E L S E V I E R Desalination 172 (2005) 207-214

DESALINATION

www.elsevier.com/locate/desal

Forecasting the economic costs of desalination technology

M o h a m m e d H.I . D o r e Climate Change Laboratory, Department of Economics, Brock University, St Catharines, ON, Canada L2S 3A1

Tel. +1 (905) 688-5550, ext 3578; Fax: +1 (905) 688-6388; email: dore@brocku.ca

Received 27 May 2003; accepted 15 July 2004

Abstract

Govemment policy, in the form of grants and contracts for desalination technology, has had a major impact on steadily declining costs of desalination. The process, reverse osmosis (RO), exhibits economies of scale, which increases its feasibility as a water treatment technology for large populations. Ultrafiltration, an RO pre-treatment, also shows economies of scale. The real economic costs of desalination technology can be forecast using an ARIMA model. If these costs fall below those of conventional water treatment processes, RO and ultrafiltration become competitive with conventional water treatment technology. Our ARIMA forecasts are validated by using independent plant level cost data.

Keywords: Federal funding; R&D; Reverse osmosis; ARIMA, Economies of scale

1. Introduction

The objective of this paper is to forecast the real economic costs of desalination technology, to determine whether the costs for this technology are increasing or decreasing, and if there is evidence of economies of scale. I f costs are decreasing, then the new technology could be feasible for all water treatment plants, particularly for improving the quality of fresh water for drink- ing and industrial use, and treating industrial water prior to discharge or reuse. This paper outlines U.S. Government policy towards desali- nation R&D from the 1950s to the present. This is followed by an ARIMA model to forecast the costs and consider the point at which this new

technology is likely to be competitive with con- ventional water treatment methods. Our forecasts are indeed consistent with information on costs at individual plants.

2. Spread of desalination technology

There are over 11,000 desalination plants in 120 countries around the world, with a combined capacity of 13.25 Mm3/d [1]. Saudi Arabia has the most capacity with 5,006,194 m3/d; the U.S. is next with 2,799,000 m3/d, and Canada has 35,629 m3/d [1]. According to the May 2003 Newsletter of the European Desalination Society, global desalination capacity at the end of 2001

0011-9164/05/$- See front matter © 2005 Elsevier B.V. All rights reserved

doi: 10.1016/j.desal.2004.07.036

208 M.H.I. Dore / Desalination t 72 (2005) 207-214

was 24 Mm3/d [2]. Generally, desalination capital and operating costs have been decreasing, but the increase in energy costs during the 1970s also increased production costs for a short period. The decrease in costs is partly attributable to grants for R&D in the US. While desalination costs have been decreasing, the cost of obtaining and treat- ing water from conventional sources has been increasing. This is due to the fact that in many countries water quality standards have become more stringent, requiring increased levels of treatment. Also, conventionally treated water costs more because of an increased demand for water, resulting in the development of more expensive conventional sources, since the less expensive sources have already been used [3].

The feasibility of desalination technology has much to do with government policy, which took the form of grants and contracts for research and development. The US's interest in desalination is long standing. Droughts in the western US began in some areas in the late 1940s and continue in some of the same areas today, although changes in precipitation patterns have also added new areas of drought. In addition, population growth has concentrated in areas where normal water sources are scarce. This includes California and the entire Southwest, south Texas, Florida and the southeast coast of the US.

In the 1950s, the very limited desalination market was dominated by European manu- facturers, which held 70-80% of the market. In 1952, the US Federal Government passed the Saline Water Act and began funding desalination research and development. By the mid 1960s, the US had built 45% of the desalination plants in operation. The European market share had fallen to 50%. In 1964 the Water Resources Research Act was introduced in the US, which provided funding for desalination R&D. During the mid- to late 1960s, much research was done on develop- ing membranes and distillation technology. The technology developed during this period was made freely available worldwide through work-

shops and published reports. This easy access to the technology contributed to the decrease in costs of desalination.

Throughout the 1960s and 1970s the US remained the leader in desalination technology. Federal support peaked in 1967 at over $100 million (1985 dollars), and steadily decreased until 1973 when funding was all but eliminated. The Saline Water Conversion Act of 1971 pro- vided about $70 million/y in the early 1970s for research grants and contracts. In 1973 the oil embargo increased distillation costs and also increased the need for energy efficiency. As a result of the dramatic increases in the price ofoil, by 1974 desalination research had been reduced to $7 million. This resulted in large reductions in ongoing R&D. However, by 1974, reverse osmosis (RO) had been commercialized, reducing the need for federal support.

