Related Pages
updated June 2010
The most common types of nuclear power plants use water for cooling in two ways:
- to convey heat from the reactor core to the steam turbines, and
- to remove and dump surplus heat from this steam circuit. (In any steam/ Rankine cycle plant such as present-day coal and nuclear plants there is a loss of about two thirds of the energy due to the intrinsic limitations of turning heat into mechanical energy.)
The bigger the temperature difference between the internal heat source and the external environment where the surplus heat is dumped, the more efficient is the process in achieving mechanical work - in this case, turning a generator[1]. Hence the desirability of having a high temperature internally and a low temperature in the external environment. This consideration gives rise to desirably siting power plants alongside very cold water.*
* Many power plants, fossil and nuclear, have higher net output in winter than summer due to differences in cooling water temperature.
1. Steam cycle heat transferFor the purpose of heat transfer from the core, the water is circulated continuously in a closed loop steam cycle and hardly any is lost. It is turned to steam by the primary heat source in order to drive the turbine to do work making electricity[2], and it is then condensed and retuned under pressure to the heat source in a closed system.[3] A very small amount of make-up water is required in any such system. The water needs to be clean and fairly pure.[4]
This function is much the same whether the power plant is nuclear, coal-fired, or conventionally gas-fired. Any steam cycle power plant functions in this way. At least 90% of the non-hydro electricity in every country is produced thus.
2. Cooling to condense the steam and discharge surplus heat
The second function for water in such a power plant is to cool the system so as to condense the low-pressure steam and recycle it. As the steam in the internal circuit condenses back to water, the surplus (waste) heat which is removed from it needs to be discharged by transfer to the air or to a body of water. This is a major consideration in siting power plants, and in the UK siting study in 2009 all recommendations were for sites within 2 km of abundant water - sea or estuary.
This cooling function to condense the steam may be done in one of three ways:
In any of these set-ups, there is no basic difference in water consumption or use between a nuclear and a coal plant. Apart from size, any differences between plants is due to thermal efficiency, ie how much heat has to be discharged into the environment, which in turn largely depends on the operating temperature in the steam generators. In a coal-fired or conventionally gas-fired plant it is possible to run the internal boilers at higher temperatures than those with finely-engineered nuclear fuel assemblies which must avoid damage. This means that the efficiency of modern coal-fired plants is typically higher than that of nuclear plants, though this intrinsic advantage may be offset by emission controls such as flue gas desulfurisation (FGD) and in the future, carbon capture and storage (CCS).
A nuclear or coal plant running at 33% thermal efficiency will need to dump about 14% more heat than one at 36% efficiency.[6] Nuclear plants currently being built have about 34-36% thermal efficiency, depending on site (especially water temperature). Older ones are often only 32-33% efficient. The relatively new Stanwell coal-fired plant in Queensland runs at 36%, but some new coal-fired plants approach 40% and one of the new nuclear reactors claims 39%.
OECD Projected Costs of Generating Electricity 2010, Tables 3.3. PCC= pulverised coal combustion, AC= air-cooled, WC= water-cooled
(No nuclear efficiency data in this report, but comparable Generation III efficiency is often quoted as about 36%, and see Table 2.)
In Europe (especially Scandinavia) low water temperature is an important criterion for power plant location. For a planned Turkish nuclear plant, there is a one percent gain in output if any particular plant is sited on the Black Sea coast with cooler water (average 5°C lower) than on the Mediterranean coast. For the new UAE nuclear power plants, because the Gulf seawater at Braka is about 35°C, instead of about 27°C as with the Shin Kori 3 & 4 reference units, larger heat exchangers and condensers will be required.
According to a 2006 Department of Energy (DOE) report discussed in the Appendix, in the USA 43% of thermal electric generating capacity uses once-through cooling, 42% wet recirculating cooling, 14% cooling ponds and 1% dry cooling (this being gas combined cycle only). The spread for coal and for nuclear is similar. For 104 US nuclear plants: 60 use once-through cooling, 35 use wet cooling towers, and 9 use dual systems, switching according to environmental conditions. This distribution is probably similar for continental Europe and Russia, though UK nuclear power plants use only once-through cooling by seawater, as do all Swedish, Finnish, Canadian (Great Lakes water), South African, Japanese, Korean and Chinese plants.
