You are on page 1of 9

!?

)i
DESALINATION
ELSEVIER Desalination 118 (1998) 13-21

Standardization of RO membrane performance


M. Safar*, M. Jafar, M. Abdel-Jawad, S. Bou-Hamad
Water Desalination Department, Kuwait Institute for Scientific Research, PO Box 24885, Safat 13109, Kuwait email." msafar@kisr.edu,kw

Received 18 June 1998

Abstract

The quality and quantity of reverse osmosis (RO) permeate may vary, depending on the applied pressure, temperature, conversion factor and the feed concentration. Actual performances are not steady under various conditions, and thus operating data are generated variably. An effective evaluation of the performance of the RO system is to compare permeate flow rate and salt rejection data under standard conditions. At the Doha desalination plant (DRP) in Kuwait, an on-line standardization software program, based on the ASTM method, has been developed to evaluate the performance data of the two RO lines - - spiral wound and hollow f'me fiber. This paper investigates the performance of the two RO lines using the program developed and comparing the results with the design data as a reference. For this purpose operating data, generated over two years by DRP, were utilized.
Keywords:

Membranes; Reverse osmosis; Plant performance; Mathematical modelling; Salt rejection; Pressure, Temperature

1. Introduction Successful long-term performance of the RO system depends on proper operation and maintenance of the system. Record-keeping and data normalization are required in order to know the actual plant performance. An RO membrane system is designed on the basis of a defined set of data such as the permeate flow, feedwater *Corresponding author.

composition and temperature. In reality, the plant operation has to be flexible to respond to changing needs or changing conditions. The performance of an RO system is influenced by the feed water composition, feed pressure, temperature and recovery. Performance data normalization involves referring the plant operation back to a standard or reference set of operation conditions. Thus at any time, the actual operating data from the plant are

Presented at the Conference on Membranes in Drinking and Industrial Water Production, Amsterdam, September21-24, 1998, International Water Services Association, European Desalination Society and American Water Works Association 0011-9164/98/$ - See front matter 1998 ElsevierScience B.V. All rights reserved. Pll S0011-9164(98)00070-8

14

M. Safar et al. / Desalination 118 (1998) 13-21

used to calculate what the operation of the plant would be under the reference conditions. Normalization with reference to the designed system performance is useful to verify that the plant gives the specified performance, while normalization with reference to initial system performance is useful to show up any performance change between day one and the actual date [ 1]. Three major plant output parameters are monitored: the permeate flow rate, the quality and the brine side pressure loss. Each of these might vary under normal variable operating conditions. The calculation of their normalized counterparts forms the main technique in monitoring a plant operation. The actual equations and methods used will vary slightly from membrane to membrane.

Table 1 Standard test conditions Conditions Feed temperature(Ts), C Spiral wound Hollowfine (train 1) fiber (train 2) 25 25

Feed conductivity(Cfs), 32,800NAC1 35,000NAC1 mg/L Feed pressure (Pfs), kPa Recovery(Y), % 5520 kPa 35 6,895 kPa 35

3. Normalized permeate flow, QPS


Water flux through an g o membrane varies with temperature due to two different mechanisms: 1. Water become less viscous as its temperature increases and tends to pass through the membrane pores more easily. 2. The membrane structure changes and increases the permeability [4]. The normalized permeate flow, Q.ps, is:
Qps = Qpa x TCFs x P C F s / (TCFa x P C F a )

2. Standardization methods
A system of analysis of plant performance has to be developed to differentiate between changes in performance that might be expected due to changing plant conditions and age, and changes that indicate malfunction or deterioration of the membrane system. To achieve this, a system of performance data "normalization" is usually used. The American Standard for Testing Materials (ASTM) provides a standardization technique for reverse osmosis data. The mathematical model consists of two main parameters: normalized permeate flow and normalized salt passage. This method can be used for both spiral-wound and hollow fine fiber membrane configurations [2]. The design values or the conditions of the initial performance as given in the start-up operation have to be chosen as a reference condition for normalization. In this paper, the design conditions of the membrane manufacturer were used as standard conditions, and they are shown in Table 1 [3]. The measured plant performance at operation conditions can be transferred to standard conditions by the equations below.

