DESALINATION

ELSEVIER

Desalination 180 (2005) 231-244

www.elsevier.corn/locate/desal

Assessment ofASTM D 4516 for evaluation of reverse osmosis

membrane performance
Yu Zhao a, James S. Taylor b*
~URS Corporation, 7650 West Courmey Campbell Causeway Suite 700, Tampa, FL 33607-1462, USA
hDepartment of Civil and Environmental Engineering, University of Central Florida,
P.O. Box 162450, Orlando, FL 32816-2450, USA
Tel. +1 (407) 823-2785; Fax +1 (407) 823-3315; email: taylor@mail.ucf.edu
Received 24 March 2004; accepted 22 November 2004

Abstract
Reverse osmosis membrane performance was evaluated using ASTM D 4516 and a modified form of the mass
transfer coefficient (MTC) as described in the homogenous solution diffusion model (HSDM) on a common data
set. Standardized salt passage and water production is used to compare both methods. ASTM D 4516 is based on
normalized pressure over time for a given set of data, considers temperature only for water production but not for
salt passage. The HSDM MTC is diffusion based and the HSDM considers variations in flux and recovery for salt
passage. Normalization of the HSDM MTCs for temperature and pressure over time provided a universal assessment
for the water and water quality for a specific diffusion controlled membrane. Assessment of water production was
identical by either method, but assessment of salt passage was different. Salt passage determined by the ASTM
method is dependent on actual net solute driving force, while HSDM determined salt passage is dependent on
MTCs, flux, recovery, temperature, net solute driving force and changes in mass transfer over time. The HSDM
MTC method of membrane evaluation is more versatile for assessment of membrane performance at varying sites
and changing operation.

Keywords: ASTM; Solution diffusion model; Reverse osmosis; Standardization

1. Introduction
Reverse osmosis (RO) and nanofiltration (NF)
are significant technologies for production o f
drinking water. RO and NF performance is
typically evaluated by the change o f water pro*Corresponding author.

ductivity and salt passage over time. Standardization of productivity and salt passage (membrane
performance) is required to compare inter and intra
site membrane performance. The American Standard for Testing Materials ( A S T M ) standard
method, ASTM D 4516 M e t h o d - - Standard Practice for Standardizing Reverse Osmosis Performance Data, provides a procedure to normalize

0011-9164/05/$-- See front matter 2005 Elsevier B.V. All rights reserved
doi: 10.1016/j.desal.2004.11.089

Y Zhao,J.S, Taylor~Desalination180 (2005)231-244

232

permeate flow (Qp) and salt passage (SP) for an

RO system [ 1] that can be used to assess membrane performance. ASTM D 4516 has been described in the literature for assessing long-term
performance of RO membranes [2-4].
ASTM method of assessing membrane performance requires normalization of actual operating conditions using standard operating conditions for pressure, temperature, conversion and
feed concentration. Consequently, the ASTM
specified Qp and SP [1], which would be inaccurate ifa significant concentration of ions passed
the membrane were independent of pressure.
Another error could be produced if there was a
large difference in the actual and standard pressures required by ASTM D-4516 [1]. As mass
transfer of ions through RO membranes are diffusion controlled, only low molecular weight and
neutral solutes would be affected by the ASTM
caveat. However, large pressure differences can
be caused by changing temperature, feed stream
water quality and membrane mass transfer
characteristics during operation.
Other methods have been developed for
assessment of diffusion controlled membrane
performance [5-9]. These methods evaluate mass
transfer coefficients (MTC) for water ( K , also
referred to as specific flux), and solutes (K,) over
time of operation. Using these methods, MTCs
are normalized for pressure, temperature over time
of operation, and provide a more universal method
of performance assessment. This paper presents
assessment of membrane performance on a
common data set using both methods.

PRETREATED
FEED(f) WATER
Qf, Cf, Pf
Fig. 1. NF or RO membraneflow diagram.