In 1976 and 1977, the western US experienced a drought that increased the interest in desali- nation technology once again. This led to the Water Research and Conversion Act of 1977, with desalination research once again focusing on RO. The studies authorized by the Act of 1977 for selecting five locations for desalting demon- strations involved some 26 locations across the country. Five were selected, but funding for construction and operation was not authorized by a federal program. However, some were con- tracted out to private companies or local agency funding.

Between 1952 and 1982, federal funding ave- raged $30 million/y (1985 dollars). Competition from European and Japanese companies increased causing sales by US companies to suffer. In the late 1970s, federal funding was partly reintro- duced, but was eventually phased out by the 1980s. However, the work did not stop com- pletely since the Bureau of Reclamation carried on with money from federal patent rights and with appropriations for their normal research efforts until the new act and funding in 1996.

The Yuma membrane desalter was constructed

M.H.I. Dore / Desalination 172 (2005)207-214 209

and included a research test train, which has been used for research and O&M training by both federal agencies as well as private industry. Throughout the years, the Bureau also partici- pated in international programs in the Middle East using funds supplied from the US State Department and US Department of Commerce funding. From 1953 to 1982, the federal govern- ment spent over $1 billion (1999 dollars) on desalination research [4]. It is generally accepted that this government involvement was responsible for the development of RO.

The desalination market in the US consists mostly of sales of small RO, electrodialysis (ED) and vapour compression (VC) equipment to the commercial and military sectors and to RO plants in coastal communities [5]. The trend is towards fewer large companies, with some small firms having been acquired by chemical corporations. The US industry has annual international sales of $200 million to $250 million, with $5-10 million spent on R&D. This does not compare favourably with the $30 million spent annually during the 30-year, federally funded desalination program. Individual firms conduct R&D programs as currently there is no industry-wide research effort [4].

In the 1990s there was a renewed interest in desalination technology, and thus the Water Desalination Act (1996), also called the Simon Act [6], was enacted. Its objective was the development of more cost-effective and efficient technologies. The federal government would share up to 50% of the total cost of the research or study activity. Financial assistance was in the range of $5,000 to $125,000, with the average grant amounting to $60,000. Beginning in 1997, program funding of $5 million/y for 6 years was allocated for research. Also, $25 million was allocated over 6 years for demonstration and development. In 2000, 15 financial assistance agreements were awarded, and by 2002, it was estimated that six to ten awards were made.

3. Desalination processes

Desalination processes can be divided into two categories: (a) thermal methods, which involve heating water to its boiling point to produce water vapour; and (b) membrane processes, which employ a membrane to move either water or salt to induce two zones of differing concentrations to produce fresh water [4]. Desalination facilities use one of five basic technologies to extract potable water from sea and brackish water. The five technologies include RO, distillation, ED, ion-exchange and freeze desalination.

Recent advancements in membrane tech- nology have allowed the cost of purifying water to drop substantially while at the same time increasing the quality of the water. In the past, RO plant operators had difficulty keeping the membrane surface clean, particularly when treat- ing seawater, surface water or wastewater [7]. Often the highly contaminated water would clog the membranes and reduce the flow capacity of the plant. As a result, higher operating costs due to cleaning chemicals, down-time and increased labour costs to clean the membranes made desali- nation a very costly method of water purification.

To prevent the RO membranes from clogging, an ultrafiltration (UF) pretreatment membrane was developed, which consists of polysulfone hollow-fibre membranes that are placed asym- metrically with the feed stream. The advantage of the polysulfone membrane is that it is chlorine tolerant. Therefore, as biological growths build up on the membrane, a backwash of water com- bined with chlorine can quickly sterilize the membrane. In addition the UF membrane has a barrier surface that is capable of removing waterborne pathogens such as Cryptosporidium, Giardia and viruses. Since the UF pretreatment apparatus can be easily cleaned and prevents waterborne pathogens, the water pressure neces- sary to force the water through the RO membrane can be substantially reduced.