where T C F is the temperature correction factor. TCF is dependent on the type of the device (spiral, hollow fiber) and on the membrane type (cellulose acetate polyamide composite). The membrane manufacturers suggested many equations to represent the temperature correction factor. Fig. 1. shows a plotting comparison between temperature correction factor calculated by ASTM and Fihntec methods. The two equations are as follows: ASTM equation:
T C F = 1.03"* (T-25)

(1)

Filmtec equation:

M. Safar et al. / Desalination 118 (1998) 13-21


2.50

15

2.00

Ill

ASTM = 1.50 [] Filmtech

1.00

0.50

0.00 0

I
10

I
20

I
30

f
40 *C

I
50 60

Temperature

Fig. 1. Plotting a comparison of the temperature correction factor by the ASTM and Filmtec methods.
T C F = e x p {U{1/298-1/(273+T)]}

(2)

Cfb

Vb
where U = 2640 T~25 and U = 3480 T_25 In this paper Eq. 2 was selected to represent the TCF. xJb ~p

= C f l n [1/(1-Y)]/Y = 0.002654 Cfb (T+ 273.15)/(1000 -Cfb/1000) [2] = [(0.0117 Cfb)-34)] (T + 320)/ (14.32 345) [1] = 0.01 rcfb

4. Normalized salt passage, % SP


The salt passage is defined as follows: % salt passage, SPa = Cpa/Cfa x 100 and the normalized salt passage is given by:
% SPs = % Spa x ( P C F a x CJbs x Cfa)/

The normalized plant performance would usually be worked out and recorded on a daily basis using these records. In this manner, it is easy to detect deviations from expected plant behavior that may signal problems of the plant on time.

5. Experimental set-up
The Doha Research Plant (DRP) is an experimental research center that deals with all aspects of desalination. Work is in progress to analyze the performance data of two types of single-stage membranes in order to establish membrane characteristics, rank membranes and identify factors that can influence membrane performance and service life. For the purpose of data evaluation, a special software was developed

(PCFs Cfba Cfs)

and that leads to % salt rejection =100-% SPs. Other parameters used in the previous equation can be calculated as follows:
PCF Cf Cp = P f - APJb/2-p- nfb + np = 7 x FCON, [5] = .55 x P C O N [5]

16

M. Safar et al. / Desalination 118 (1998) 13-21

Pan,e

8eachwea

Feed
: :

na~ pro~ct

1,Td08/gll WO-O05C

Pmm~. Flew Ten~. Pressure eoad. SOl

PBEO PROOUCT 34ASm~/h 12.22m)/11


24"C 58 bar

Fluid SystemsTFCL Spiral Wound 2822SS

t.Obar f~n~an 7~cm 1.3e

Fig. 2. Typical flow diagram of the spiral-wound RO] plant.

lrtn~

Beachwel

Oate Project

: :

15/06/98 WD.O05C

Patm.

Film RROOUCT 34.35m3/1~ 12.21ml/h 24"C )0.4 bar 1.0bar 5,1mS/era607p.~Jcm 1.34

Temp. Pressure Co~l. SDI

D u P o n t Hollow F i b e r 6880"1" B -10 - T w i n

Fig. 3. Typical flow diagram of the hollow fiber RO 2 plant.

M. Safar et al. / Desalination 118 (1998) 13-21

17

to effectively evaluate system performance under standard conditions. To implement the proposed plan, research facilities were designed and installed at DRP. The plant is comprised of two different seawater RO module systems capable of desalting Gulf seawater through single stage. Each line is designed to operate independently to produce 300m3/d using raw seawater from a common beachwell system or from a conventional surface seawater pretreatment system. The two RO lines have been operated for 3 years till May 1998. RO Train 1 is using a spiral-wound membrane (Fluid system 2822 SS) whereas RO Train 2 is using the hollow fine fiber twin membrane (Dupont 6880T-B-10-twin). Figs. 2 and 3 show a typical flow diagram of the spiral-wound and hollow fine fiber (twin) RO systems installed at DRP.
6. Results and diseussion