2. Theory

2.1. Homogenous solution diffusion model

(HSDM)
The HSDM describes water flux, J , and mass
solute flux J , through diffusion controlled (RO/
NF) membranes as shown in Eq. (1) and Eq. (2)
[10,11]. Eq. (3) is the HSDM and has successfully
described steady-state permeation of water and
solutes through diffusion controlled membranes
[12-14]. A membrane flow diagram showing
influent and effluent flow, concentration and
pressure of the feed, permeate and concentrate
streams is presented in Fig. 1.
The HSDM has been developed by mathematically relating the average feed stream concentration to system recovery using a mass balance
approach [10]. The HSDM was the first model
developed that considered recovery, which allowed
the feed stream concentration to remain constant
and allowed accurate prediction of permeate
concentration for varying flux and recovery [15].
HSDM can be utilized to predict permeate concentrations for any RO or NF membrane application given the feed stream concentration, flux,
recovery and MTCs ( K , K),

Jw=Kwx(AP-A=) =Qp

(1)

L=K.aC=J.C,,

(2)

Cp

K,C
(3)

PERMEATE(p)
Op, Cp, Pf

CONCENTRATE(c)
Qc, Cc, Pc

Y Zhao,J.S. Taylor/Desalination180 (2005)231-244

The osmotic pressure gradient is the difference
between the feed-brine and permeate osmotic
pressure, it can be estimated by the ASTM method
as described previously, or by using total dissolved
solids (TDS) as shown in Eq. (4) [11].

7t = Kms x TDS

(4)

Permeate flow and salt passage increase as

temperature increases [16,17]. K can be compensated for changes in viscosity and membrane
film, e.g. membrane pore radius by normalization
with respect to temperature [7,11]. Eq. (5) and
Eq. (6) were used to normalize temperature. If
Eq. (5) is developed only from the viscosity of
water, 0 wwill be equal to 1.03, which is what is
used in ASTM D 4516 to adjust flux of water for
changing temperature. Such adjustment assumes
the membrane film does not change with temperature. If this equation is developed by non-linear
regression of actual operating data, 0 w will be
normalized for changes in the membrane film as
well as the viscosity of water, and a more accurate
representation of the effect of temperature on
water flux in an RO process will be obtained. K
and K can be normalized for temperature using
non-linear regression of a data set that includes

water and solute mass transfer at varying

temperatures and Eq. (5) and Eq. (6).

Kwr = 0~'-2s) x Kw25

(5)

K, r = 0(y-zs) x Ks2s

(6)

2.2. ASTM standardization methods

ASTM D 4516 utilizes Qp and SP indices as
shown in Eq (7) and Eq. (8). [1]. As shown in
these equations, normalization of permeate flow
is achieved using the standard and actual (a) net
driving forces (b) temperature correction factors
and (c) permeate stream flows. SP is normalized
using standard and actual (a) net driving force,
(b) feed and brine (concentrate) stream concentrations and (c) SPs. Standard values of these
variables are determined by average. Consequently large variations in temperature, conversion (recovery), pressure (flux or fouling) or
feed stream concentration are compensated for by
averaging the associated parameters, which limits
the utility of this approach. This method is
recommended for both spiral-wound and hollow
fine fiber membrane systems.

(7)

QeS=(PY~ AP'2 PP"-na'+nP~)

Pyo

s =(ps

233

TM

Pm -Irma +rcr~

(seo)

(8)

Osmotic pressure is related to temperature and concentration as NaCI (in mg/L) using Eq. (9).
n =0.2654 C (T + 273.15)/(1000- C/1000)

(9)

Y. Zhao, d.S. Taylor~Desalination 180 (2005) 231-244

234

ductivity, pH, turbidity, UV254and chloramine

residual were recorded at the same time. Additional water quality was measured weekly and
included chloride, bromide, sulfate, silica, sodium,
calcium, barium, strontium, total iron and NPDOC.
The data shown in this paper is for X20 membrane.
The membrane was replaced after 2370 h of
cumulative operation.

3. Methods

The simultaneous performances of 4"40"

Filmtec BW30FR, Filmtec BW30LE, Trisep X20
and Osmonics SG elements were evaluated in
large-scale pilot study from July 16, 2002 to April
2, 2003. These four membranes were low-pressure
reverse osmosis membranes. A flow diagram for
the pilot study is shown in Fig. 2. The raw water
was a highly organic and brackish surface water
from St. Johns River at Sanford, Florida, USA.
The surface water was pretreated by ferric sulfate
coagulation using Super Pulsator and Actiflo technologies and dual-media filtration. Sulfuric acid
was used for pH adjustment. Chloramine addition
and 5-micron cartridge filtration preceded membrane filtration. As shown in Fig. 2, a single-stage
membrane system consisting of a single pressure
vessel containing three 4"x40" spiral wound
modules was used for the field study.
The permeate flux and system recovery were
maintained at 22 L.h-l.m-2 (13 gsfd) and 30%.
Feed flow was controlled by a valve on the feed
stream, which was typical ly opened during operation. Feed stream pressure increased as time of
operation increased due to fouling. Pressure, flow
and temperature of the feed, concentrate and permeate streams were recorded twice daily. ConRaw
Water
SCADA DATA
SAHPLE DATA

