Page 450 -

MEMBRANE PROCESSES 13.47

have been used for water treatment for more than 30 years, and when properly manufac-
tured, installed, operated, and maintained, the systems are very reliable.
Membrane treatment facilities commonly have multiple parallel process trains.
Where required, additional membrane units can be installed to provide needed re-
dundant capacity when one or more units are out of service for maintenance or mem-
brane cleaning. In many cases, particularly where the membrane system provides ex-
pansion to an existing water supply system, only critical system components are
provided with redundancy, rather than the entire membrane unit. This not only min-
imizes the financial investment, but also allows the membrane system to operate
as continuously as possible. If there is no complete standby membrane unit, the
need to prepare and feed membrane storage solutions is minimized among other
advantages.
If a product water quality problem is detected in a membrane system, the specific
membrane modules within the system having the problem can be identified for re-
medial action. With many membrane systems, the deficient membrane modules can
be isolated and the rest of the system operated until there is a convenient time for
maintenance.

Bench and Pilot Testing
Bench testing is often performed in conjunction with a feasibility study to determine which,
if any, membrane process is applicable to treatment requirements. If a particular process
has been selected, bench testing results can develop basic characterization data on the pro-
cess. Bench testing is typically 1 day to 2 weeks long and uses a simple membrane char-
acterization apparatus consisting of a feed pump; a pressure assembly containing a fiat-
sheet, hollow-fiber, or tubular membrane; and valves to set flow rates, pressures, and
recovery. Typical operating data derived during bench testing include initial membrane
feed pressure to produce a given flux and initial permeate quality (representative of low-
recovery operation) at a given membrane flux.
Compared with bench testing, membrane pilot testing is lengthy, often taking 1 to 4
months of operation. Test duration is generally governed by the amount of time the pro-
posed membrane and pretreatrnent systems must be operated to correctly quantify longer-
term trends in membrane performance, such as flux decline rate and rate of change of
solute transport.
Bench and pilot testing both have limitations that the designer must consider. For ex-
ample, source water may not be representative of water quality over the life of the mem-
brane system. The designer should consider extra allowances, or safety factors, for criti-
cal design parameters.

Designer and Vendor Interface Issues
Many membrane products are available for water treatment applications, and new prod-
ucts continue to enter the marketplace. It is important that the membrane facility designer
contact membrane manufacturers and system suppliers to identify the best candidate mem-
brane products for the application and to identify all critical design issues applicable to
each product. This not only minimizes potential problems with the constructed facility
and allows the best candidate membranes to be considered, but also provides for the de-
signer the benefit (directly or indirectly) of the latest computer design software to mini-
mize design calculation time.