Producing fuel cells bipolar plates and other devices such as microscale heat exchangers for electronics requires both macroscale and microscale forming processes. At the macroscale, typically, mechanical properties of sheet metal are determined by performing tensile tests. In addition, it has long been recognized that bi-axial tension tests, dome tests, and hydroforming or viscous bulge tests provide the basis for improved understanding of the mechanics of sheet metal forming. At the microscale strain gauges are too large for measuring strains in small regions and membrane theory is only valid at the poles of the bulge. Continuum mechanics models are useful but require tedious thickness measurements for multiple work pieces, requiring extensive sample preparation and analysis.
In this paper experimental results from hydroforming tests for 0.2-mm thick annealed ASTM 304 stainless steel sheet in 11 mm, 5 mm, and 1 mm diameter open dies at various pressures were evaluated. The height of the bulge at the pole and strains based upon measurements of 127 micron strain grids were determined. These dies represent the transition from a small macroscale process to a microscale forming process. Two methods were used to estimate material properties: an analytical model and an iterative method which compared experimental strain results with the strains from a finite element model where the Holloman constitutive properties of the sheet were varied. The problems estimating material properties based upon grid strain measurement, membrane theory, and the iterative finite element approaches were investigated and the results were compared. This study indicates that membrane theory will provide adequate predictions for Holloman constructive properties provided the assumptions for membrane theory are not violated. However, using measured microscale grid deformation strains does not produce very good agreement estimates of the Holloman constitutive model when comparing experimental results with FEA strains. It is believed that while the grid strain measurement method used results in strain measurement errors of less than 1.5% of strain, this error is sufficient to result in enough uncertainty to produce results that are inconsistent with other methods.