Presented at the 11th Aachen Membrane Colloquium, 28–29 March, 2007, Aachen, Germany.

power plant

Karen Gerstandta, K.-V. Peinemanna*, Stein Erik Skilhagenb, Thor Thorsenc,Torleif Holtc

aInstitut für Polymerforschung, GKSS-Forschungszentrum Geesthacht GmbH,Max-Planck-Straße 1, 21502 Geesthacht, Germany

Tel. +49 (4152) 872420; Fax: +49 (4152) 872466; email: klaus-viktor.peinemann@gkss.debStatkraft Development AS, PO Box 200 Lilleaker, N-0216 Oslo, Norway

cSINTEF, N-7465 Trondheim, Norway

Received 14 January 2007; Accepted 9 February 2007

Abstract

The idea to generate power through osmosis between river and ocean water has been known since the 1970s. Thepotential power that can be produced worldwide through osmotic power is estimated to be 1600 TWh/a. But due toinefficient membranes, little effort has been put into research for this type of renewable ocean energy. In 2001,Statkraft, one of the major energy providers in Norway, invited GKSS-Forschungszentrum to develop a suitableosmosis membrane for pressure-retarded osmosis (PRO). Two different types of membranes were optimised: thin-film composites (TFC) and asymmetric cellulose acetate. To make PRO profitable, the power density of the mem-brane was determined to be between 4–6 W/m2. Starting with power production from 0.1 W/m2 for the TFCmembrane, a power density of 3.5 W/m2 with a potential of 5 W/m2 was measured. The starting value for the CA-typewas approximately 0.5 W/m2, and the best measured performance was 1.3 W/m2. However, if it is possible toimprove PRO membranes further, PRO will move to the idea of a profitable application, generating green, emission-free energy.

Keywords: Osmotic power; PRO membrane

1. Introduction

The emission of greenhouse gases such ascarbon dioxide (CO2), causing global warming, is

*Corresponding author.

mainly caused by the combustion of fossil fuels.The shrinking reserves of fossil fuels and theincreasing energy demand due to the develop-ment of Third World countries are only somereasons for the urgent need to search for alter-native emission-free energy sources.

Membrane processes in energy supply for an osmotic

Desalination 224 (2008) 64–70

doi:10.1016/j.desal.2007.02.080

0011-9164/08/$– See front matter © 2008 Published by Elsevier B.V.

K. Gerstandt et al. / Desalination 224 (2008) 64–70 65

Solar and wind power are already well estab-lished as part of our day-to-day life. The oceanas an energy source has not yet been fullydeveloped, being one of the renewable energysources with a very high potential [1]. Oceanenergy sources were accessed during the energycrisis beginning in 1973 regarding energy densityand potential power. Accessed were ocean waves,ocean currents, thermal gradient, tides and sali-nity gradients [2]. Salinity gradient powersystems include reverse electrodialysis and pres-sure-retarded osmosis (PRO).

In theory, a stream flowing at 1 m3/s couldproduce 1 MW of electricity [3]. The globalpotential for PRO is calculated to be on the orderof 1600 TWh/y and for Europe 200 TWh/y [4].Scientists have known this energy source fordecades. Significant research efforts took placebetween 1974 and 1985 [5–9], but due to ineffec-tive membranes, the key part of an osmotic powerplant, not much effort took place to establish thistype of energy.

PRO has the potential to be a huge energyresource. PRO produces no CO2 or other emis-sions that may interfere with the global climateand it is a stable energy form compared to solaror wind. Statkraft, one of the leading energyproviders in Norway, started their research onPRO in 1997. In 2001 Statkraft invited GKSSForschungszentrum to participate as a partner formembrane development.

2. Basic principles: osmosis and pressureretarded osmosis (PRO)

Osmosis is a natural process and for centuriesit has been known that when salt water and freshwater are portioned in two chambers separated bya semi-permeable membrane, made for exampleof a biological membrane, e.g. of hog bladder,fresh water will permeate through the membrane.The driving force is capable of elevating the saltwater level above the fresh water level, whereas

the potential energy is obtained from the staticwater height. The process stops when the hydrau-lic pressure is equal to the potential osmoticpressure (Πosmotic) of the salt water.

