Membrane fouling monitoring by 3ω sensing

Abstract

Membrane fouling significantly reduces membrane permeability, leading to higher operational expenses. In situ monitoring of membrane fouling can potentially be used to reduce operation cost by optimizing operational parameters and cleaning conditions. In this study, a platinum wire with a diameter of 20 µm was attached to the surface of a ceramic ultrafiltration membrane, and by measuring the voltage across the wire while applying an AC current, the amplitude of the third harmonic wave, the so-called 3ω signal, was obtained. Results showed increasing 3ω signals during formation of fouling layers, which correlates directly to the hydraulic resistance of the formed fouling layer in semi-dead end filtration of polymeric core shell particles and crossflow filtration of diluted milk. This is explained by the insulating effect of the fouling layers which reduces heat convection by crossflow and the different thermal conductivity in the fouling layer compared with the feed. After membrane cleaning, the permeability and the magnitude of the 3ω signal were partly restored, showing that the 3ω method can be used to monitor the effect of cleaning. The frequency of the AC current was varied so it was possible to measure the heat conductivity in the fouling layer (high frequency) and heat convection due to cross-flow (low frequency). This may potentially be used to get information of the type of fouling (heat conductivity) and thickness of the fouling layer (AC frequency where heat conductivity becomes dominating).

Figure 1

Microscopy image of platinum wire on ceramic membrane consisting of a selective ZrO₂ layer with a SiC support, captured with Dino-Lite microscope (a) and the membrane filtration cell with platinum wire (Ø = 20 µm) integrated on the membrane (b). The Fourier transform of the measured voltage shows amplitudes at 1ω and 3ω; i.e., the U_3ω is determined from the amplitude at a frequency of 3 × ω (c). Ù_3ω signals measured at 1 Hz AC at varying currents on a platinum wire on a membrane in air and in water with varying stagnant, crossflow and permeation conditions (d). Ù_3ω signals were determined at 75 mA AC at different AC frequencies and varying crossflow velocities (e). (f) shows an illustration of a cross-section of the membrane with the thermal waves from the heated platinum wire on the ceramic membrane, and how the heat is released to the surrounding feed suspension, consisting of a laminar boundary layer and a crossflow beyond this, and membrane. Ù_3ω are averages with standard deviations of three replicates.

Figure 2

Ù_3ω signals (average values with standard deviations of three replicates) at varying frequencies obtained by AC of 75 mA through platinum on a clean membrane with crossflow of water and without permeation, a membrane with one layer of acrylic (11.6 g m⁻²) w/ and w/o crossflow, a membrane with two layers of acrylic (29.1 g m⁻²) w/ and w/o crossflow, a membrane with three layers of acrylic (50.4 g m⁻²) w/ crossflow and after removal of acrylic layers by acetone cleaning (a). The theoretical thermal penetration depths of water and acrylic polymer are plotted as a function of AC frequency, resembling a likely range of penetration depth in the fouling layer. Ù_3ω signals are plotted against estimated thicknesses of acrylic varnish on the membranes, showing an increasing signal with layer thickness for the applied AC frequencies (b). The illustration (c) shows how heat is released to the fouling layer and feed suspension from the wire at varying AC frequencies.

Figure 3

Development in flux and Ù_3ω measured at 75 mA AC with 1 Hz frequency over time during semi-dead-end filtration of PS-PAA particles at 1 bar (a) and Ù_3ω plotted against total hydraulic resistance to filtration (b).

Figure 4

Development in flux (top) hydraulic resistance (middle) and Ù_3ω (bottom) over time during filtration of DI water (left), dilute milk (middle) and DI water after cleaning with NaOH (right).

Figure 5

Ù_3ω plotted against hydraulic resistance to filtration from filtration of DI water, filtration of milk, and filtration of DI water after membrane cleaning. Plots of DI water filtrations are average values with standard deviations of the entire filtration.