Sensitivity analysis of the pressure-based direct integrity test for membranes used in drinking water treatment

We conducted a sensitivity analysis of the commonly employed pressure-based direct integrity test (DIT), the most sensitive test for defects in low-pressure hollow fiber (LPHF) microfiltration and ultrafiltration systems used in drinking water treatment. Incorporating uncertainty to assess the practice of DIT, we find the resolution in some tests may be insufficient to verify the presence of a barrier to oocysts of Cryptosporidium. Applying distributions and boundaries derived from literature and practice, we solved for the defect size resolution (DSR) using Monte Carlo and Probability Bounds Analysis for five commercial membrane designs. Surface tension was modeled using annual temperature profiles from three rivers. Contact angle measurement error and variability were derived from literature, respectively, as a standard deviation of 5.7 degrees and +/- 9.6 degrees median change due to natural organic matter (NOM) fouling. These measures of contact angle uncertainty and variability were combined in a normal distribution with the discrete values currently applied. Additionally we considered model uncertainty, applying the maximum bubble pressure method, an established method of surface tension measurement in liquids in which the maximum air pressure in a submerged capillary is developed after the contact angle becomes zero prior to bubble formation. Where the DSR exceeds 3 microm the test design is not compliant with applicable drinking water regulations. Implications include uncertain and variable log-removal values (LRV) as determined by DIT due to the possible emergence of defects large enough to allow oocysts to pass without detection by the DIT. Specifically, we found the DSR may exceed 3 microm and may be as large as 8 microm. With the variable contact angle model, all lower bound possibilities are compliant, whereas the upper bound is over 80% noncompliant for three of five commercial designs. Using the Maximum Bubble Pressure Method, the lower bounds in three designs start to exceed 3 microm for between 50 and 100% of the produced water, whereas the upper bounds of the DSR completely exceed 3 microm for four of five commercial designs examined.