Osmotic pressure can be calculated by thevan’t Hoff equation:

Πosmotic = 2.cNaCl R.T (1)

(for dilute solutions), where R is the gas constant,T, the absolute temperature and c the concentra-tion of NaCl. For a 35 g/l NaCl solution, Eq. (1)gives a theoretical osmotic pressure of 29 bars(2.9×106 Pa) at 20°C. This corresponds to a watercolumn of 296m.

2.1. PRO concept

The PRO plant can be described as a desalina-tion plant running backwards. A desalinationplant uses a RO process operating against theosmotic force. However, a PRO plant utilizes thesame osmotic force to produce energy frommixing fresh water and salt water. A simplifiedPRO process is shown in Fig. 1.

Freshwater is fed into the plant and filteredbefore entering the membrane modules contain-ing spiral-wound or hollow fibre membranes. Inthe membrane module, 80–90% of fresh water istransferred by osmosis across the membrane intothe pressurised seawater. The osmotic processincreases the volumetric flow of high pressurewater and is the key energy transfer in the plant.This requires a membrane that has a high waterflow and high salt retention. Typical membraneperformance should be in the range of 4–6 W/m2.

The brackish water from the membranemodule is split into two flows. About 1/3 of thewater goes to the turbine to generate power and2/3 returns to the pressure exchanger to pressurisethe seawater feed. Typical operating pressure is inthe range of 11–15 bars. This is equivalent to awater head of 100–145 m in a hydropower plant,

K. Gerstandt et al. / Desalination 224 (2008) 64–7066

Fig. 1. Osmotic power process [4].

generating about 1MW/m3 freshwater. The freshwater operates in ambient pressure.

Some pre-treatment of the water is necessary.Experience from Norwegian water treatmentplants shows that mechanical filtration down to50 µm in combination with a standard cleaningand maintenance cycle is enough to sustain themembrane performance for 7–10 years. Similarlifetime data are assumed for osmotic powerplants.

2.2. Plant designs

Several plant designs have been developed forPRO power generation. Fig. 2 shows a typicalplant placed at sea level. Fresh water is takenfrom a river close to its outlet. Seawater is fedinto the plant by underground pipes. The brackishwater is let to the natural brackish water zone ofthe estuary thus maintaining the flow of water inthe river. In many respects this PRO process canbe designed as a run-of-river hydropower plant.

Another concept utilises the gravity instead ofthe pressure exchanger to pressurise the incomingseawater. By placing the whole plant 100–130 mbelow sea level, the efficiency of the process canbe increased significantly. The concept comprisesa normal hydropower plant running on waterfrom a river or a lake utilising the extra water

Fig. 2. Sea level PRO power plant [4].

Fig. 3. Sub-sea PRO power plant [4].

K. Gerstandt et al. / Desalination 224 (2008) 64–70 67

head. A membrane plant pumps the water out ofthe sub-sea cavern.

3. PRO membrane theory

The membrane in the PRO process is the keycomponent. The requirements for a good PROmembrane are high water flux, Jw together with alow salt permeability, Js. Fig. 4 illustrates themass transfer of water and salt through an osmo-tic membrane by PRO. The water flux, Jw, isdetermined by the water permeability coefficientA, the effective osmotic pressure difference, Δπand the hydrostatic pressure difference across themembrane, Δp. For an ideal osmotic membraneΔπ is 26 bars for a typical seawater concentrationof 32 g NaCl/l.

The flux of salt is proportional to its permea-bility coefficient, B and Δcsalt. The concentrationprofile of salt through the system is illustrated inthe figure. Only the salt concentration differenceacross the membrane skin will give water trans-port through the membrane.

The PRO performance can be improved bydeveloping membranes with high A-values andlow B-values (Fig. 5), but acceptable performancealso implies that there is a true osmotic transportmechanism in the membrane skin. Further, andalso important, is the inner structure of themembrane, which must not allow significant saltconcentration to build up inside the membrane. Ina technical unit the membrane is placed in amodule. The module must have a design thatreduces the thickness of the boundary layer to aminimum without requiring too much energy forpumping water through the module.

Structure parameter, S (m) — An osmosismembrane consists of one thin non-porous layer,the diffusion skin, and at least one layer of porousmaterial as shown in Fig. 4. The structure of theporous layer is important for an efficient mem-brane in PRO in order to minimise the concen-tration of salt into the porous substructure.

Fig. 4. Water and salt transport through an osmoticmembrane, Jw, Js, s.

