Sustainable seawater desalination: stand-alone small scale windmill and reverse osmosis system
S.G.J. Heijman1,2, E. Rabinovitch1, F. Bos1, N. Olthof1, J.C. van Dijk1,
1 Delft University of Technology, Stevinweg 1, 2628CN Delft, The Netherlands, email@example.com
2 Kiwa Water research, Groningerhaven 7, 3430BB Nieuwegein
Abstract: In coastal areas with a shortage of fresh drinking water, but enough wind, the combination of wind energy and reverse osmosis may provide a sustainable way to produce drinking water. Especially in remote areas the combination is also cost effective. At the moment there are already systems that use windmills in combination with RO installations. But in these systems the wind energy is first transferred to electricity, electricity is stored and used to drive a high pressure pump. Our system uses a direct mechanical drive of the RO-pump by a windmill and storage of clear water covers periods of low wind speed. No electricity is required and the system can be used in remote areas. A commercial windmill, normally used for irrigation purposes, is connected to the RO-piston pump. The installation has a mechanical dry-run protection, low speed and high speed limitation. The energy recovery system is not only to recover the energy from the concentrate flow, but also to control the water recovery (between 10-25%). The permeate production in the first prototype is about 5 m3/day at an average wind speed of 7 m/h. The estimated cost price is 2.3 euro per m3 of produced drinking water. Improvements will be implemented in a second prototype.
Keywords: desalination, reverse osmosis, windmill, sustainable.
The energy consumption for seawater reverse osmosis decreased during the past 20 years. The main developments in the technology were the introduction of the thin film composite membranes and an efficient energy recovery systems. These systems recover the energy from the concentrate which has still a pressure of about 50 bar. Nevertheless the energy costs are still about 50% of the production costs of RO-permeate . A number of publications can be found in literature dealing with the possibility to provide energy for the desalination process with sustainable wind energy. Garcia-Rodriguez  gives an overview of the economic possibilities of the sustainable combination of seawater RO and wind energy. Mathioulakis  explains the problem with the variations in wind speed and the problems with the storage of electrical energy when using wind mills to produce electricity for the RO-system. Lund  shows the feasibility of overcoming periods of low wind speed with the storage of electricity. The goal is to maintain the operational conditions (flux, recovery) at a constant level to have optimal conditions in the RO. It appears that storage of electricity is too expensive to overcome substantial periods of low wind speed. From these results we can conclude that it is worthwhile to investigate if it is better to store the produced water instead of the electricity. The consequence of not storing electricity is that the water production will vary with the wind speed and the RO will not operate at constant flux and/or recovery. Because of the energy losses by the transfer of kinetic energy to electric energy and back to kinetic energy again and because of the costs of storage or back up of electricity, two researchers investigated the direct drive of a high pressure pump with the kinetic wind energy [5,6]. Witte  built a prototype of a vapour compression desalter based on a compression pump directly connected to a windmill. In the other paper case  a prototype of reverse osmosis/windmill system was build for brackish water. The pressure provided by the high pressure pump was depending on the wind speed. The recovery and flux were kept constant by regulating the feed pressure: below 5 bar the RO was shut down. Above 7 bar a pressure release valve opened. Solar cells were used to obtain electricity for the pressure measurements and the process controls. From the literature we conclude that no research was done on a very simple setup for the windmill/RO combination for seawater desalination. We designed a system with a direct drive of the RO-high pressure pump by the mechanical wind power. The pump speed will vary with the wind speed. The permeate production will vary also with the wind speed. A permeate storage is used to overcome periods of low wind speed. The permeate production is about 5 m3/day at an average wind speed of 7 m/s. Only mechanical controls are used to protect the system.
Design of the prototype
A commercial windmill (Tarrago M5015, Figure 1) is used in the first prototype. From the shaft connected to the wings a 90 degrees gear box is transferring the energy to a vertical shaft running down to ground level. At ground level the energy is again transferred with a 90 degrees gear box to a horizontal shaft (see Figure 2). On this horizontal shaft a disk brake is mounted with a manual brake and an automated brake used for the run-dry-protection. A belt-drive is transferring the energy to the shaft to which the high pressure pump (Danfoss APP1.5) and the energy recovery motor (APM1.2) are connected (see Figure 2). An important protection of the pump is the run-dry-protection. The high pressure pump is vulnerable for a run-dry because the pistons are lubricated with water. The run-dry protection is based on the weight of the feed-vessel. If the feed vessel is empty the lack of weight will put the brake on the disk-brake (see Figure 2). A full feed vessel will pull the break of the disk and the mill can turn freely.