Gas combined cycle (combined cycle gas turbine - CCGT) plants need only about one third as much engineered cooling as normal thermal plants (much heat being released in the turbine exhaust), and these often use dry cooling for the second stage.*
* CCGT plants have an oil or gas-fired gas turbine (jet engine) coupled to a generator. The exhaust is passed through a steam generator and the steam is used to drive another turbine. This results in overall thermal efficiency of over 50%. The steam in the second phase must be condensed either with an air cooled condenser or some kind of wet cooling.
Combined heat and power (CHP) plants obviously need less engineered cooling provision that others since the by-product heat is actually used for something and not dissipated uselessly.
If a coal or nuclear plant is next to a large volume of water (big river, lake or sea), cooling can be achieved by simply running water through the plant and discharging it at a slightly higher temperature. There is then hardly any use in the sense of consumption or depletion on site, though some evaporation will occur as it cools downstream. The amount of water required will be greater than with the recirculating set-up, but the water is withdrawn and returned, not consumed by evaporation. In the UK the water requirement for a 1600 MWe nuclear unit is about 90 cubic metres per second (7.8 GL/d).
Many nuclear power plants have once-through cooling, since their location is not at all determined by the source of the fuel, and depends first on where the power is needed and secondly on water availability for cooling. Using seawater means that higher-grade materials must be used to prevent corrosion, but cooling is often more efficient. In a 2008 French government study, siting an EPR on a river instead of the coast would decrease its output by 0.9% and increase the kWh cost by 3%.
Any nuclear or coal-fired plant that is normally cooled by drawing water from a river or lake will have limits imposed on the temperature of the returned water (typically 30°C) and/or on the temperature differential between inlet and discharge. In hot summer conditions even the inlet water from a river may approach the limit set for discharge, and this will mean that the plant is unable to run at full power.
Sometimes a supplementary cooling tower is used to help, giving a dual system, as with many inland plants in France and Germany and at the Huntly plant in New Zealand, but this means that some water is then lost by evaporation.
Where a power plant does not have abundant water, it can discharge surplus heat to the air using recirculating water systems which mostly use the physics of evaporation.
Cooling towers with recirculating water are a common visual feature of power plants, often seen with condensed water vapour plumes. Sometimes in a cool climate it is possible to use simply a pond, from which hot water evaporates.
Most nuclear power (and other thermal) plants with recirculating cooling are cooled by water in a condenser circuit with the hot water then going to a cooling tower. This may employ either natural draft (chimney effect) or mechanical draft using large fans (enabling a lower profile but using power). The cooling in the tower is by transferring the water's heat to the air, both directly and through evaporation of some of the water. In the UK the water requirement for a 1600 MWe nuclear unit is about 2 cubic metres per second (173 ML/d), this being about half for evaporation and half for blow-down (see below).
The most common configuration for natural draft towers is called counterflow. These towers have a large concrete shell with a heat exchange 'fill' in a layer above the cold air inlet at the base of the shell. The air warmed by the hot water rises up through the shell by convection (the chimney effect), creating a natural draft to provide airflow to cool the hot water which is sprayed in at the top. Other configurations include crossflow, where the air moves laterally through the water, and co-current, where the air moves in the same direction as the water droplets. These towers do not require fans and have low operating but significant maintenance costs. They are used in large nuclear and coal-fired plants in Europe, eastern USA, Australia, and South Africa
Mechanical draft cooling towers have large axial flow fans in a timber and plastic structure. The fans provide the airflow and are able to provide lower water temperatures than natural draft towers, particularly on hot dry days. However, they have the disadvantage of requiring auxiliary power, typically about 0.5% of the plant's output. Mechanical draft towers are used exclusively in central and western USA since they can provide a more controlled performance over a wide range of conditions, ranging from freezing to hot and dry.