210 M.H.1. Dore /Desalination 172 (2005) 207-214

4. Determinants of costs of desalination

The three factors that have the largest effect on the cost of desalination per unit of fresh water produced are the feedwater salinity level, energy costs and plant size, which show economies of scale [8]. An increase in the salt content of the feedwater increases the operating costs as desali- nation takes longer and/or uses more equipment. The cost of desalting seawater is three to four times the cost of desalting brackish water, with RO being the least expensive process for this application. By 1999 in some areas of the US, the costs ofdesalting brackish water became less than transferring large amounts of conventionally treated water by long-distance pipeline [9]. The energy required for desalination can represent 50-75% of operating costs, with RO having the lowest energy demand. The distillation processes benefit most from economies o f scale, while for RO such economies of scale lead to a fall in unit costs at a lower rate. However, RO has lower unit water costs due to lower energy demands, making it the most economical of all the desalination methods [10]. Furthermore, although RO has higher up-front investment costs, the unit cost of desalted water is determined by membrane life and energy cost [11]. In the southeastern US, desalination technology is becoming more prevalent for brackish water applications.

5. Costs of desalination and economies of scale

During the late 1980s, increasing reports of contaminated ground water combined with poor secondary fresh water sources prompted the US Office of Technology Assessment to conduct a through investigation of desalination technolo- gies. Among the many site-specific criteria examined, the US Office created a summary of real desalination unit costs.

It was determined that the real unit costs for desalination are tiered. For brackish water, according to one estimate, in 1988 RO unit costs ranged from $0.32 to $0.44 per m 3 [4], which in 2004 have now fallen to a range of $0.22 to $0.28. Our own calculations (see Fig. 1) show that the current costs for brackish water lie in the range of $0.08 and $0.07 per m 3, including capital and operating costs. According to the same source, for seawater, which contains ten times the number of contaminates, RO has unit costs rang- ing from $1.57 to $3.55 per m 3 [4]. However, our own calculations show that RO costs for seawater lie in the range of $0.48 and $0.42 per m 3, includ- ing capital and operating costs (in 1985 constant dollars). Our most recent information of the costs in 2004 dollars is in the range of $0.50 to $0.70 per m 3. The range of course reflects economies of scale due to size of the plant.

In addition to the unit costs being tiered, we see that economies of scale can reduce real unit costs by as much as 55% when using RO to treat seawater. In order to determine if economies of scale are present, we used the cost assumptions of

The RO membrane technique is considered the most promising for brackish and seawater desalination [ 12]. It has several advantages over other desalination technologies including lower energy requirements, fewer problems with cor- rosion, higher recovery rates for seawater and less surface area for the same amount of water production [13]. The ability to produce potable drinking water for significantly less than $0.50/m 3 [14] is by far its greatest asset.

0 . 5

0,3 . . . . . . . . . . .

~o 0.2

0.1 . . . . . . . . . . . . . . . . IlIIIIIF ... . . . | " Sea Water

0 ~ rRT r rnT~ u H = ~ l . r r r ~ ~l ~t u , , I , =qTT r r r rTT~T rnT~T rTn~TTCFr rm~q~T~ r r r r r r q l ~ ,-r r r n r r r rm

0 1 O0 200 300 400 500

Capacity in thousands of cubic meters

Fig. l. Reverse osmosis economies of scale.

M.H.I. Dore / Desalination 172 (2005) 207-214 211

O

0.5100

0.4900

0.4700

0.4500

0.4300

0.4100

0.3900

0.3700 tO t o tO tO

¢O ¢.D O5 tO tO tO t o tO t o t o tO t o tO tO tO t o 0~1 tO QO ~ ~1" I ~ 0 O~ ~0 ~ ~ tO CO

Capacity in thousands of cubic meters

Fig. 2. Capital and operating costs of RO with UF pretreatment (in thousands of m3).

UF Economies of Scale Boise River and Rannie Collector

$I,00

$0.80

o $0,60

$o.4o

$o,2o

$o.oo 379 3790 18930

Capacity in cubic metres

Fig. 3. UF economies of scale, a case study. Source: http://www.awwarf.com/exsums/90603.htm.

Truby [15], and graphed economies of scale for brackish water and seawater using RO and UF technologies (Fig. 1) (see Appendix 1 for details).