I
II

Settled
Pretreated

4. Theoretical interpretations

The ASTM method of standardizing permeate

flow is based on Eq. (7) and can be developed
from the equations used to develop the HSDM.
However ASTM method of standardizing salt
passage (SP) as shown in Eq. (8) could not be
developed from the solute mass transfer equations
used to develop the HSDM and was likely developed empirically. The actual and standardized
permeate flows are shown in Eq. ( ! 0) and Eq. (11 ).
Solving these two equations for Qr, produces
Eq. (12) and demonstrates water productivity as
described for RO or NF processes using either
the ASTM or HSDM methods are identical.

Qm=J~.xA=K~zsx(AP~-A1r.)A

Filtered
Pretreated

I
I

(lO)

x O~r-25)

Breaktank Permeate
Concentrate

Feed

ii

, ~

Ferric sulfate

............................

Ammonia

Booster
pump

SuP Transfer
AF Pump

Pressure
Filter

Break
Tank

Transfer
Pump

H Ig h
Cartridge Pressure Membrane
Filter
Pu m p
Vessel

Fig. 2. Integratedmembranesystemshowingsuper pulsator (SUP)and actiflow(AF) pretreatment.

Y.Zhao,J.S. Taylor/ Desalination 180 (2005)231-244

Cp Cp

Qr,, =J.', xA=Kwz, x (Ap., - ATt,)x A

(II)

x 0~;-25)
(

(12)
= (aeo

a=o)

K,

C----~AC:Kx(Ap-Arc)
SP = Cp

- A=., )

C:

235

K, x Cj~
Kw(AP-An)C:

(13)

(14)

, "

The ratio of the permeate concentration to the

bulk concentration is shown in Eq. (13). The
HSDM can be developed from Eq. (13) keeping
the bulk feed concentration constant and incorporating recovery as shown in Eq. (3). If the
permeate concentration is dropped from the concentration gradient, a similar ratio of permeate
concentration to the average bulk concentration
can be developed as shown in Eq. (13). A 1.5%
concentration gradient error was produced in this
work by dropping CP, however the error would
increase as C: or K (nanofiitration) increased.
However, ASTM D 4516 is only meant for standardization of SP and Qp from RO membranes,
which have relatively low K s and Ks.
The left hand side of Eq. (13) is developed by
assuming the permeate to average bulk ratio is
simply the ratio of the solute mass transfer coefficient and water flux. The equation for SP calculation in shown in Eq. (14) and is a simple ratio
of diffusion controlled solute flux divided by
solvent flux. Note, HSDM solute flux is the product of the average concentration gradient and
solute mass transfer coefficient. This is a reasonable approach initially but errors in that recovery and permeate solute concentration is not considered. Consequently a recovery term and solute
MTC (K) appear in the denominator of Eq. (3)
and are absent from Eq. (8). Consequently, the
ratio of permeate and feed stream concentration
or SP using the ASTM approach considers SP to
be the ratio of diffusion controlled solute transport
and flux and did not develop SP fully using mass
balances. The associated degree of error depends
on application and would be significant in some
cases.

The ASTM approach does not rely on any

MTCs and does not provide a means of predicting
the impact of different membranes or operating
conditions on SP. It does provide a means of comparing permeate production and SP for any set of
operating conditions and any environments. The
membrane permeate production and SP would be
normalized for comparison to any other membrane
permeate and production in any environment.
However, the actual SP for any given condition
of temperature, flux, recovery or feed stream concentration in a different environment could not
be predicted unless the standard values specified
in Eqs. (7) and (8) were known. The HSDM considers five different independent variables, is
derived from a fundamental diffusion controlled
mass transfer approach and offers a easy method
of considering the impact of different membranes
or operating conditions on SP. The ASTM approach for evaluation of SP was likely postulated
from rational thinking but not derived.
Temperature is not considered in ASTM D
4516 for SP, but it is considered for Q.. The assumption is that temperature does not a~ect solute
diffusion. However, solute diffusion increases at
a higher rate with respect to temperature than the
viscosity of water decreases with respect to temperature. Consequently, salt rejection will increase
as temperature decreases.
Temperature, normal use (water loading),
organic loading (UV2s4), turbidity loading, and
monochloramine loading are independent variables that could potentially affect productivity and
solute mass transfer. Eq. (15) and Eq. (16) are
direct mass loading expressions that consider the
impact of temperature and feed water quality on
water and solute mass transfer. K and K are

