Multi-criteria Decision Analysis is Useful for Determining Optimal De-salination Design

While desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> processes can generally be applied interchangeably to different situations, the different methods vary in cost effectiveness depending on external conditions. Two categories of desalination methods are included in this study—thermal and membrane. Thermal desalination technologies include multistage flash (MSF) distillation, multiple-effect distillation (MED), and vapor compression (VC). Membrane technologies include reverse osmosis (RO)<!–[if supportFields]>XE “reverse osmosis (RO)” <![endif]–><!–[if supportFields]><![endif]–> and electro-dialysis (ED). For any of these technologies, optimization of plant design is based primarily on the salinity<!–[if supportFields]> XE “salinity” <![endif]–><!–[if supportFields]><![endif]–> of the water (usually categorized into low salinity brackish water and high salinity seawater), quality of water product, and volume of production.  For this study, however, other factors such as environmental impact and political preference were also taken into consideration. Afify (2010) looked at five different location types including three aquifers (brackish water) and two coastal settings (seawater). Each location differed in the type of water usage, ranging from small scale use along the desert fringes of Egypt to coastal tourism and resorts. Afify used Multi-criteria Decision Analysis (MCDA) to determine the best technology for each location. He found that, independent of water usage, ED is preferred for brackish water treatment, though RO would be better for larger desalination volumes, while MED is slightly preferred for seawater followed closely by MSF and RO. For any of the technologies, a higher water usage is preferred as it is more cost effective over small scale plants. —Erin Partlan
Afify, A., 2010. Prioritizing desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> strategies using multi-criteria decision analysis. Desalination.250, 928–935.

Afify used the method of MDCA to evaluate the selected desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> alternatives for a range of location scenarios in Egypt. MDCA uses weighted numerical values for each evaluation criterion so that criteria of varying importance can be used altogether. Furthermore, numerical valuations of qualitative aspects allow them to be included in this method. However, there is no set method for determining the weighting values, and as a result, the interpretation of importance or quality as numerical values can be fairly arbitrary.
 Afify looked at five different water sources—low-usage desert fringe aquifers, the moderate-usage Nile<!–[if supportFields]> XE “Nile” <![endif]–><!–[if supportFields]><![endif]–> aquifer, low-usage coastal aquifer, seawater for tourism, and seawater for coastal city development. The last water source is based on plans to construct new villages and cities along the Red Sea in order to accommodate a growing population (10.7 million by 2025). This scenario incorporates medium to large scale desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> plants into these establishments during initial construction. In comparison, seawater for tourism would utilize small scale desalination systems. For the three aquifers, though they are all brackish water, they vary in salinity<!–[if supportFields]> XE “salinity” <![endif]–><!–[if supportFields]><![endif]–>. The Nile-fed aquifers have the lowest salinity, whereas the coastal aquifers have high salinity bordering on seawater. The desert aquifers are medium salinity oases. However, Afify notes that as aquifer water is a relatively non-renewable resource, and are generally in areas of low development in Egypt, use of aquifer water for desalination should be limited.
Afify assigned values for each alternative (permutations of water source, desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> technology, and plant size) across five categories—investment costs, operation costs, quality of produced water, environmental impacts of brine outflows, and political preference. Costs were measured in Euros, water quality in ppm, and political and environmental factors were rated out of ten. To rank the categories, Afify used percentages of these values to create two weighting scenarios. In both scenarios, the best technology for each situation was the same: large-scale MED for coastal cities, small-scale MED for coastal resorts, small-scale ED for oasis aquifers, medium-scale ED for Nile<!–[if supportFields]>XE “Nile” <![endif]–><!–[if supportFields]><![endif]–>-fed aquifers, and medium-scale ED for coastal aquifers. Afify uses these results to recommend a plan of action for coastal development—one large MSF desalination plant for the city of Suize and five MED plants for all other cities along the Red Sea.