Such cooling towers give rise to water consumption, with up to 2.5 litres being evaporated for each kilowatt-hour produced[7], depending on conditions[8]. This evaporative water loss by phase change of a few percent of it from liquid to vapour is responsible for removing most of the heat from the coolant water at the cost of only a small fraction of the volume of the circulating liquid.
Cooling towers with recirculating water reduce the overall efficiency of a power plant by 2-5% compared with once-through use of water from sea, lake or large stream, the amount depending on local conditions. A 2009 US DOE study says they are about 40% more expensive than a direct, once-through cooling system.
Water evaporating from the cooling tower leads to an increasing concentration of impurities in the remaining coolant. Some bleed - known as "blowdown" - is needed to maintain water quality, especially if the water is recycled municipal wastewater to start with - as Palo Verde, Arizona. Replacement water required is thus about 50% more than actual evaporation replacement, so this kind of system consumes (by evaporation) about 70% of the water withdrawn.
Even with the relatively low net water requirement for recirculating cooling, large power plants can exceed what is readily available from a river in summer. The 3000 MWe Civaux nuclear plant in France has 20 GL of water stored in dams upstream to ensure adequate supply through drought conditions.
Despite many coal and nuclear plants using wet cooling towers, in the USA electric power generation accounts for only about 3% of all freshwater consumption, according to the US Geological Survey - some 15.2 gigalitres per day (5550 GL/yr). This would be simply for inland coal and nuclear plants without access to abundant water for once-through cooling. Australian coal-fired power plants consume about 400 GL/yr[9] - the equivalent of Melbourne's water supply.
Where access to water is even more restricted, or environmental and aesthetic considerations are prioritised, dry cooling techniques may be chosen. As the name suggests, this relies on air as the medium of heat transfer, rather than evaporation from the cooling circuit. Dry cooling means that minimal water loss is achieved. There are two basic types of dry cooling techniques available.
One design works like an automobile radiator and employ high-flow forced draft past a system of finned tubes in the condenser through which the steam passes, simply transferring its heat to the ambient air directly. The whole power plant then uses less than 10% of the water required for a wet-cooled plant,[10] but a lot of power (around one to 1.5 percent of power station's output) is consumed by the large fans required.[11] This is direct dry cooling, using air-cooled condenser (ACC) and it is not currently in use on any nuclear power plant.
Alternatively there may still be a condenser cooling circuit as with wet recirculating cooling, but the water in it is enclosed and cooled by a flow of air past finned tubes in a cooling tower.* Heat is transferred to the air, but inefficiently. This technology is not favoured if wet cooling depending on evaporation is possible, but energy use is only 0.5% of output.
* Some mechanical draft towers are a hybrid design incorporating a dry section above the wet section. The mode of cooling used depends on the season, with dry cooling being preferred during the colder months.
In both cases there is no dependence on vaporization and hence no evaporative loss of cooling water. The use of fans also allows for greater control over cooling than relying simply on natural draught. However the heat transfer is much less efficient and hence requires much larger cooling plant which is mechanically more complex. Eskom in South Africa quotes dry-cooled plants as having total station water consumption under 0.8 lites/kWh, this being for steam cycle losses (cf about 2.5 L/kWh for wet-cooled plants). Hardly any US capacity uses dry cooling, and in the UK it has been ruled out as impractical and unreliable (in hot weather) for new nuclear plants. A 2009 US DOE study says they are three to four times more expensive than a recirculating wet cooling system.
Both types of dry cooling involve greater cost for the cooling set-up and are much less efficient than wet cooling towers using the physics of evaporation[12] since the only cooling is by relatively inefficient heat transfer from steam or water to air via metal fins, not by evaporation. In a hot climate the ambient air temperature may be 40 degrees C, which severely limits the cooling potential compared with a wet bulb temperature of maybe 20ºC which defines the potential for a wet system.
Australian projected figures for coal* show a 32% drop in thermal efficiency for air cooling versus water, eg from 33% to 31%.
* In OECD Projected Costs of Generating Electricity 2010, Tables 3.3.
Environmental and social aspects of cooling
Each of the different methods of cooling entails their own set of local environmental and social impacts and is subject to regulation. In the case of direct cooling, impacts include the amount of water abstracted and the effects upon organisms in the aquatic environment, particularly fish and crustaceans. This latter includes both kills due to abstraction and the change in eco-system conditions brought about by the increase in temperature of the discharge water.