Y Zhao, J.S. Taylor/Desalination 180 (2005) 231-244

236

empirically related to the change in membrane

mass transfer by specific consideration of degradation (by water), oxidation (by NH2CI) and
organic (by NPDOC or UV-254) or particulate
(by turbidity) fouling. 0 wand 0s as well as all of
the water quality coefficients (x) are determined
by non-linear regression. These equations much
like the ASTM SP equation are postulated rational
n

equations for prediction of the impact of several

variables on RO membrane mass transfer. But
unlike the ASTM Eq. (7) and Eq. (8), these
equations when combined with Eq. (3) provide a
means of predicting productivity and SP or permeate concentration at any time, temperature, flux,
recovery, feed concentration or fouling conditions.

K~25 + Xw~e~~ J~t, + Xuvy~ JwCuv2,_,t,

K~

0(~/'-25)

i=1

i=1

(15)

+XturbZJwCturb_iti q- XNH2CIZJwCNH2CI_iI ,
i=!

i=I

K, = 0!r-z~)

r 25+Xw=EJ.t,+XuvEJ.C
v2 _,t,
i=l
i=I

?1

(16)

+x~ZJwC~eo_,t, + XN.,C.ZJ~CN.2C._,t,
i=1

i=1

5. Results and discussion

5.1. Mass loading model development

The standard conditions for normalization are
shown in Table 1 and were set to the average actual
conditions as specified in ASTM D 4516. Since
the X20 membrane was replaced at 2370 h, the
average conditions were determined before and
after 2370 h as shown in Table 1. The feed water
quality, recovery, flow and flux are similar but
the feed pressure and device pressure drop were
different for each period. These standard conditions are averaged for more than 2000 h of operation of each membrane.
The effects of temperature and water quality
on mass transfer were assessed by regressing K
and K over time using Eq. (15) and Eq. (16) and
the water quality data set corresponding to periods
one and two. Initially, all independent variables
shown in Eqs. (15) and (16) were regressed in the
mass loading models to determine their relative

significance. The most insignificant variable was

determined by the highestp or ct values over 0.05,
and was dropped from the model and the regression was repeated. This continued until there were
no insignificant terms remaining. Both 0 wand 0
were significant and are shown in Table 2. The
model coefficients indicate K is more influenced
by temperature than Kc

5.2. Standardized ASTM Qv and HSDM K w

The normalized K and ASTM standardized
Qp vs. membrane run time for the X20 membrane
are shown in Fig. 3. ASTM standardized Qpwas
calculated from Eq. (7) and the normalized K was
calculated from Eqs. (4) and (5). The X20 membrane used in period one is described as the first
membrane. The X20 membrane used in period two
is described as the second membrane. Clearly, both
the ASTM PF and normalized K decreased with
time. The ASTM PF and K changed propor-

Y Zhao, J.S. Taylor/Desalination 180 (2005) 231-244

237

Tablel
Standard conditions for normalization of SP for LR2 system
Standard condition
NaCI feed, mg/L
Pressure drop device, kPa (psi)
Pressure feed, kPa (psi)
Pressure permeate, kPa (psi)
Recovery
Flow feed, m3.d-1 (gpm)
Flux, L.m-2.h-I (gsfd)
Temperature,C

1st membrane
350
303 (44)
1020 (148)
138 (20)
0.3
42.0 (7.7)
23.3 (13.7)
25.0

2nd membrane
350
152 (22)
1344 ( ! 95)
138 (20)
0.3
39.8 (7.3)
22.1 (13.0)
25.0

1 kPa = 0.145 psi; 1 m3-d-~ = 0.183 gpm; 1 L.m-2.h-l = 0.589 gsfd

Table 2
Non-linear regression coefficients of Kw,K mass loading model
System

Kwh,Ks25

Xw-~r3

Kw
K~

0.68 (L.d-l'm-2"kPa-~)1
0.21 (L-h-l-m-2)2

-3.90E - 07
3.60E - 06

XUV

Xturb

XNH,C1

0w, 0s

R2

1.041
1.006

0.78
0.29

~1 L.m-2.h-~ = 0.589 gsfd; 21 L'd-Lm-2 kPa-~ = 0.169gsfd/psi; 3Cumulativewater flux in mS/m2.

tionaily over time; there is no difference in the
normalized HSDM K and ASTM Qp as shown in
Fig. 3, which was stated previously in the theory
section. Therefore, both methods provided equivalent assessments of membrane productivity over
time of operation.
The actual and predicted K s are shown in Fig. 4.
K was predicted using Eq. (15) and as shown in
Fig. 4 is accurately predicted. The model coefficients for Eq. (15) are listed in Table 2. The variation o f K over time was caused by membrane
fouling and deterioration and represents actual not
normalized mass transfer of water.

