are some links to such studies.

Hi, Dr. Dietiker

Thank you for the documents. I read them today. But i still have a question about independence check of PIC simulation. Can you please give me some advice?

For a fluidized bed reactor, as you know, the phase fraction distribution in the reactor is usually unknown, and only the yield/conversion of gas compositions at the outlet can be obtained directly. In this case, how can we get reasonable grid size and particle statistical weight? We noticed that in some papers, the author used the measured solid fraction distribution to validate the rationality of the selection of grid size and particle statistical weight. However, it seems very difficult for our case, due to the difficulty in measuring solid fraction. So, is there another path to make the check?

It is mentioned in the document, named “Sensitivity Analysis of Particle-In-Cell Modeling Parameters in Settling Bed, Bubbling Fluidized Bed and Circulating Fluidized Bed”, that the influence of particle statistical weight on the observed quantity is correlated to the size of the grid, which is consistent with our simulation results. In the simulation, both refining the grid and increasing the value of particle statistical weight can decrease the conversion of gas reactant. Compared to the single independence check of grid, two aspects, particle (parcel independence) and fluid (gird independence), are involved here for PIC simulation and they are not independent. Therefore, what is the rational procedure for checking the independence for PIC simulation? Especially when the phase fraction in the reactor can not be obtained directly.

Thank you!

I am afraid I don’t have a good answer for you. This is a bit of a balancing act and depends on how much time you can spend on such studies. The link below gives some rule of thumb to select the statistical weight:
https://mfix.netl.doe.gov/doc/mfix/22.1/html/reference/pic_tips.html

I am not sure what you mean by “difficulty in measuring solid fraction”, you have access to all the data, so you can get a distribution of solids fraction.

I believe some people keep the same number of parcels per cells when refining/coarsening the mesh, using a representative solids fraction.

Hi, Dr. Dietiker

Thank you for your helpful link. It’s too long since I last used the manual. I am sorry for forgetting there is a guide for PIC simulation in the manual. I think you are right. We have to find a reasonable value of particle statistical weight (PSW) first and then check the grid sensitivity, though different value of PSW may result in different choice of grid size. What i meant by “difficulty in measuring solid fraction” is that it is difficult to measure solid fraction in a experimental fluidized bed. As you know, for a reactor, it is usually hard to obtain its flow filed directly.