In the case of wet cooling towers, impacts include water consumption (as distinct from just abstraction) and the effects of the visual plume of vapour emitted from the cooling tower. Many people consider such plumes as a disturbance, while in cold conditions some tower designs allow ice to form which may coat the ground or nearby surfaces. Another possible problem is carryover, where salt and other contaminants may be present in the water droplets.
Over time, knowledge of these effects has increased, impacts have been quantified and solutions developed. Technical solutions (such as fish screens and plume eliminators) can effectively mitigate many of these impacts but at an associated cost that scales with complexity. In a nuclear plant, beyond some minor chlorination, the cooling water is not polluted by use - it is never in contact with the nuclear part of the plant but only cools the condenser in the turbine hall.
On a regional and global scale, less efficient means of cooling, especially dry cooling, will lead to an increase in associated emissions per unit of electricity sent out. This is more of a concern for fossil-fuel plants but arguably carries implications for nuclear as well in terms of waste generated.
On the policy side, one US DOE report notes that a major effect of the US Clean Water Act is to regulate the impact of cooling water use on aquatic life, and this is already driving the choice towards recirculating systems over once-through ones for freshwater. This will increase water consumption unless more expensive and less efficient dry cooling systems are used. This will disadvantage nuclear over supercritical coal, though flue gas desulfurization (FGD) demands for coal will even out the water balance at least to some extent.
In France, all but four of EdF's nuclear power plants (14 reactors) are inland, and require fresh water for cooling. Eleven of the 15 inland plants (32 reactors) have cooling towers, using evaporative cooling, the other four (12 reactors) use simply river or lake water directly. With regulatory constraints on the temperature increase in receiving waters, this means that in very hot summers generation output may be limited.
With one exception, all nuclear power plants in the UK are located on the coast and use direct cooling. In the UK siting study of 2009 for nuclear new build, all recommendations were for sites within 2 km of abundant water - sea or estuary.
Fresh water is a valuable resource in most parts of the world. Where it is at all scarce, public opinion supports government policies, supported by common sense, to minimise the waste of it.
Apart from proximity to the main load centres, there is no reason to site nuclear power plants away from a coast, where they can use once-through seawater cooling. Coal plant locations need to have consideration for the logistics of fuel supply (and associated aesthetics), with over three million tonnes of coal being required per year for each 1000 MWe plant.
If any thermal power plant - coal or nuclear - needs to be sited inland, the availability of cooling water is a key factor in location. Where cooling water is limited, the importance of high thermal efficiency is great, though any advantage of, say, supercritical coal over nuclear is likely to be greatly diminished due to water requirements for FGD.
Even if water is so limited that it cannot be used for cooling, there is the option of dry cooling (with extra cost) for either coal or nuclear plants. Alternatively the plant can be sited away from the load demand and where there is sufficient water for efficient cooling (accepting some losses and extra cost for transmission[14]).
Generation III+ nuclear plants have high thermal efficiency relative to older ones, and should not be disadvantaged relative to coal by water use considerations. If water is very scarce, dry cooling can be used, though this has some cost and efficiency penalty.
Considerations of limiting greenhouse gas emissions will, of course, be superimposed upon the above.
A further implication relates to cogeneration, using the waste heat from a nuclear plant on a coastline for MSF desalination. A lot of desalination in the Middle East and North Africa already uses waste heat from oil- and gas-fired power plants, and in future a number of countries are expecting to use nuclear power for this cogeneration role. See also WNA Nuclear Desalination information paper.
APPENDIX: Critique of US reports
It is evident that apart from any difference in thermal efficiency which affects the amount of heat to be dumped in the cooling system, there is no real difference in the amount of water used for cooling nuclear power plants, relative to coal-fired plants of the same size. However, some US studies quote a significant difference between coal and nuclear plants, this evidently being related to the (unstated) thermal efficiency of selected examples. The studies exclude nuclear plants on the coast, which employ salt water for cooling. See Appendix for comment on two of these, from EPRI (2002) and NETL (2006).