5.3. A S T M and H S D M standardized SP

The predicted TDS permeate concentration m~d
predicted SP vs. time are shown in Fig. 5. The
HSDM as shown in Eq. (3) with incorporation of
Eq. (13) for K and K was used to predict C and
calculate predicted S~P.As shown in Fig. 5p, the

HSDM accurately predicts permeate TDS and salt

passage. There is significantly more variation in
C than SP, which is due to Fig. 5 scale differences.
ASTM SP is not shown in Fig. 5 because ASTM
predicted SP methodology is not defined in ASTM
D 4516 and is probably not developed.
Membrane replacement can be determined by
SP. Membrane life can be estimated by predicting
SP. Simple linear regression equations relating
standardized SP to time are shown in Fig. 6 for
each period of X20 operation. SP was standardized
in order to compare the ASTM D 4517 and HSDM
SP over time on an equivalent basis. The conditions for standardization are given in Table 1.
Simple trend lines are shown in Fig. 6 for ASTM
and HSDM SP with time for period I and period 2.
The slope oftheASTM trend line was greater than
the slope o f the HSDM trend line in period one,
both slopes were positive in period one. The inference o f increasing SP with time, is that the membrane will have to be replaced at some time due

Y. Zhao, J.S. Taylor~Desalination 180 (2005) 231-244

238

15

r
J

Norm Kw

14
13
12

11
tA

10

92~,

0.6

'E

0.5

al

70

._1
v

1st membrane

0.4

2nd membrane

5 ~<

0.3

0.2

0.1

I
I

0.0

500

1000

. . . .

1500

i i , , , t l

2000

. . . .

2500

1 , 1 1 , 1

3000

. . . .

3500

i , l , i

4000

0
4500

Hours
Fig. 3. A S T M standardized PF and normalized K vs. membrane run time, X 2 0 m e m b r a n e .

1.0
0.9 ,A t=l,_
'=-aA ~.
O~
I~ .

d
II
"7
t~

o_

~ - ~ll I ~
~,m~',ll~il'

i
i

" PredictedKw
A ActualKw

0.8

,~

0.7

p"= :

. =

0.6

E
05
._1
"7

O..
E

0.4

1st membrane

0.3

2nd membrane

"O
_.1

ff

0.2

0.1
0.0

500

1000

1500

2000
2500
Hours

3000

3500

Fig. 4. Actual K w and H S D M predicted K w vs. m e m b r a n e run time for the X20 m e m b r a n e .

4000

4500

Y. Zhao, J.S. Taylor / Desalination 180 (2005) 231-244

20

239

20

;
ActualTDS
A PredictedTDS
-i- ActuaIMembranesp
Replacement

1
t

15

15

Predicted SP

lOu~
or)
{3
}-

ZX

SP

tt~ !
[]

[]

[]

500

1000

1500

2000
2500
Hours

3000

3500

4000

Fig. 5. Actual and HSDM predicted TDS and SP vs. membrane run time, X20 membrane.

0.05

.....................................................................................................................................................

............................................................................................................................................

ti

A ASTM Standardized SP
- Membrane Replacement
----. HSDM Standardized SP
--

0.04
y = 9E-06x + 0.0074
t~
u)

~1'

0.03
r =

t~

o,.
~o 0.02
u')

4E-06x + 0.0118 1'

'

T~J

y = -4E-07x + 0.0149

l ~ ~ 1 1 ~

y=5E-O6x+O.OOQ5 ~

0.01
1st membrane
0.00

1000

2nd membrane

|
|
I

2000
Hours

3000

4000

Fig. 6. ASTM standardized SP and HSDM SP normalized for temperature vs. run time.

0
4500

240

Y. Zhao, J.S. Taylor/Desalination 180 (2005) 231-244

to excess SP. In period two, the slope of the SP

trend line was negative whereas the slope of the
HSMD trend line was positive. A negative slope
indicates the membrane can be used indefinitely
to reject salt. The results of the two methods are
different. Normalization is essential to see what
is happening to SP during time, however the lack
of consideration of temperature in ASTM D 4516
standardization of SP likely caused the negative
trend of SP with time in period two. SP can decrease with time, but it is noted that actual SP did
not decrease with time as shown in Fig. 4.