Figure 1: Windmill M5015 in a Dutch winter landscape at the test location
The provided mechanical power of this windmill at different wind speed is shown in Figure 3. Also the energy demand of the pump at different wind speed is shown. This energy demand is based on the information of both the pump as well as the windmill manufacturer. The information of the windmill manufacturer about the rotation speed of the windmill is probably not correct because these data are based on a set-up with a conventional irrigation pump. At the start of the project this was the only information available to choose the write windmill/pump combination. The windmill will start at a wind speed of about 7 m/s. The pressure in the RO-system is of course low when starting the high pressure pump and the energy recovery system will not recover energy yet. When the windmill slows down it will stop at a wind speed of about 5 m/s because then the energy recovery system will still be functioning. At a wind speed between 7 and 12 the provided energy will be larger than the consumed energy and the windmill will increase in speed. The speed will increase until either the energy demand of the pump is covered or the vanes are hindered by the turbulent air of the preceding vane. At a wind speed above 12 m/h the windmill turns itself out of the wind in order to avoid damage of the installation. This is a gradual process: At about 17 m/s the windmill will stop completely because the blades will be directed perpendicularly to the wind.
Figure 2: High pressure pump(APP1.5) energy recovery (APM1.2); belt drive and disk brake ready for connection to the windmill.
Feed vessel and permeate vessel are situated at 8 meters. The feed vessel maintains a pressure necessary for the high pressure piston pump. Also it functions as a run-dry protection. The permeate vessel (also at about 10 meters) is storing a small amount of permeate which can be used for flushing the seawater from the feed site of the membrane. At this moment the fresh water flush is done manually after a stop of the mill (either by low wind speed or other reasons). It is the intention to open valve mechanically in the case the windmill stops. It is expected that the fresh water flushes will also prevent membrane fouling. Scaling is prevented by flushing the membranes with permeate without calcium, magnesium or barium. Scaled salts will dissolve during low wind periods. Biofouling is prevented by the osmotic shock effect of the low saline concentration. Flushing is also important to prevent reverse flux by forward osmosis (from permeate to feed).A feed pump is filling the feed vessel continuously. At the moment an electric feed pump is used. In the future this feed pump will be connected to the windmill as well. The capacity of this windmill is enough to drive a extra (low) pressure pump. Another option is to use a separate small windmill near the water source (for instance near a beach well or near a brackish well) to pump the water to the feed vessel.
To protect the system the following mechanical controls were used:
· A run dry protection based on the weight of the feed vessel and the disk brake.
· A high speed protection
· A permeate flush started after a shut down of the windmill
Figure 3: Capacity of the windmill and needed capacity of the high pressure pump with and without energy recovery. (data from windmill and pump manufacturers)
Maintaining a fixed recovery
When dealing with a variable pump speed there is a choice for a constant recovery or a constant cross flow velocity. In our system it was decided to maintain a constant recovery because a constant (low) recovery is very important to prevent scaling and to prevent too high osmotic pressures. At a low recovery the use of an antiscalant is perhaps not necessary, because sparingly soluble salts are only slightly concentrated . A constant recovery is maintained by using the energy recovery system of Danfoss. The piston motor (APM1.2) displaces about 20% less volume compared to the piston pump (APP1.5) simply because the capacity is 20% smaller. The pump and the motor have always the same rotation speed because both are connected to the same shaft (see Figure 4). The advantage of this approach is that the recovery can not exceed a safe value and scaling is prevented. No regulators/controllers are needed to maintain the recovery.