In the case of both brackish water and sea- water, an increase in treatment capacity results in a per unit decrease in cost. For the purposes of comparison, and in order to maintain the same capacity ranges as in the RO case, the UF costs were generated using an exponential smoothing function (Fig. 2). As in Fig. 1, UF pretreatment with RO decreases costs as capacity increases.

A case study further illustrates the point that economies of scale exist in UF technology. The Boise Water Corporation undertook a pilot pro- ject to study low-pressure membranes for the removal of particles from the Boise River and the Marden Street ranney collector. The project pro- duced total cost data for low-pressure UF [ l 0].

The capital and operating costs for three plant capacities clearly show economies of scale in the Boise River case study (Fig. 3). The lower unit costs reflect the fact river water is being used as the feed.

6. Forecasting desalination costs

We used a statistical forecasting technique to generate data on how these costs are expected to change in the near future. The source of these data is "Using desalination technologies for water treatment, a report published by the US Office of Technology Assessment for water treatment [4], which is given in 1985 constant dollars. In order to forecast the real unit cost of desalination for the year 2000 and beyond, we reproduced this time series on real desalination costs (Fig. 4). The 1970s spike in seawater distillation has been eliminated because this event was an anomaly created by rising energy costs from the oil embargo imposed by the Organization of the Petroleum Exporting Countries (OPEC).

Next, we used an autoregressive integrated moving average model (ARIMA) to forecast the change in real seawater desalination unit costs on our time series. Letting p denote the order of autoregressive terms, d the number of times the data are differenced, and q the order of the mov- ing average terms, we found that the (p, d, q) best model was (0,1,1) (see Appendix 1).

212

24

2O

16

8

M.H.L Dote/Desalination

1940 1945 1950 1955 1960 1965 1970 1975 1980 1985

Year

Fig. 4. Real desalination cost time series in 1985 constant dollars. Source: US Office of Technology [15].

172 (2005) 207-214

constant dollars. When we convert this range to 2002 dollars, it becomes $0.00 to $1.40, where the higher number is the upper 95% confidence limit. Our forecast value of the cost for 2003 was $0.71 in 2002 constant dollars. For 2004, our forecast was between $0.25 and $0.71 per m 3 in 2002 dollars. This compares well with the cost range of $0.50-$0.70 in 2004 dollars quoted above. Our projected costs are, of course, industry-wide average projections, not plant level costs.

24

20

16

"~ 8

4

0 1{

-4 10 1955 1970 1985 2000"-

Year

Fig. 5. Integrated moving average model forecast.

The forecast of the model is shown graphic- ally in Fig. 5. The equation is:

(1-B) ]I, = -.31149899 + v, -.80700050 v,_ 1

(-8.7116) (-8.5764)

In Fig. 5 the dashed line above the fitted line represents the upper 95% confidence interval of the forecast and the dashed line below the fitted line represents the 95% lower confidence interval of the forecast. The fitted line is the predicted value. From Fig. 5 we see that real unit costs of desalination are expected to continue their down- ward trend. For the year 2000 the real unit cost of seawater desalination was expected to lie within $0.00 to $0.93 per m 3. This is, of course, in 1985

7. Comparing forecast values with actual data

We now compare our (industry-wide) fore- casted values with the actual costs of desalination at plants in Ashkelon (Israel), Singapore and a new plant to be built by Dankcr-Ellern Infra- structures at a kibbutz, Palmachim, in Israel, for which the contract was awarded in August 2002. For the Ashkelon plant the 2002 cost per m 3 was quoted as being between $0.50 and $0.53. For Singapore the cost per m 3 is $0.50, and for the new plant at the kibbutz, the tendered price was $0.53, which Tal [17] considered the lowest desalination tendered price ever. The forccast costs are also consistent with costs at Fujairah, Unitcd Arab Emirates, (April 2003). If we assume that these prices are given in current 2002 dollars, then the actual prices all fall within the forecast range given by our ARIMA model. In fact, as these plant level cost data are exogenous, they serve to validate our forecasting model [18].

8. Conclusions

The US Federal Government's policy of in- vesting in R&D from 1952 to 1982 resulted in the unit cost of desalted water to decrease steadily. Our ARIMA model forecast indicates that desali- nation costs will continue to decline. Costs are likely to continue to fall due to the development and improvement of desalination technologies.

