Mechanical behavior of powders: Tester design, load-response measurement, and constitutive modeling

Load-response of powders under various stress paths and over wide range of stresses are not well understood due to: (1) lack of capable test apparatus, i.e., existing apparatuses are either lack of capacity in testing pressure or limited in the loading boundary conditions, and (2) lack of comprehensive research in constitutive modeling of powders' behavior. This thesis addresses these two issues. In this study, a medium pressure flexible boundary cubical triaxial tester was designed. With this tester, up to 35 MPa air or gas pressure can be applied to the six faces of a 50.8 mm. cube-shaped powder specimen via flexible rubber membranes with virtually unlimited number of stress paths. Usage of flexible rubber membranes as a means of applying pressure minimizes the undesired frictional effect and improves loading boundary conditions. Several types of tests were conducted on different powders. The results were highly consistent and repeatable which proved that the tester designed in this study can measure the 3D load-response of powders. Two elasto-plastic constitutive models, modified Cam-clay (MC) and the DiMaggio and Sandler (D & S) cap models, were identified as the leading constitutive models for predicting the static mechanical behavior of powder. Heckel and Kawakita equations were identified as two of the leading empirical equations for powder compaction. Material parameters were determined. One of the most notable phenomena, the change of material properties with consolidation pressure, is evidenced by the large differences in the parameter values at different pressures. MC and D & S cap models were verified by back-predicting the experiment results. Both were found capable of predicting the powders' load-response but the hardening parameter function in D & S cap model needed improvement. MC and D & S cap models and Heckel and Kawakita equations were compared. When pressure <14 NTa, Kawakita equation predicted the relative density well and Heckel equation gave almost linear predictions and did not follow the experimental curves. While superior in many other ways, the continuum mechanics based MC and D & S cap models performed better than or as well as the empirically formulated Heckel and Kawakita equations at the only task the latter are capable of.