5.4. Temperature effects on HSDM and ASTM SP

ASTM SP standardization is essentially based
on actual and standard net driving forces and
actual and standard feed and bulk stream salt concentrations. Only osmotic pressure is corrected
for temperature in the ASTM model; however
temperature impacts solute more than water
transport because of the relative changes in ion
diffusivity and the viscosity of water. SP increases
as temperature increases because ion diffusivity
increases more rapidly than the viscosity of water
decreases with increasing temperature. Although
water flux increases as temperature increases, the
diffusion of ions increases more rapidly and SP
increases as temperature increases. AI-Bastaki and
AI-Qahtani also noted that SP increases with
temperature due to an increase in pore size with
temperature [ 16]. As noted previously, feed stream
pressure was varied in order to maintain constant
Flux.
In order to calculate osmotic pressure terms in
Eq. (8), a specific correlation between conductivity and standardized NaCI concentration was
established. Lab water quality data from April
2002 to March 2003 were analyzed in UCF ESI
LAB that included all major cation and anion in
feed water. Salt in NaCI was calculated using
summation ofmolarities of Ba 2+,Ca 2,Mg v, Na +,
SIO2, Br-, CI-, SO~- and Alkalinity in the feed
water. Feed water conductivity and temperature
were recorded twice every day in the field. A

correlation between field measured conductivity

(in ms/m) and NaCI in mg/L was developed and
is shown as Eq. (17).
NaCI = 4 . 1 6 . ~

(17)

where NaC1 = mg/L NaCi; p, = conductivity in

ms/m @ 25C
The effect of temperature on SP is shown in
Table 3 for the HSDM, ASTM D 4516 and the
ROSA program, which is distributed by Dow for
Filmtec membranes. The same membrane and
operating conditions are represented in all cases.
Eq. (9) and Eq. (17) were used to determine the
effect of temperature on ASTM D 4516 SP, which
essentially did not vary with temperature. ROSA
predicted SP was affected the most by temperature
variations and varied from 0.55% at 5C to 1.33%
at 40C. The data in Table 3 clearly shows that
ASTM D 4516 normalization removes any significant impact of temperature on SP.

5.5. SP Comparison at varyingflux and recovery

The predicted ASTM D 4516 SP and HSDM
predicted SP at different K s of 0.5, 0.85 and
1.71 L.m-2.h -1 are shown in Fig. 7. Predicted SP
using the HSDM model was calculated using a
feed stream TDS of 400 mg/L at 25C. Flux and
recovery were varied from 8.5 to 51 L.m-2.h-1 (5Table 3
Comparison of ASTM and HSMD SP at varying temperature
Temperature,
C
5
10
20
25
30
35
40

HSDM-SP,
%
1.50
1.55
1.64
1.69
1.74
i.79
1.84

ASTM-SP, ROSA-SP,
%
%
1.442
0.55
1.443
0.63
1.454
0.83
1.455
0.93
1.456
1.05
1.456
1.18
1.457
1.33

ROSA-SP: Prediction using design software ROSA 4.3

for a BW30-4040 membrane, 3 vessels single stage.

Y. Zhao, J.S. Taylor / Desalination 180 (2005) 231-244

30 gsfd) and from 5 to 90 % respectively. K was

assumed to be constant at 0.59 L.d-t.m-2 kPa -~
(0.1gsfd/psi), SP was calculated using a HSDM
predicted C which was developed from mass
balance and did not neglect the permeate concentration. The required pressure and concentration
terms for ASTM D 4516 were calculated from
mass balances and water fluxes assuming the same
K as for the HSDM. Neglecting the permeate
concentration at higher recovery and lower flux
introduces significant error in ASTM SP standardization.
As shown in Fig. 7, the predicted SP using
ASTM D 4516 does not change for varying flux
and recovery, however the predicted SP using the
HSDM does change with flux and recovery. The
essential point ofASTM D 4516 is illustrated by
the lack of variation of SP with flux and recovery.
This method of standardization removes nearly
all effects of feed concentration, flux and recovery
on SP, and can be used to compare SP in different
environments and for different RO membranes.
ASTM D 4516 can be used accurately for determining standardized SP on RO membranes, which
are membranes with K s less than 0.59 L/d-m2

to

0.25
. . . . .

