TY - JOUR
T1 - Replicating cohesive and stress-history-dependent behavior of bulk solids
T2 - Feasibility and definiteness in DEM calibration procedure
AU - Mohajeri, M. Javad
AU - van Rhee, Cees
AU - Schott, Dingena L.
PY - 2021
Y1 - 2021
N2 - This paper presents a multi-step DEM calibration procedure for cohesive solid materials, incorporating feasibility in finding a non-empty solution space and definiteness in capturing bulk responses independently of calibration targets. Our procedure follows four steps: (I) feasibility; (II) screening of DEM variables; (III) surrogate modeling-based optimization; and (IV) verification. Both types of input parameter, continuous (e.g. coefficient of static friction) and categorical (e.g. contact module), can be used in our calibration procedure. The cohesive and stress-history-dependent behavior of a moist iron ore sample is replicated using experimental data from four different laboratory tests, such as a ring shear test. This results in a high number of bulk responses (i.e. ≥ 4) as calibration targets in combination with a high number of significant DEM input variables (i.e. > 2) in the calibration procedure. Coefficient of static friction, surface energy, and particle shear modulus are found to be the most significant continuous variables for the simulated processes. The optimal DEM parameter set and its definiteness are verified using 20 different bulk response values. The multi-step optimization framework thus can be used to calibrate material models when both a high number of input variables (i.e. > 2) and a high number of calibration targets (i.e. ≥ 4) are involved.
AB - This paper presents a multi-step DEM calibration procedure for cohesive solid materials, incorporating feasibility in finding a non-empty solution space and definiteness in capturing bulk responses independently of calibration targets. Our procedure follows four steps: (I) feasibility; (II) screening of DEM variables; (III) surrogate modeling-based optimization; and (IV) verification. Both types of input parameter, continuous (e.g. coefficient of static friction) and categorical (e.g. contact module), can be used in our calibration procedure. The cohesive and stress-history-dependent behavior of a moist iron ore sample is replicated using experimental data from four different laboratory tests, such as a ring shear test. This results in a high number of bulk responses (i.e. ≥ 4) as calibration targets in combination with a high number of significant DEM input variables (i.e. > 2) in the calibration procedure. Coefficient of static friction, surface energy, and particle shear modulus are found to be the most significant continuous variables for the simulated processes. The optimal DEM parameter set and its definiteness are verified using 20 different bulk response values. The multi-step optimization framework thus can be used to calibrate material models when both a high number of input variables (i.e. > 2) and a high number of calibration targets (i.e. ≥ 4) are involved.
KW - Consolidation-penetration test
KW - DEM calibration
KW - Elasto-plastic adhesive contact spring
KW - Ring shear test
KW - Surrogate modeling-based optimization
UR - http://www.scopus.com/inward/record.url?scp=85105697940&partnerID=8YFLogxK
U2 - 10.1016/j.apt.2021.02.044
DO - 10.1016/j.apt.2021.02.044
M3 - Article
AN - SCOPUS:85105697940
SN - 0921-8831
VL - 32
SP - 1532
EP - 1548
JO - Advanced Powder Technology
JF - Advanced Powder Technology
IS - 5
ER -