For fluid particle interations, specially sedimentation problems, when I am using the direct forcing immersed boundary method, the calculation of the torque is not exactly accurate. I have read in papers that for IBM, the effective radius is slightly smaller than the physical radius and hence the placement of the lagrangian nodes shoould be slightly lesser (by about 0.5dx) than the physical radius. What is the basis for this and can someone please refer to e some solid literature?