1,

(0.1 gsfd/psi) and Ks less than 0.51 L.d-~.m-2 kPa-j

(0.3 gsfd), which represent the upper end of water
and solute mass transfer coefficients for RO membranes. IfASTM D 4516 was used to standardize
SP o f a NF membrane the resulting plane would
not be fiat as shown in Fig. 8. Note, the plane
generated by the lowest K and K~ is relative flat,
but more plane curvature is generated in Fig. 8
when the highest and lowest K and K are used to
determine ASTM SP. Hence, the effects of flux
and recovery on SP using a NF membrane would
not be removed using this method and SP standardization would not be achieved. Moreover, if
the predicted SP plane is not fiat, the SP does not
meet the ASTM D 4516 criteria for standardization. Consequently, NF SP using different membranes in different environments could not be
accurately compared. Additionally ASTM D 4516
cannot be easily used if at all used to predict actual
SP. The HDSM is more flexible than ASTM D
4516 and can be used to predict SP for varying
concentration, flux, recovery and temperature for
any membrane once the solute and water MTCs
are known [ 17,18].

"d-l~rn-2.d;l

0.30

241

005

000t

e
~u 45 50

Fig. 7. ASTM standardized SP and HSDM SP vs. recovery and flux.

so " > < ~

... .-:'<.. t

2"

...

_,,-=',~

Plane1:Ks=0.SL.hr-1.m-2;Plane2:Ks=0.85L.hr-1.m-2
Plane3:Ks=1.7L.hr-1.m-2;Kw=0.00103L-d-1-m-2.d-1

Fig. 8. ASTM standardized SP c,nd HSMD SP vs. flux

and recovery for varying K.

242

Y Zhao, J.S. Taylor / Desalination 180 (2005) 231-244

mass transfer approach and develop solute and

water MTCs that can be modified to accommodate changes in feed water quality, foulants
and temperature during operation and easily
predict actual membrane performance.
ASTM D 4516-85 consideration of temperature significantly limits comparison of RO
processes in cold and warm weather environments.
There was no difference in comparison of X20
productivity using ASTM D 4516-85 or
HSDM predicted permeate flow.
The trend of salt passage over time of production using ASTM D 4516 was negative and
the HSDM trend of salt passage over time was
positive. Negative trends of salt passage over
time indicate acceptable membrane performance indefinitely; positive trends infer membrane will be eventually required.

In summary ASTM D 4516 provides a method

of evaluating membrane productivity and salt
passage for any environment and any set of operating conditions by normalization. Production (Q,)
is normalized to temperature, pressure and permeate flow. Salt passage is normalized to pressure,
salt concentration and salt passage. Hence, normalized production and salt passage can be compared for any environment and any set of operating
conditions. ASTM D 45 i 6 is not meant to be used
and can not be easily used to predict actual production for different feed stream concentrations, flux,
recovery or temperature. ASTM D 4516 assessment of water production was verified by derivation from basic mass transfer equations, but the
assessment of salt passage was probably developed
by rational postulation. Prediction of productivity
and salt passage for varying temperature, foulants,
feed concentrations, flux and recovery for different membranes is easily done using the HSDM.

Acknowledgements
6. Conclusions
6.1. General conclusions
ASTM D 4516 and HSDM methods of assessing membrane productivity and solute mass transfer are different. The HSDM method considers
water quality mass solute and water MTCs, fluxes,
recoveries, foulants and temperatures which are
directly transferable to any other water quality
environment. ASTM D 4516 is based on a ratio
of operating results to average operating conditions and does not consider major factors that
influence mass transfer. However the ASTM D
4516 does provide standardized measures of production and salt passage that can be directly used
to assess membrane performance among any environments and operating conditions. ASTM D
4516just can not be easily used to predict actual
performance.

6.2. Specific conclusions

ASTM D 4516 should be modified to reflect a

This pilot study was supported by the St. Johns

River Water Management District and contracted
directly to CH2M Hill, who subcontracted analytical, interpretation and modeling work to UCF.

Symbols
A
C

C~,~
C~s

jP
j
W

Membrane surface area

NaCI equivalent concentration, mg/L
Feed concentration
Actual feed concentration, mg/L
NaCl
- - Actual linear or log mean of feedbrine concentration, mg/L NaCI
- - Standard linear or log mean of feedbrine concentration, mg/L NaC!
Standard feed concentration, mg/L
NaCI
-_ -_ Permeate concentration
Solute flux
_
Water flux
Solute MTC