Economics of Desalination

Cost estimation for industrial plants is usually performed with the assistance of aggregated data and estimates by knowledgeable bodies. Younos (2010) discusses the many factors that affect desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> plant cost and evaluates their impact on total cost. In addition, he describes several cost estimation models that are used for desalination plants. Costs are broken into one time construction costs, including both direct and indirect costs, and recurring operation and maintenance costs. The primary factors that determine the magnitude of these costs are feedwater salinity<!–[if supportFields]> XE “salinity” <![endif]–><!–[if supportFields]><![endif]–>, plant capacity and the location of the plant. One of the more costly aspects of plant location is brine removal as coastal plants can use cost effective surface water disposal while inland plants must use more expensive alternatives. Younos concludes based on aggregated desalination plant cost estimates that fixed costs play a large role compared to maintenance costs, and that brackish water treatment only differs from seawater treatment in energy costs.—Erin Partlan
Younos, T., (2010). The Economics of Desalination. Journal of Contemporary Water Research and Education 132, 39–45.

Younos defined the following costs pertinent to the construction and operation of a desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> plant. The direct construction costs of implementing a desalination plant include costs for land, production wells, surface water intake structure, equipment (i.e. for water treatment), buildings and brine disposal. Indirect costs include construction overhead such as labor costs and tools, owner’s costs such as administrative fees, freight and insurance costs, and resources reserved for contingency. The recurring operations and maintenance costs are separated into fixed insurance and amortization costs and all other costs, including energy, equipment replacement and cost of chemicals. For each site, these costs will be based on the quality of the feedwater, the plant capacity, the location of the plant, and any regulation requirements that may exist. For example, low salinity<!–[if supportFields]> XE “salinity” <![endif]–><!–[if supportFields]><![endif]–> feedwater will cost less in terms of energy usage, large capacity plants cost more initially but are more efficient in the long run, and costs associated with water intake, pretreatment and brine disposal will depend on the plant surroundings. Younos points out that brine disposal plays a large role in desalination plant design due to its dependence on site location and regulations. In addition, there are multiple options for brine disposal including surface disposal, disposal with wastewater plant effluent, deep well injection, on-land dispersal such as evaporation ponds, spray irrigation, percolation, and zero liquid discharge. For coastal plants, surface disposal to large bodies of water is common and cost effective. However, this is generally not an option for inland plants. Options for inland plants depend on the characteristics of the plant location and techniques such as deep well injection and zero liquid discharge can be costly.
Younos also describes three cost estimation models that are particularly useful for evaluating desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> projects. WTCost, a model from the Bureau of Reclamation, includes thorough and detailed plant cost estimates for the following desalination technologies: reverse osmosis (RO)<!–[if supportFields]> XE “reverse osmosis (RO)” <![endif]–><!–[if supportFields]><![endif]–>, mechanical vapor compression (MVC), multiple effect distillation (MED), multi-stage distillation (MSF), nanofiltration (NF), and electrodialysis reversal (EDR). The the Desalination Economic Evaluation Program (DEEP)<!–[if supportFields]> XE “Desalination Economic Evaluation Program (DEEP)” <![endif]–><!–[if supportFields]><![endif]–> developed by the International Atomic Energy Agency (IAEA)<!–[if supportFields]>XE “International Atomic Energy Agency (IAEA)” <![endif]–><!–[if supportFields]><![endif]–> is a model for evaluating the effects of various energy sources on a desalination plant, particularly looking at nuclear sources versus other alternative energies. However, this model was not intended for industrial use and is not as detailed in non-technical costs. The third model is the Reverse Osmosis Desalination Cost Planning Model, a product of Water Resources Associates (WRA)<!–[if supportFields]> XE “Water Resources Associates (WRA)” <![endif]–><!–[if supportFields]><![endif]–>, which includes 33 different parameters for desalination plants and is similar to WTCost. While Younos did not perform the cost estimation himself, Sandia National Laboraties aggregated the work of others who have used such methods to estimate the costs of desalination plants. Based on this report, which differentiated estimates by both technology and feedwater quality (brackish vs. seawater), Younos concludes overall that fixed costs are actually a large component, equipment replacement is actually relatively small, and that the only major difference between brackish water and seawater desalination is the cost of energy.

