Page 593 - Fundamentals of Water Treatment Unit Processes : Physical, Chemical, and Biological

548 Fundamentals of Water Treatment Unit Processes: Physical, Chemical, and Biological

reject all particles above a certain size. These membranes
are given an ‘‘absolute’’ rejection rating based on the largest TABLE 17.4
pore size. Representative Pore Sizes and MWCO’s for Various
Asymmetric membranes (for NF=RO) are usually given Membranes
a ‘‘nominal’’ rejection rating; the pore size of the ‘‘selective’’
Effective Pore
layer governs the shape of the distribution curve, with narrow Type Diameter (mm) Material a
a
curves for both anisotropic and isotropic membranes. Associ-
Hollow fiber 0.2 Polypropylene
ated with the distribution curve for a given membrane are
(shell feed)
membrane ratings, that is, absolute cutoff and nominal cutoff,
Tubular 0.2–0.8 Alumina
also illustrated in Figure 17.11. The term ‘‘absolute cutoff’’
means that all particles of a given size and larger will be MWCO (Da) a
‘‘rejected’’ by the membrane. The ‘‘nominal cutoff’’ rating Hollow fiber, tubular 6,000–13,000 Acrylonitrile, glass
means that only a portion of the particles (albeit about Tubular, spiral wound 8,000 Polyamide
98%–99% usually) is rejected. Membrane manufacturers gen- 5,000 Polysulfone
erally state nominal rejections based on experimental data from 18,000 PVDF
ASTM tests (ASTM Method D4194-85, ASTM, 1987), which Hollow fiber 1,000–300,000 Polysulfone
use a standard solution, e.g., a salt, an organic compound, or Tubular, spiral wound 300 Da–0.1 mm PTFE (teflon),
polysulfone, cellulosic,
particles, to test for the fraction of the test substance rejected. As
thin-film composites
an example, if a standard fructose solution was tested, with
Tubular 0.6 mm Alumina
MW ¼ 180 daltons, with an observed test rejection of 98%,
Spiral wound, flat 2,000–500,000 Thin-film composite,
such a result be the basis for comparing performances of differ-
(DDS) polysulfone
ent membranes. An alternative designation for the nominal
Spiral wound 10,000–100,000 Polysulfone
rejection rating is molecular weight cutoff (MWCO) of solute
Hollow fiber 100,000 Cellulosic
molecules, which is applicable to NF and RO membranes. Spiral wound 1,000 Da–0.1 mm Cellulosic, polysulfones,
Cheryan (1986, p. 57) shows a distribution for a PS VF, polypropylene
UF membrane similar to Figure 17.11 with low range about Tubular 100,000 PDF
10 Å and with the high about 150 Å with mean about Hollow fiber 100,000 Polysulfone
60 Å. As a matter of interest, he determined that the pore Tubular 0.02–0.05 mm Zirconia
11 2
density was about 4 10 pores=cm of membrane surface Hollow fiber (bore 600–800 Polysulfone
area. He calculated the flux density of water through a feed)
similar membrane by the equation of Poiseuille, that is, Hollow fiber (shell 400–600 Polysulfone
2
j (cm=s) ¼ ed Dp=32dm (Rouse, 1946, p. 158), in which feed)
p
e ¼ porosity, d p ¼ mean pore diameter, Dp ¼ pressure differen- Tubular 100,000 PVDF
tial, and m ¼ dynamic viscosity of water. For N ¼ 3 10 9 Source: Adapted from Wiesner, M. R., An overview of DP membrane
2
pores=cm (from a photomicrograph) and d p ¼ 175 Å, the processes, in: Association of Environmental Engineering
2
2
product, N pd =4 ¼ 7.216 10 3 cm pore openings=cm
p Professors, San Antonio, TX, p. 14, June 7, 1993.
membrane surface 0.0072 fraction of membrane surface is a A dalton is a unit of mass equal to 1=12 the mass of a carbon-12 atom,
occupied by pores ¼ e. The membrane skin thickness, that is, about the same as a hydrogen atom. The atomic radius of carbon is
d ¼ 0.2 mm and m (208C) ¼ 10 2 g=cm=s. His calculations given as 0.77 Å (see glossary, Dalton); therefore, to infer that a
2
(cgs units) gave j ¼ 3.45 10 3 cm=s ¼ 124 L=m =h. For dalton is about 0.1–1 Å is probably a reasonable inference as to the size
2
comparison, experiment gave 80 L=m =h. The Poiseuille associated with a dalton. For a more definitive reference, Silberberg
m, while
model is for straight circular cylinders; by contrast, the pores (1996, p. 51) gives the nominal size of an atom as about 10 10 6
14
of a membrane are quite complex. The calculation demon- a nucleus size is about 10 m. Also, for reference, 1 mm ¼ 10 m and
1Å ¼ 10 9 m.
strates, however, that general notion of the Poiseuille model
is likely to be applicable. For the NF=UF membranes, ‘‘pores’’
as such may not exist and so the water permeates by a pressure
17.1.8 APPLICATIONS
gradient, but apparently through the molecular structure of
the membrane. Initially, in the 1960s, membranes were considered for desalt-
ing of seawater. Since the 1990s, the scope of membrane
filtration has been expanded to include removals of organisms
17.1.7.6 Variations in Manufacturer’s Products (cysts, oocysts, bacteria, and viruses), organic compounds,
Table 17.4 lists variations in type, pore size, and materials for and ions (e.g., selected ones, such as nitrates, boron, etc.,
MF and UF membrane elements representative of those manu- or all that are in the feed water). The purposes have included:
factured in 1993. The membranes listed are subject to change drinking water treatment, industrial process water, ultrapure
and the intent is to provide an indication of the variation that industrial water for electronics, reuse of wastewaters, and
exists. As seen, the membrane types include hollow fiber (shell separation of solids in wastewater treatment by immersed
feed and bore feed), spiral wound, tubular, and flat. membranes.