Y Zhao, J.S. Taylor / Desalination 180 (2005) 231-244

Standardized solute MTC at 25C

Solute MTC at temperature T, C
g T
69 Pa/(mg/L TDS) or 0.01 psi/(mg/
KTDS
n
LTDS)
gf
m Solvent MTC
Standardized solvent MTC at 25C
K,25
Solvent MTC at temperature T, C
KT
Actual feed pressure
Standard feed pressure
Actual permeate pressure
Standard permeate pressure
f
_
Feed stream flow
Permeate stream flow
Actual permeate flow
Standard permeate flow
Recovery = Qp/Qs
Actual salt passage
SP a
SP
- - Standard salt passage
Temperature, C
T
Time
t
TCF a - - Actual temperature correction factor
TCF s - - Standard t e m p e r a t u r e correction
factor
Appropriate regression coefficient
Xi
AC
m Concentration gradient
AP
m
Pressure gradient
Actual average device pressure drop
Standard one half device pressure
/2
drop
Osmotic pressure, kilopascals
ATt
Osmotic pressure gradient = n ~ - n
K temperature correction factor P
0s
K temperature correction factor
0 w
Actual feed-brine osmotic pressure
Standard feed-brine osmotic
pressure
Actual permeate osmotic pressure
pa
Standard permeate osmotic pressure
7t
UV254 mass loading, M3/cmm 2
~ w C UV254/
Combined
chlorine
mass
XJ,,C~,2clt
loading, g/m 2
~,Jwft urbt- - Turbidity mass loading, m3ntu/m 2
~_J,t
- - Water mass loading, m3/m 2

K25

- -

- -

- -

243

R e f e r e n c e s

[1] The Annual Book of ASTM Standard, Designation:

D 4516-85, 1993.
[2] M. Safar, M. Jafar, M. AbdeI-Jawad and S. BouHamad, Standardization of RO membrane performance. Desalination, 118 (1998) 13.
[3] B.A.Q. Darwish, M. Abdel-Jawad and G.S. Aly, On
the standardization of performance data for reverse
osmosis desalination plants. Desalination, 74 (1989)
125.
[4] H.I. AI-Qahtany and N.M.S. AI-Bastaki, Effect of
aging on the performance of RO hollow fiber membranes in a section of an RO plant. Desalination,
101 (1995) 177.
[5] C.R. Reiss, J.S. Taylor and C. Robert, Surface water
treatment using nanofiltation - - pilot testing results
and design considerations. Desalination, i 25 (1999)
97.
[6] L.A. Mulford, J.S. Taylor, D. Nickerson and S.S.
Chen, Comparison of full- and pilot-scale nanofiltration on plant performance. J. AWWA, 91 (1999) 64.
[7] M.F.A. Goosen, S.S. Sablani, S.S. AI-Maskari, R.H.
Al-Belushi and M. Wilf, Effect of feed temperature
on permeate flux and mass transfer coefficient in
spiral-wound reverse osmosis systems. Desalination,
144 (2002) 367.
[81 W.A. Lovins, Correlation and modeling of laboratory and field scale integrated membrane system
productivity and water quality, Ph.D. Dissertation,
University of Central Florida, 2000.
[9] J.S. Taylor and S.K. Hong, SJWD Eastern I-4 Corridor Water Project Phase-IB report, 2002.
[10] J.S. Taylor, Membrane, ch. 11 of Water Quality and
Treatment, A Handbook of Community Water Supplies. Denver, CO, AWWA, 1999.
[11] W.J. Weber, Physicochemical Process for Water
Quality Control. John-Wiley and Sons, New York,
1972.
[12] K.S. Spiegler and O. Kedem, Thermodynamics of
hyperfiltration (RO): Criteria for efficient
membranes. Desalination, 21 (1966) 203.
[13] R.I. Urama and B.J. Marifias, Mechanistic interpretation of solute permeation through a fully
aromatic polyamide reverse osmosis membrane. J.
Membr. Sci., 123 (1997) 267.
[14] D. Van Gauwbergen and J. Baeyens, Modeling
reverse osmosis by irreversible thermodynamics.
Separ. Purif. Technol., 13 (1998) 117.

244

Y Zhao, J.S. Taylor~Desalination 180 (2005) 231-244

[15] J.S. Taylor, L.A. Mulford, S.J. Duranceau and W.M.

Barrett, Cost and performance of a membrane pilot
plant. J. AWWA, 81(11) (1989) 52.
[16] N.M. AI-Bastaki and H.I. AI-Qahtani, Assessment
of thermal effects on the reverse osmosis of salt/
water solutions by using a spiral wound polyamide
membrane. Desalination, 99 (1994) 159.

[17] X.S. Chen Modeling of membrane surface chemistry

and mass transfer. Ph.D. Dissertation, University of
Central Florida, 1999.
[ 18] Y. Zhao, Modeling of membrane solute mass transfer
in NF/RO membrane systems, Ph.D. Dissertation,
University of Central Florida, 2004.