The March 2002 EPRI Technical Report: Water and Sustainability (volume 3): US Water Consumption for Power production - the next half century aims to estimate future water consumption associated with power generation in the USA to about 2020. It uses some "typical" figures for water withdrawal and consumption which show marked differences between coal and nuclear, without giving the source of these or explaining any reason for their magnitude. It focuses on freshwater only, and ignores plants with seawater cooling. Its conclusions are presented on a regional basis in the light of projected increased generations and likely changes in generation technology such as from coal to combined cycle gas.
No empirical basis for the numbers is given or cited. The 2006 DOE report refers to the EPRI figures as "estimated" and based on data reported by operators to EIA. It would seem that some inefficient nuclear plants are being compared with much more efficient coal plants, without this being made explicit. In fact any projection needs to take account of the thermal efficiencies of plants currently being built, but this is evidently not done.
The second report is August 2006 National Energy Technology Laboratory, Department of Energy: Estimating Freshwater Needs to Meet Future Thermoelectric Generation Requirements DOE/NETL-2006/1235, its 2008 update DOE/NETL-400/2008/1339 and also a more general 2009 report Water Requirements for Existing and Emerging Thermoelectric Plant Technologies DOE/NETL-402/080108. The first two look out to 2030 and are also focused on fresh water needs and use five cooling scenarios applied to regional projections for additions and retirements. Here the assumptions for future coal plants are 70% supercritical[13] and 30% subcritical, the former having very high thermal efficiency, beyond that of any Generation III nuclear plant. However, coal plants are assumed to need flue gas desulfurization (FGD), which usually increases water use.
Cooling water requirements for each type of plant were calculated from NETL data and are tabulated as follows for "model" plants' consumption of fresh water:
Coal, once-through, subcritical, wet FGD
0.52 litres/kWh
Coal, once-through, supercritical, wet FGD
0.47 litres/kWh
Nuclear, once-through, subcritical
Coal, recirculating, subcritical, wet FGD
1.75 litres/kWh
Coal, recirculating, supercritical, wet FGD
1.96 litres/kWh
Nuclear, recirculating, subcritical
2.36 litres/kWh
The figures are puzzling in that supercritical coal should use significantly less than less-efficient subcritical coal-fired plants, and for recirculating use of cooling towers the large difference between subcritical coal and nuclear is unexplained. Clearly there are significant variables which are not accounted for though they must surely be relevant to NETL's projections.
The 2009 DOE/NETL report shows a diagram (Fig 3-6) citing the 2002 EPRI report giving net consumption of 2.27 to 3.8 L/kWh for nuclear*. This is a lot more than that of the figures in the subcritical coal-fired diagram with FGD (Fig 3-2) - 1.9-2.5 L/kWh (0.505-0.665 gal/kWh) with similar blow-down.
* cooling water make-up of 3.0 to 4.1 L/kWh (0.8-1.1 gal/kWh), less blow-down of 0.06-0.20 gal/kWh.
Another diagram (Fig 3-1) citing EPRI 2002 gives net 2.7 L/kWh (0.72 gal/kWh) for nuclear and 2.0 L/kWh (0.52 gal) for subcritical coal. In explanation the text says: "Nuclear plants have a higher cooling tower load relative to net power generation. This is because the steam conditions are limited by metal brittleness effects from the nuclear reactor thereby reducing efficiency." However, neither it nor the EPRI report justify the large difference, which should be directly related to thermal efficiencies. There would have to be a huge difference in average thermal efficiencies for these numbers to be credible - eg 25% versus 36%* - perhaps compounded by climatic differences which affected the proportion of cooling which was actually evaporative. In any case it would appear that some very old nuclear plants are being compared with new coal-fired ones, which is a poor basis for forecasting to 2030.
* Even then the contrast is between 3.8 L/kWh and 2.8 l/kWh respectively, if all cooling is attributed to evaporation, or 2.66 and 2.0 L/kWh if 70% is purely evaporative.
[1] At theoretical full efficiency and considering only the vapour phase this is known as the Carnot cycle. The Carnot efficiency of a system refers to the difference between input and output heat levels and is more generally referred to as thermal efficiency.
