Improving membrane filtration performance through time series analysis

Abstract

AbstractFor ultrafiltration, and membrane filtration more generally, the quantitative determination of the modes of fouling remains a subject of great interest. Herein an integral method for determining the modes from a time series of volumetric flux $$J\left(t\right)$$ J t is given and illustrated with previously published filtration data of bergamot juice (Ruby-Figueroa et al (J Membr Sci 524:108-116, 2017)). The integral method of fouling analysis has the potential to become the cornerstone of a robust empirical process. In addition to determining, in a clear-cut manner, the point at which there is a switch from one mode to another, the robust methodology yields characteristic $$J\left(t\right)$$ J t equation for each mode that are an excellent fit to the data. The emphasis is upon the creation of a robust methodology which is best viewed as being a semi-empirical method that is indicative of the modes of fouling. For the example chosen, the initial 4 L/m2 generates some pore blocking after which the main mode of fouling is cake build-up. The variation of overall resistance with time is also informative and analysis of this series was used to check the result for the initial phase of fouling as determined from the time series of volumetric flux. A comparison against the ARIMA (Autoregressive integrated moving average) method, which has never been previously undertaken, is given herein. The integral method of fouling analysis was found to be superior, in part because of the quality of fit to the data and in part because it enables one to establish whether the initial fouling is different in character from the subsequent fouling. Having this information can improve membrane selection and overall membrane filtration performance.