Fig. 5. Target values for the A- and B-permeability in agood PRO membrane, fibre or flat [10].

The membrane structure parameter S isdefined as:

S = x @ τ/n (2)

with the porosity, n, thickness of the porouslayer, x, and the tortuosity, τ. The lower thestructure parameter, the better the performance ofthe membrane under PRO conditions. The desiredvalue for the structure parameter is lower than0.0015 m [10].

Calculation of power performance, W (W/m2)— Membrane power performance can be calcu-lated by multiplying the water flux, Jw, and thehydrostatic pressure difference across the mem-brane, Δp.

W= Jw @ Δp (3)

K. Gerstandt et al. / Desalination 224 (2008) 64–7068

Fig. 6. SEM cross section showing an asymmetric CA membrane (a) and the cross section of a TFC membrane (b) [11].

4. Results and discussion

Several approaches to the membrane develop-ment process have been followed that involvedboth asymmetric membranes made of celluloseacetate (CA) (Fig. 6a) and thin-film composite(TFC) membranes (Fig. 6b). Over 50 differentsupport materials for the TFC membrane and arange of asymmetric membrane types made frompolymers other than CA were produced to findthe optimal PRO membrane.

4.1. Cellulose acetate membrane

A cellulose acetate membrane was prepared asfollowing: the casting solution is cast on a glassplate and immersed in ice cold water after solventevaporation. After solidification the membrane isannealed between 80° and 95°C. A typical castingsolution, according to a GKSS patent, consistsout of cellulose diacetate, cellulose triacetate,dioxane, acetone, acetic acid and methanol. Thiscomposition was kept, but due to changing thecasting parameters, both in the lab and in pilotscale, the performance was improved. Castingparameters like casting speed, changes in the tem-perature of the coagulation bath and also thechanges of the support material led to the

improved performance. Starting with a membraneperformance of approximately 0.5 W/m2, thistype of membrane was improved to a perfor-mance of close to 1.3 W/m2.

4.2. TFC composite membranes

The chemistry for manufacturing the mem-brane was basically the same as described byCadotte in 1981 (US Pat. 4,277,344), which is theinterfacial polymerisation of trimesoylchlorideand m-phenylene diamine. The essential differ-ence between a reverse and a direct osmosismembrane is the following: a reverse osmosismembrane has to withstand pressures of 60 or80 bars and requires therefore a compaction resis-tant sub-structure. In RO water and salt are flow-ing in the same direction. In the PRO processdescribed here, pressures are much lower and saltand water are flowing in opposite directions.When a high-performance RO membrane is usedin the PRO process, its performance is usuallypoor due to salt concentration build-up in thesupport layer (internal polarization). Therefore,we made the microporous support layer as thinand open as possible, resulting in a reduction ofthe structure parameter, S, and in an improvementof the PRO membrane (see Section 3).

K. Gerstandt et al. / Desalination 224 (2008) 64–70 69

A further improvement step was achieved withthe improvement of the diffusive layer, the poly-amide, PA. A formic acid post-treatment step hasalready been described by Verissimo et al.[11,12]. Inner-coated TFC hollow-fibre mem-branes showed an increase of water permeabilityup to four times without losing the salt retention.An improvement of water permeability up 18times was achieved with a concentrated acid anda long acid treatment. This is remarkable; on theother hand, it caused a decrease of the salt reten-tion, which is undesirable.

For flat TFC membranes an improvement ofwater permeability of a factor of approximately 4was achieved, without decreasing the salt reten-tion. Similar to the results presented by Verissimoet al., a too long acid treatment results in adecrease of the salt rejection, which woulddecrease the power performance. An explanationfor the increased water permeability after acidtreatment might be due to partly dissolving thepolyamide, the surface layer, and hence thereduction of the membrane thickness respectivelythe membrane density. It could also be explainedby the extraction of un-reacted chemicals and bythe dissolution of polymer particles which werenot formed in a complete film.

4.3. Power production of membrane prototypesMeasurements of CA membranes have proven

that many of these membranes behave as, or closeto, ideal osmotic behaviour (Fig. 7). Some of theprototype membranes exhibit very high waterpermeabilities, but the corresponding salt reten-tion is still too low, which restricts the powerproduction in PRO.

