Reliability of membrane test cell measurements

Membrane test cells are extensively used for quality assurance, screening tests and in many research projects. However, many practitioners will agree that test cell results may vary considerably and their accuracy can be inadequate for scale-up to larger membrane units. The reliability of test cells is, besides other factors, influenced by small-scale variations of membrane material properties. The performance of membrane modules, in contrast, depends only on the average properties of a large membrane area. In this paper, we study the variance of test results and the magnitude of scale-up errors caused by the difference between local and average membrane properties. Four flat sheets of 0.5–0.8 m2 of commercial nanofiltration and reverse osmosis membranes are tested for local water and salt permeability, using 18–40 test cell samples per sheet. A statistical analysis of this experimental data indicates the most likely and worst-case errors for scale-up from randomly placed test cell samples to the entire sheet. We demonstrate further, how much the reliability of test results can be increased using either a larger sample size or more than one sample.