M.H.L Dore / Desalination 172 (2005) 207-214 213

UF, a desalination pretreatment technology, also exhibits economies of scale. As these costs fall below those of conventional water treatment technologies, state-of-the-art desalination can be utilized in all water treatment plants. The extent of the use of desalination in the future depends on the decreasing costs, the increasing demand for drinking water, the decreasing viability of alter- natives, and stricter drinking water and discharge standards. The decline in energy costs, especially when using renewable energy, will make mem- brane technology for water production even more attractive in the future.

A case in point is the Tampa Bay desalination plant, which uses gravity as part of its energy requirements, making it perhaps the most effi- cient plant when it reaches full capacity in 2008. It is expected that the cost o f water at Tampa Bay will fall to $0.49 per m 3.

Acknowledgement

The Canadian Water Network (CWN), under the Infrastructure Theme, funded this research project. The author would also like to thank the editor and anonymous referees of the journal for helpful comments. However, the author alone is responsible for any remaining deficiencies.

References

[1] J.E. Mielke, Desalination R&D: The New Federal Program, CRS Report for Congress, 1999.

[2] K. Wangnick, The development in seawater desali- nation, EDS Newsletter, No. 18, May 2003.

[3] Reuters, Thirst for water drives desalination boom, Environmental News Network, March 21,2001.

[4] US Congress, Office of Technology Assessment, Using desalination technologies for water treatment, US Government Printing Office, Washington, 1988.

[5] H. Kunze, Desalination, 139 (2001) 35-41. [6] US Government, Water Desalination Act of 1996,

Public Law 104-298, 42 U.S.C. 10301. [7] California Costal Commission, Seawater Desali-

nation in California, 1993. [8] D.J. Winter Pannell and L. McCann, The economics

of desalination and its potential application in Australia, Sustainability and economics in agricul- ture working paper, Agricultural and Resource Economics, University of Western Australia, 2001, pp. 1-7.

[9] O.K. Buros, The ABCs of Desalting~ International Desalination Association, Massachusetts, 2000.

[ 10] R.V. Azpitarte, A.A. Mesa and C.M. Gomez, Desali- nation, 108 (1996) 43-50.

[11] A.C.F. Ammerlaan, Desalination, 40 (1982) 317- 326.

[12] R. Serniat, International Water Resources Associ- ation, 25 (2000) 54-65.

[13] A.M. Abulnor, M.H. Sorour, F.A. Hammauda and A.M. Abdel Dazen, Desalination, 44 (1983) 189- 198.

[14] K. Klinko and M. Wilf, Desalination, 138 (2001) 299-306.

[15] R. Truby, European Desalination Society Newsletter, May 2001, pp. 2-4.

I16] J.M. Montgomery, J.G. Jacangelo, N.L. Patania and J.M. Lain, Low pressure membrane filtration for particle removal, 1992, http://www.awwarf.corn/ exsums/90603.htm.

[17] D. Tal, Via Marls, Carmel desalination wins 60 MCM desalination tender, EDS Newsletter, 16, September 2002; also in www.globes.co.il, August 1, 2002.

[18] G.G. Pique, Breakthroughs allow seawater desali- nation for less than $0.50/m 3, EDS Newsletter, No 16, September 2002.

214 M.H.1. Dore / Desalination 172 (2005) 207-214

Appendix 1

The following assumptions used by Truby [15] were also used here. These assumptions are:

Interest rate

Amortization period

Membrane replacement cost

Cost of electricity

Membrane life expectancy

Plant life expectancy

Brackish water capital costs/ seawater capital costs

8% 30 years

17.50%

$0.06

7 years

30 years

16.352%

Used ratio to calculate % difference: capital cost, brackish water $130 per m3; capital cost, seawater $795 per m 3.

Some relevant details of the ARIMA model are:

No. of residuals: 45 Standard error: .13405252 Log likelihood: 27.067109 AIC: - 50.134218 SBC: -46.520893

Analysis of variance: D F adj: 43 Sum of squares: .79101485 Residual variance: 0.1797009

Variables in the ARIMA model:

B SEB Tratio Approx. prob.

MA 1 - .80700050 .09409578 - 8.576371 .0000000 Constant - .31149899 .03575684 - 8.7115926 .0000000