Wind-powered Membrane Desalination Feasible with Minimal Energy Storage

Reverse osmosis (RO) is a desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> process that uses pressure to force high salinity<!–[if supportFields]> XE “salinity” <![endif]–><!–[if supportFields]><![endif]–> water through a semi-permeable membrane. Since the production rate depends on the pressure difference, which in turn depends on the power supplied, the fuel of choice is energy-dense coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–> and oil. However, not only are fossil fuels unsustainable, but polluting as well. Park et al. (2011) studied the possibility of powering an RO plant with renewable wind power. The advantage to using wind power is that there is a steady supply of wind along coastal areas where desalination plants are most commonly placed. However, without energy storage as a buffer between the wind turbine<!–[if supportFields]> XE “turbine” <![endif]–><!–[if supportFields]><![endif]–> and the plant, changes in wind speed and direction translate directly into changes in RO pressure and flow rate. While changes in power supply are expected with any power source, this is particularly troublesome for wind energy as wind flux<!–[if supportFields]> XE “flux” <![endif]–><!–[if supportFields]><![endif]–> typically changes by 12% per second, compared to the 1% change on average of solar flux. Park et al. tested their RO model with brackish feedwater using both a programmable power supply and an actual wind turbine. Issues arose such as a maximum power output of 300 W from the turbine, system shutdown under low wind speeds, salt diffusion across the membrane during shutdown, and a wind speed threshold for feedwater with a high osmotic pressure. However, the authors found that desalination performance under wind conditions of more than 7.0 m/s and turbulence of less than 0.4 was similar to that of steady state conditions, thereby concluding that directly connected wind-powered desalination is feasible with energy buffering to prevent system shutdown. —Erin Partlan
Park, G., Schäfer, A., Richards, B., 2010. Renewable energy powered membrane technology: The effect of wind speed fluctuations on the performance of a wind-powered membrane system for brackish water desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–>. Journal of Membrane Science 30, 34–44.