[2] This thermodynamic process of turning heat into work is also known as the Rankine Cycle, or more colloquially as the steam cycle, which can be considered a practical Carnot cycle but using a pump to return the fluid as liquid to the heat source.
[3] The function of the condenser is to condense exhaust steam from the steam turbine by losing the latent heat of vaporisation to the cooling water (or possibly air) passing through the condenser. The temperature of the condensate determines the pressure in that side of the condenser. This pressure is called the turbine backpressure and is usually a partial vacuum. Decreasing the condensate temperature will result in a lowering of the turbine backpressure which will increase the thermal efficiency of the turbine. A typical condenser consists of tubes within a shell or casing.
There may be primary and secondary circuits, as in pressurized water reactors (PWRs) and two or three other types. In this case the primary circuit simply conveys the heat from reactor core to steam generators, and the water in it remains liquid at high pressure. In a boiling water reactor and one other type, the water boils in or near the core. What is said in the body of the paper refers to the latter situation or the secondary circuit, where there are two.
[4] Within a nuclear reactor water or heavy water must be maintained at very high pressure (1000-2200 psi, 7-15 MPa) to enable it to remain liquid above 100ºC, as in present reactors. This has a major influence on reactor engineering.
A more detailed treatment of different primary coolants is in the Nuclear Power Reactors paper.
[5] A US Geological Survey report in 1995 suggested 98% of withdrawal is typically returned to source.
[6] For a given electrical output, because the plant needs to be bigger (for given output @36% 1.78 times as much heat needs to be dumped, at 33% 2.03 times as much heat has to be dumped - a 14% difference). If one simply looks at the proportion of heat lost in a particular plant at the two efficiencies the difference is 5% and there is 8% less electricity produced.
[7] For each kWh electrical output, at 33% thermal efficiency 7.3 MJ of heat needs to be dumped. At 36% thermal efficiency 6.4 MJ is dumped. With latent heat of vaporization 2.26 MJ/L, this gives rise to 3.2 litres or 2.8 litres per kWh respectively evaporated if all the cooling effect is simply evaporative. This would amount to 77 or 67 megalitres per day respectively for a 1000 MWe plant if all cooling were evaporative only. In practice, about 60-75% is evaporative, depending on atmospheric factors.
[8] The 2006 DOE report critiqued below shows 2.9 litres/kWh as typical.
[9] On basis of 70% of 255 TWh total produced at water cost of 2.25 litres/kWh (80% of electricity is from coal, mostly using evaporative cooling).
[10] About 0.25 L/kWh at Kogan Creek, including supplementary small amount of wet cooling, 0.15 L/kWh at Milmerran.
[11] 48 fans each 9 metres diameter at Kogan Creek.
[12] In Australia Kogan Creek (750 MWe supercritical) and Milmerran (840 MWe supercritical) coal-fired power stations use dry cooling with ACC, as do Matimba and Majuba plants in South Africa. Kendal in South Africa uses indirect dry cooling system. Dry cooling is apparently also used in Iran and Europe. South African experience puts ACC cost as about 50% more than recirculating wet cooling and indirect dry cooling as 70 to 150% more.
[13] These use supercritical water around 25 MPa which have "steam" temperatures of 500 to 600ºC and can give 45% thermal efficiency. Over 400 such plants are operating world-wide. One stream of development for Generation IV nuclear reactors involves supercritical water-cooled designs. At ultra supercritical levels (30+ MPa), 50% thermal efficiency may be attained.
Supercritical fluids are those above the thermodynamic critical point, defined as the highest temperature and pressure at which gas and liquid phases can co-exist in equilibrium, as a homogenous fluid. They have properties between those of gas and liquid. For water the critical point is at 374C and 22 MPa, giving it a "steam" density one third that of the liquid so that it can drive a turbine in a similar way to normal steam.
[14] In the UK all nuclear plants are on the coast and total transmission losses in the system are 1.5%.
Sources:
UK Environment Agency, 2010, Cooling Water Options for the New Generation of Nuclear Power Stations in the UK