A comprehensive operating data base generated over a period of 2 years by DRP-Kuwait, has been utilized to investigate the normalized performance data of the two RO lines and to compare the results with the design conditions of the two membrane configuration systems, as reference. Table 2 Feed-brine osmotic pressure, ~fb using the ASTM method Conductivity, mg/L Temperature, C 5 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000 13.12 16.48 19.88 23.31 26.78 30.28 33.82 37.40 41.02 10 13.39 16.82 20.29 23.79 27.33 30.91 34.52 38.18 41.87 15 13.66 17.16 20.70 24.27 27.88 31.53 35.22 38.95 42.72 20 13.93 17.50 21.11 24.75 28.44 32.16 35.92 39.72 43.56

It can be seen from Fig. 1 that the plottings of the TCF of ASTM and Filmtec overlap between the temperature range 20-40C. The difference between the two methods is less than 5.2% for the temperature range between 20-50 C, whereas for the range between 5 and 20, the deviation between the plottings increases to almost 22%. Luckily, the actual operation of the RO system in the Gulf region is implemented, most of the time, between 20 and 35C. Therefore, it is important to use the manufacturer's recommended TCF, especially in the low temperature range, i.e., below 20C. As the operation temperature of the work described in the paper is carried out at temperature between 20 and 40C (beachwell seawater temperature is almost steady at 27C), the ASTM correcting factor was applied. The feed-brine osmotic pressure was also calculated using different equations as mentioned above. Tables 2 and 3 show the osmotic pressure of the feed-brine flow between 20,000 and 60,000mg/l and a temperature range of 5-45C for the ASTM and Filmtec methods, respectively. Table 4 shows details of the calculated osmotic pressure using the two mentioned equations. It is clear from Table 4 that the difference between the

25 14.20 17.84 21.52 25.23 28.99 32.78 36.62 40.49 44.41

30 14.47 18.18 21.93 25.72 29.54 33.41 37.32 41.27 45.26

35 14.74 18.52 22.34 26.20 30.10 34.03 38.02 42.04 46.10

40 15.01 18.86 22.75 26.68 30.65 34.66 38.71 42.81 46.95

45 15.28 19.20 23.16 27.16 31.20 35.28 39.41 43.58 47.80

18

M. Safar et al. / Desalination 118 (1998) 13-21

Table 3 Feed-brine osmotic pressure, r~fb using the Filmtech method Conductivity, mg/L Temperature, C 5 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000 13.16 17.01 20.85 24.70 28.55 32.40 36.25 40.10 43.94 10 13.36 17.27 21.17 25.08 28.99 32.90 36.80 40.71 44.62 15 13.56 17.53 21.50 25.46 29.43 33.40 37.36 41.33 45.30 20 13.76 17.79 21.82 25.84 29.87 33.89 37.92 41.95 45.97 25 13.97 18.05 22.14 26.22 30.31 34.39 38.48 42.56 46.65 30 14.17 18.31 22.46 26.60 30.75 34.89 39.04 43.18 47.32 35 14.37 18.57 22.78 26.98 31.19 35.39 39.59 43.80 48.00 40 14.57 18.84 23.10 27.36 31.62 35.89 40.15 44.41 48.68 45 14.78 19.10 23.42 27.74 32.06 36.39 40.71 45.03 49.35

Table 4 Percentage difference between the nfb calculated by ASTM and Filmtech Conductivity, mg/L Temperature, C 5 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000 0.31 3.20 4.92 5.98 6.62 6.99 7.16 7.20 7.12 10 0.20 2.66 4.38 5.43 6.07 6.44 6.61 6.64 6.57 15 0.70 2.15 3.86 4.90 5.54 5.91 6.08 6.11 6.04 20 1.18 1.66 3.36 4.40 5.03 5.40 5.57 5.60 5.53 25 1.64 1.19 2.88 3.91 4.55 4.91 5.08 5.11 5.04 30 2.08 0.73 2.41 3.45 4.08 4.44 4.61 4.64 4.57 35 2.50 0.30 1.97 3.00 3.62 3.98 4.1 4.18 4.11 40 2.92 0.13 1.54 2.56 3.19 3.54 3.71 3.75 3.68 45 3.31 0.53 1.13 2.15 2.77 3.12 3.29 3.32 3.25

50

_ ~ -~---"Cond.- 60 g/L
45

4O

_--- ~ - - ~ - - ~ - - ' ~ n d "

50 B/L

__ ----,x C o ~ . - 40 Z/L

30

~ 0 2O
.._____ .-.---~__~-----

. , ~ . ~ . . A . _ _ _ _ _ _ ~

Cond.- 30 ICL

-------*

~_-4.~------4.