Park et al. used a test bench model of a reverse osmosis desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> system. They examined variables of average wind speed, oscillations in wind speed, and turbulence. Most situations were created with a programmable power supply and then verified using a wind turbine<!–[if supportFields]> XE “turbine” <![endif]–><!–[if supportFields]><![endif]–> in a wind tunnel. Two brackish water concentrations were used–2750 and 5500 mg/L NaCl–and testing occurred under wind speeds ranging from 3.7–8.7 m/s. Using the power supply, they tested steady state wind speeds (no turbulence or oscillation) and programmed oscillating wind speeds with a turbulence intensity of 0.4 (0.6 being extreme fluctuations and 0.0 being no fluctuation).
They found that under steady state conditions, the optimal power outputs were 120 W for low concentration feedwater and 180 W for high concentration feedwater. The maximum power output for their experimental turbine<!–[if supportFields]> XE “turbine” <![endif]–><!–[if supportFields]><![endif]–> was 300 W as safety mechanisms were activated under high wind speeds. They also found that while all wind speeds produced permeate flows with acceptable salt concentrations using low concentration feedwater, high concentration feedwater required a minimum of 120 W (corresponding to a steady state wind speed of approximately 5.3 m/s). Under oscillating conditions, they found that low wind speeds with low frequency oscillations produced the lowest permeate flows. This was caused by system shutdown due to low membrane pressure as a direct result of low power supply. They found that shutdown occurred at power outputs of less than 40 W for three seconds. Also, due to the low wind speeds, the system had difficulty restarting, thus further reducing productivity under these conditions. Another issue with system shutdown was the diffusion of salt across the membrane, resulting in a raised permeate salt concentration. Due to this diffusion, the maximum shutdown period for high concentration feedwater is three minutes as the permeate will reach unacceptable salt concentrations at this point. To re-achieve permeate salt concentration within two minutes after a shutdown period, 240 W was needed for high concentration feedwater compared with 120 W for low concentration feedwater. In contrast, high frequency oscillations did not permit shutdown even during low wind speeds as the pressure always returned quickly and thus did not differ greatly from steady state conditions.
The third variable tested was turbulence, measured by the amplitude of the oscillating wind speeds. The authors found that wind conditions of more than 7.0 m/s and turbulence of less than 0.4 adequately resembled steady state conditions. At less than 7.0 m/s, the system was able to reach low enough pressures to reach shut-off. In the extreme case, the osmotic pressure of the high concentration feedwater combined with the low power supply of low wind speeds resulted in zero permeate production.
Lastly, the authors used a wind turbine<!–[if supportFields]> XE “turbine” <![endif]–><!–[if supportFields]><![endif]–> inside a wind tunnel to perform verification tests. They found that the wind speed did not always correspond with the power output due to complexities in the system, though the membrane pressure still depended directly on the power output. However, the system performance still compared well with the steady state test results using the power supply. The authors were also able to demonstrate an exponential decay in the membrane pressure after system shutdown, thus providing an explanation for the buffer time during low wind speeds. In addition, the wind tunnel tests displayed a 50% production loss with large wind speed fluctuations. However, the authors note that this type of turbulence is not typical and only observed in extreme conditions. They also note that large power fluctuations should not be significantly detrimental as their test membrane has been used for over 250 hours under extreme turbulence. Overall, the authors conclude that wind-powered membrane desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–> is feasible despite any drawbacks.
We saw that both wind and solar power are feasible options for powering desalination<!–[if supportFields]> XE “desalination” <![endif]–><!–[if supportFields]><![endif]–>, though in different applications. We also saw that technology is constantly changing, both in the manipulation of existing processes to the invention of new ones. However, desalination is not a standalone process. Issues such as the ethical disposal of concentrated brine or the effect of environmental conditions are just some of the concerns that arise in operating a desalination plant.

Heat-absorbing Materials Useful for Increased Solar Still Efficiency

The solar still is the simplest form of solar-powered desalination. It uses the mechanisms of evaporation and condensation—the same processes used other forms of distillation—to purify the water. In a solar still, water is kept in an airtight container. As the water heats up, it evaporates and becomes water vapor. The lid of the still serves as the condenser to transform the purified water vapor back into water and the water slides down the slope of the lid to a collection point. Murugavel et al. (2010) built and tested a solar still with a roof-like glass lid, shallow basin, and insulation. They investigated the effects of various insulating and heat-absorbing materials on the efficiency of the still, since operation of the still depends on the amount of water evaporated, which depends on the amount of heat added to the water, among other things. The materials tested here included rocks, brick, metal, and cloth. The results of their testing showed that a ¾ inch layer of quartzite rock on the bed of the still performed the best. Murugavel et al. also performed theoretical calculations using energy balances and heat transfer equations to determine the theoretical efficiency possible for the chosen parameters. While the quartzite rock performed the best in tests, the actual efficiency was still nowhere near the theoretically possible one.—Erin Partlan
Murugavel, K., Sivakumar, S., Ahamed, J., Chockalingam, K., Srithar, K., 2010. Single basin double slope solar still with minimum basin depth and energy storing materials. Applied Energy, 87, 2, 514–523.
Murugavel et al. built and tested a solar still in Kovilpatti, India. They crafted a basin from mild steel plate, created a glass cover with a north and south slope, and insulated it with glass wool. In testing, a minimal water depth of 0.5 cm was used. Measurements were taken of the influx and outflux of water and of the temperature of the body of water and the water vapor. Also, atmospheric conditions were monitored to ensure that factors were controlled between test days. From incident solar radiation and ambient temperature data, the authors conclude that this is a valid assumption. The materials used to collect extra heat on the basin were ¼ inch quartzite rock, ¾ inch quartzite rock, ¼ inch washed stones, 1½ inch cement concrete pieces, 1¼ inch brick pieces, mild steel trimmings, and a light black cotton cloth. Multiple trials were run with each material, and while the overall productions hovered around 3.5 L/day of water, the ¾ inch quartzite material performed slightly better than the rest at 3.66 L/day of water.

