TFM Bug with Guo-Boyce model

MFiX
1

Hi,

I think that there is a bug with the Guo-Boyce TFM model. When taking the original .mfx file from the paper (located at GitHub - guoq52ce/MFiX-TFM-frictional-solids-stress-model-by-GZPXKB: Necessary udfs bulit into MFiX for a new frictional solids stress model that can capture structured bubbling in fluidized beds with vertical vibration or gas flow oscillation) and running it using MFiX-21.2 (as in the paper), the structuring looks very different to if you use MFiX-25.3 (most recent version) and run it using the Guo-Boyce model. Further, if you then use the UDF’s given by Qiang in the Github with MFiX-25.3, you get very similar structuring to what is presented in the PNAS paper (https://www.pnas.org/doi/10.1073/pnas.2108647118). Note - I have not made any other changes to the .mfx files, except remove some keywords that became depreciated during this time.

I think that this suggests that there is something wrong with the implementation of the Guo-Boyce model. I have attached an image showing the change in structure for the different cases: MFiX-21.2 with Qiang’s UDF showing structured bubbling, MFiX-25.3 with Qiang’s UDF showing structured bubbling, and MFiX-25.3 with Guo-Boyce model showing no structured bubbling.

All files are uploaded in the attached .zip - any help would be greatly appreciated!

Oscar

TFM_GuoBoyce_Bug.zip (876.8 KB)

2

This is indeed a bug, thanks for reporting! The problem is a_ps is defined both in a module and in calc_mu_s and they have different values. This will be fixed in the next release. In the meantime, please use the attached or patch the file based on the diff below.

calc_mu_s.f (113.0 KB)

index 8e39451a7..45b4217bb 100644
--- a/model/calc_mu_s.f
+++ b/model/calc_mu_s.f
@@ -4,6 +4,7 @@ MODULE CALC_MU_s_MOD
 
    USE calc_trd_s_mod, only: calc_deriv_vel_solids
    USE compar, only: ijkstart3, ijkend3
+   USE constant, only: a_ps, b_ps
    USE constant, only: C_e, pi, switch, sqrt_pi
    USE constant, only: alpha
    USE constant, only: eps_f_min
@@ -1956,8 +1957,6 @@ CONTAINS
 
       DOUBLE PRECISION :: n1_t
       DOUBLE PRECISION :: shearRate_gamma
-      DOUBLE PRECISION, PARAMETER          :: a_ps = 1D25
-      INTEGER, PARAMETER                   :: b_ps = 10
 
 !---------------------------------------------------------------------
3

Hi Jeff,

Thanks for that. I have now run the code using MFiX-25.3 with the Guo-Boyce model and your attached calc_mu_s.f file and found that it now gives very similar structured bubbling results to the original.

Oscar

image
image942×260 56.3 KB