Figure 4: energy recovery system also used for maintaining a low water recovery
In Figure 5 we can see the measured recovery at different pump speed. At low pump speed the recovery is decreasing to 10% probably because the (small) leakage of water around the pistons is relatively high compared to the leakage at higher pump speeds. With a fixed recovery and changing feed flow and flux, the cross flow velocity is also changing. A low cross flow can provide a risk of a high concentration polarisation. Fortunately a low cross flow coincides with the low permeate flux at low pump speed. Concentration polarisation is a function of both the cross flow as well as the flux. The cross flow decreases the boundary layer near the membrane surface and therefor limiting the back diffusion of ions. The flux is driving the ions towards the membrane. A lower flux will give a lower concentration polarisation. In Figure 5 the concentration polarisation is calculated with the projection program of the membrane manufacturer. The concentration polarisation has a low value and is almost independent of the pump speed. With a maximum recovery of 25% and a concentration polarisation of about 1 the concentration of sparingly soluble salts at the membrane wall is at the maximum only 25% higher compared to the seawater. The expectation based on  is that no scaling will occur even without the dosing of anti-scalant. In future experiments with the prototype on seawater we will monitor the possible development of scaling in the last membrane element.
Figure 5: Permeate production, recovery and concentration polarisation as a function of the pump speed.
The windmill was transported and rebuild at the Curacao, an island in the Caribbean. The windmill was installed near the sea water RO plant of Auqualectra, the water and electricity company of the island. The prototype was fed with the same water as the sea water RO-plant (pre-treatment with coagulation and multimedia filtration). The pressure varied between 30 bar and 50 bar. During periods when the windmill stopped, the pressure dropped to zero within a minute. The permeate production varied between 1,5 m3 per day and 6 m3 a day. During this testing period, the wind speed was not very high (between 4.5 and 7.5 m/s). The average wind speed in this area is 7 m/s, so we expect a variation of wind speed between 4.5 and 9.5 m/s and corresponding with permeate productions between 1.5 and 9 m3 a day. Perhaps the location in the neighbourhood of the RO-building and a small hill influenced the average wind speed. Nevertheless Figure 7 proves that with an average wind speed of 7 m/s the production is about 0.2 m3/h corresponding with 5 m3/day.
Figure 6: vafriation in windspeed, permeate production and feed pressure
Figure 7: production of the first prototype as a function of windspeed.
A prototype of a windmill/reverse osmosis system was designed and built. The goal was to use no electricity. Not for energy storage and not for process control. Storage of electricity is very costly and it is less expensive to store the produced fresh water. The consequence of not storing energy is a variable permeate production. The challenge was to design an RO-system which is able to deal with variable process conditions. A simple energy recovery system is used to maintain a constant recovery of about 20%. A variable flux and a variable cross flow velocity are not giving problems in the membrane installation. According to the projection program of the membrane manufacturer the concentration polarisation is hardly influenced and at a low level. Several mechanical controls are introduced to protect the system.
The author(s) would like to express there gratitude to mr. G.N.M. Grootscholten, Hatenboer Water BV and Aqua-for-All for their financial support.
 C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin. (2007)“State-of-the-art of reverse osmosis desalination” Desalination 216 p1-76.
 L. García-Rodríguez, V. Romero-Ternero, C. Gómez-Camacho (2001) “Economic analysis of wind powered desalination”,Desalination 137 p259-265
 E. Mathioulakis, V. Belessiotis, E.Delyannis, (2006) Desalination by using alternative energy: review and state of the art, Desalination 203 p346-365.
 . Lund,P.D. (2006),”Energy storage options for improving wind power quality” , Proceedings “Nordic wind power conference”, 22-23 may 2006, Espoo Finland.
 Tomas Witte, Sönke Siegfriedsen, Magdy El-Allawy, (2003) WindDeSalter Technology, direct use of wind energy for seawater desalination by vapor compression or reverse osmosis, Desalination 156 p275-279. .
 Clark C. K. Liu , Jae-Woo Park, Reef Migita and Gang Qin (2002) C.C.K “ Experiments of a prototype wind-driven reverse osmosis desalination system with feedback control” Desalination 150 p277-287.
 C. A. C. van de Lisdonk, B. M. Rietman, S. G. J. Heijman, G. R. Sterk and J. C. Schippers (2001) ”Prediction of supersaturation and monitoring of scaling in reverse osmosis and nanofiltration membrane systems” Desalination 138 (2001) 259-270