----------*

Cond.= 20 8/L

10

15

20

25

30

35

40

45

50

T e m p e r a t u r e , C

Fig. 4. Osmotic pressure (rrfb) calculated using the ASTM method at different temperatures and conductivity.

M. Safar et al. / Desalination 118 (1998) 13-21


3O 25 Actual

19

"41-- Normnliz~

20

15

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

2000

4000

6000

8000

10000
R u n n i n g

12000 Hours

14000

16000

18000

20000

22000

Fig. 5. Spiral-wound actual, normalized and designed permeate flow rate vs. running hours.

! 00.00

99.50

99.00
I 0

u r.~

98.50

-*-^c~l
- m - Normalized 98.00 ~Dcsigncd

9?.50 0 2000 4000 ~ 8~ 10000 12000 ! 4000 16000 18000 20000 22000

Runningilours

Fig. 6. Spiral-wound actual, normalized and designed salt rejection vs. running hours.

20
25.00

M. Safar et al. / Desalination 118 (1998) 13-21

Actual

~1111
20.00 -

--m-- Normalized

~ 15.00

i 10,O0
[h

5.00

o.oo
0

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

2000

4000

6000

8000
Running Hours

10000

12000

14(X}O

16000

18000

Fig. 7. HFF actual, normalized and designed permeate flow rate vs. running hours.
I00.00

99.50

99.00 98.50

2"
o

98.00
97.50

97.00 96.50 96.00 0

uc~Ign,o

', 2000

', 4000

', 6000

', 8000

', 10000

', 12000

', 14000

', 16000 18000

Running Hours Fig. 8. HFF actual, normalized and designed salt rejection vs. running hours.

M. Safar et al. / Desalination 118 (1998) 13-21

21

results of the two equations vary between 0.31% and 7.2%. Hence, the variation between the methods can be considered acceptable. In this paper the ASTM method was used to represent the osmotic pressure of the feed-brine flow. Fig 3 represents the osmotic pressure of the feed-brine flow at different concentrations and different temperatures. Operational parameters for product flow rate and salt rejection vs. time for RO1, spiral-wound membrane, are shown in Figs. 5 and 6, respectively, whereas those for RO2, the hollow fine-fiber twin, are shown in Figs. 7 and 8, respectively. For over 20,000h of operation the normalized permeate flow was maintained according to the design figures, as can be seen from Figs. 5 and 7 for SW and HFF, respectively. Hence, the normalized flow is always higher than the actual flow due to the effect of the pressure correction factor. With regard to the salt rejection, RO1 is still in operation according to the design value with slight drop of less than 1% (98.75% normalized value vs. 99.6% design value), whereas for RO2, the salt rejection suffered a drop of 2.65% (from 99.65% to 97.0% in 2 years of operation. It can be concluded that the pressure correction factor has a profound effect in normalizing performance data of RO plants using beachwell seawater intake, as the temperature of the water well is almost constant throughout the year.

~ps

Qpa SP

--

SW T
TCF y

---

Normalized permeate flow, m3/h Permeate flow, m3/h Salt passage, % Spiral-wound Temperature, C Temperature correction factor Recovery ratio

Greek
"iT,

Osmotic pressure, bar

Subscripts

a b f p s

------

Actual condition Brine Feed Permeate Standard condition

Acknowledgments
The authors wish to thank the Ministry of Electricity and Water (MEW) for the support of this project (No. KISR WD005) and the Kuwait Foundation for the Advancement of Sciences (KFAS) for their support.

References
[1] J. Huisman,FILMTECMembraneTechnicalManual, Dow Europe Separation Systems, Germany, 1993. [2] ASTM, The Annual Book of ASTM Standard, Designation: D4516-85, 1993. [3] FluidSystemsTechnical Manual. [4] W.T.Hanbury, T. Hodgkiess and R. Morris, Desalination Technology,Porthan, UK, 1992. [5] V.R.G.Walton, Desalination, 72 (1989) 275.

7. Symbols
C HFF p
PCF AP

----

Concentration, mg/L as NaC1 Hollow fine fiber (twin) Pressure, bar Pressure correction factor Device pressure drop, bar

You might also like