Investigations of commercial CA membranesshowed a verified power of below 1 W/m2; insome cases a power production of 1.6 W/m2 wasreported [13,14]. In the present study an improve-ment from 0.6 W/m2 up to 1.3 W/m2 was achievedfor both the potential and also for the measuredpower, illustrated in Fig. 7. However, the

Fig. 7. Power production from prototype membranes.

achieved improvement is still far from the target,5 W/m2.

4.4. TFC membrane

Starting with membranes designed for RO,with an energy density of less than 0.1 W/m, onlyvery recently an energy density of 3.5 W/m2 hasbeen verified. However, the potential of this typeof membrane was calculated using a membraneperformance simulator to be close to 5 W/m2,which is the target value for the membrane tomake PRO commercial attractive.

The difference of the calculated and themeasured performance was not or only slightlypresent for the CA membranes. For the TFCmembranes this is an undesired but present effect.Intensive studies are currently in progress inorder to find the reason for this problem.

5. Conclusions and future work

PRO is one of the most promising renewableocean energy sources. In order to establish thisform of green energy, the membrane, the heart ofthe process, has to be optimised. The break-evenpoint for the membrane performance is of 5 W/m2

In this study it was presented that the mem-brane performance was optimised for both types

K. Gerstandt et al. / Desalination 224 (2008) 64–7070

of membrane CA and more impressive for theTFC, reaching in theory the goal of 5 W/m2 andwith a measured power performance of 3.5 W/m2.

These achievements show that the day forPRO to become a real application is gettingcloser. However, membrane performance still hasto be improved and the up-scaling of the mem-brane production process might be problematic.

Another aspect that should be taken intoaccount is that getting close to the realisation ofPRO, a new market will be opened. To realise a1 MW power plant approximately 200,000 m2 ofmembranes are required, which opens up a newmarket for membrane producers.

Statkraft’s next step is the planning of con-struction of an osmotic prototype power plant,with 2000 m2 of membrane installed.

6. Symbols

A — Water permeability coefficient, m/sPa

B — Salt permeability coefficient, m/sc — Concentration, mol/m3

Js — Molar salt flux, mol/m2 sJw — Volumetric water flux, m3/m2 s =

m/sR — Gas constant, J/mol KS — Structure parameter, mT — Absolute temperature, KW — Power, W/m2

x — Thickness of the porous structure, m

Δosmotic — Osmotic pressure, bar or PaΔπ — Effective osmotic pressure differ-

ence, Paτ — Tortuosityn — Porosity

References

[1] A.T. Jones and W. Rowley, Global perspective:economic forecast for renewable ccean energytechnology, Mar. Technol. Soc. J., 36 (2003) 85–90.

[2] A.T. Jones and W. Fineley, Recent eevelopment insalinity gradient power, Doherty Lecture, 2001.

[3] G.L. Wick and W.R. Schmitt, Prospects for renew-able energy from sea, Mar. Technol. Soc. J., 11(5–6)(1977) 16–21.

[4] Statkraft, Osmotic Power — A huge renewableenergy source, Information material, 2006.

[5] R.W. Norman, Water salination: a source of energy,Science, 186 (1974) p. 350.

[6] O. Levenspiel and N. de Vevers, The osmotic pump,Science, 183 (1974) p. 157.

[7] S. Loeb, Osmotic power plants, Science, 189 (1975)654–655.

[8] S. Loeb, Production of energy from concentratedbrines by pressure-retarded osmosis, 1. Preliminarytechnical and economic correlations, J. Membr. Sci.,1 (1976) 49–63.

[9] S. Loeb, F. Vanhessen and D. Shahaf, Production ofenergy from concentrated brines by pressure-retardedosmosis, 2. Experimental results and projectedenergy costs, J. Membr. Sci., 3 (1976) 249–269.

[10] Statkraft patent, European Patent Application No.01961437.9.

[11] S. Verissimo, K.-V. Peinemann and J. Bordado,Thin-film composite hollow fibre membranes: Anoptimized manufacturing method, J. Membr. Sci.,264 (2005) 48–55.

[12] Carsten Blicke, personal communication.[13] K.L. Lee, R.W. Baker and H.K. Londsdale, Mem-

branes for power generation by pressure-retardadosmosis, J. Membr. Sci., 8 (1981) 141.

[14] H.H. Jellink and H. Masuda, Osmo-power. Theoryand performance of an osmo-power pilot plant,Ocean Engng., 8(2) (1981) 103.