(601e) Online Detection of Compromised Hollow Fiber Modules Using Membrane Integrity Sensor

One of the key issues with operating
membrane treatment plants is ensuring that the membranes maintain their
integrity throughout the process. Most methods of detecting defects (Pressure
Decay Test, Bubble Tests) can only be conducted offline. The membrane integrity
sensor is an online instrument for detecting adverse changes in membrane
integrity by accurately monitoring the permeate stream. This reduces the cost
of operating UF membrane systems, improves reliability and allows for higher
efficiency in the plant.

A novel approach for characterizing the
particulate load and fouling propensity of water has been developed in the Temasek Professor programme at
NTU. This process has been patented and initially designed as an ?Integrity
Sensor' (IS)[1].
The current arrangement (Figure 1) for the IS uses two membrane discs (M1 and
M2) connected in series with pressure transducers measuring TMP1 (P₂
? P₁) and TMP2 (P₃ ? P₂). Sample is pumped in crossflow across M1 and a steady permeate stream is pulled
through M1 and M2. For a ?clean' sample stream the ratio TMP1/TMP2 remains
constant, whereas a solids-laden stream causes TMP1/TMP2 to steadily rise,
signifying a loss of integrity down stream.

A tweak in the design of the original
integrity sensor has increased the versatility and the potential lifespan of
the sensor.

Figure 1 Membrane Based Sensor ? Old Design (left)
and New Design (right)

There have been two major changes to the
design. The first is replacing the second membrane with a needle valve which
mimics the resistance of the second membrane. The second is the installation of
a bypass line on the sample line. The function of the second membrane is to
provide a reference resistance for the fouled first membrane. A needle valve
provides the required resistance with some advantages. One of the advantages is
that a needle valve is much less prone to fouling than a membrane due to its
much larger orifice. This increases the time required before the first membrane
must be replaced. Without a second membrane, it is also much easier to design a
backwash regime to keep the first membrane clean. Most importantly, a needle
valve allows the operator to easily change the ?reference? resistance. The
utility of this feature will be explained in the next section. The bypass line
allows the operator to change the driving pressure in for the sensor, up to a
maximum of the line pressure. One way the sensor loses sensitivity is when the
flux through the membrane drops to such a low value that only tiny amounts of foulants are deposited onto the surface of the membrane. By
starting with a low pressure, and cranking up the pressure when the flux drops,
the first membrane does not need to be replaced as often.

Trial
runs of the sensor have been conducted at the Bedok NEWater Factory. Figure 2 shows a successful test of the
integrity sensor in the plant. Using feed from the plant, it was possible to
detect a hollow fiber UF module that had 0.07% of its
fibers broken. This is evidenced by an increase in
the sensor reading, π.

Figure
2 Detection of Compromised hollow fiber modules

In the full paper, we will discuss other results
obtained from the plant, including long term tests (>1 week), the
correlation of the sensor reading to SDI as well as the detecting breakages in
hollow fiber modules. The implications of using the sensor for improving the
reliability of the entire plant will be explored. References

J. Phattaranawik, A.G. Fane, F.S. Wong, Detection apparatus
and method utilizing membranes and ratio of transmembrane
pressures, PCT/SG2007/000130, WO/2007/129994, Filing date 10 May 2007