In the theoretical testing, the authors use thermodynamic equations to model the heat influxes and outfluxes of the system. They note that they are novel in their approach as they use a variable term for the transmittance of solar energy through the glass cover, a term usually assumed to be constant. The resulting equations in their modeling are expressions for the instantaneous and overall water production of the solar still. However, when the parameters from the test of the ¾ inch quartzite rock are used, it was found that four-fold increase in the production rates was theoretically possible. While the authors note several areas of discrepancy—the change in water volume and depth over time, a higher proportion of water vapor inside the still, and differences in the absorptivity of the different testing materials—these results imply that the heat-absorbing material used has a minimal impact on improved efficiency, and that instead, focuses should be made on improving the design of the solar still itself to better utilize the incident energy. 

Chlorine By-products from Reverse Osmosis Desalination Not Likely to Pose Health Risks to Humans or Aquatic Environments

Reverse osmosis (RO) is a common method of water filtration, and in particular, desalination. Of commercial desalination using filter membranes, RO makes up 80%. It operates by applying pressure to brackish water on one side of a membrane that is only permeable to water, thus creating purified water on the other side of the membrane. In practice, however, the intake water is chlorinated in order to remove organic matter such as algae that can cause problems with the system. This in turn results in chlorine by-products, some of which can cause chronic problems or even be lethal at sufficient concentrations. In addition, desalinated water is often blended with available freshwater, which may also contain the same organic compounds, thereby making total concentration levels unacceptable. The concern addressed in this study is whether these by-products are released in harmful concentrations either for human consumption or the aquatic environment (Agus and Sedlak 2010). The study focused on a pilot plant in Carlsbad, CA and though the study found that some of the organic compounds produced were prevalent enough to be tasted, none were harmful.––Erin Partlan

Agus, E., Sedlak, D., 2010. Formation and fate of chlorination by-products in reverse osmosis desalination systems. Water Research 44, 1616—1626.

            Eva Agus and David Sedlak looked at the Carlsbad desalination plant and also obtained water samples from the coastal regions of California, Florida, and Singapore. These samples were used as a comparison to the Carlsbad data and were chosen as they are likely locations for the placement of water-processing plants. In the Carlsbad tests, samples were taken at two different times (summer and winter) and with two different chlorination dosages. To model blending of desalinated water with freshwater, samples were taken from the Colorado River, NV and the San Pablo Reservoir, CA, which are rich and poor in bromide respectively, a compound that can pass through RO filters relatively well. For the samples from the various coastal regions, after controlling for pH, one chlorination dosage was used to match the Carlsbad tests. For all samples, measurements were taken at various times over a period of three days.
            The Carlsbad plant samples had less organic compounds than expected, and is thus attributed to the fact that it was designed for higher water purity and also, as a pilot-scale plant, gives less opportunity for the formation of by-products due to a shorter holding time. The organic compounds that were detected were found to vary both seasonally and with chlorine concentration. As a result, the authors recommend that tests be run during different times of the year in order to ensure that none of the water produced is harmful. The study also found that even after blending the desalinated water with the chosen freshwater samples, while the bromide concentrations increased greatly, they remained safe for consumption. It is also noted that the results of this test were conservative since no pre-treatment of the freshwater was performed, a process that would normally be done in a commercial plant. 
For the samples from the coastal locations, none were found to exceed standards for organic compounds after chlorination. However, it was found that the types and proportions of compounds varied by locations, even though the method and dosage of chlorine were held constant. It is likely that the initial variation in organic compounds dictates the type of reactions that occur after chlorination. Therefore, the authors recommended that tests be performed at the site of a desalination plant to ensure that the compounds present in its outflows are not harmful. In particular, they note that some compounds would be tasted at the levels produced and may be a factor in the consumer desirability of the desalinated water.