User:Berkut/PBMR

Arguments on the Construction of PBMR Reactors in South Africa

Steve Thomas

SPRU (University of Sussex)

February 1999

This paper examines the arguments for and against the development by the nationally owned utility, Eskom, of a small modular nuclear power reactor, the Pebble Bed Modular Reactor (PBMR), for construction in South Africa and for export. It examines the case from five perspectives:

· the technology;

· why electricity liberalisation and nuclear power do not mix

· the economics of nuclear power;

· the world market for nuclear power plants and the prospects for exports from South Africa; and,

· waste disposal.

There are other important arguments which should be considered, in particular those related to safety. However, I am not qualified to make judgements on this issue and they are referred to only in passing. Prior to discussing the arguments on the PBMR, it is useful to explain briefly the main principles of nuclear power.

The Principles of Nuclear Power

· In naturally occurring uranium, 0.7% of uranium is of a particular type (isotope) of uranium (U235) which spontaneously splits (fissile material) to emit a tiny particle (a neutron). If this neutron hits another U235 atom, it too will split (a fission) to produce two more neutrons (chain reaction).

· If the concentration of U235 is sufficient (a critical mass), the process will be self-sustaining (the plant is 'critical'), producing large quantities of heat in the 'core' of the reactor.

· Two important ingredients are needed to control the process and to utilise the heat, the moderator and the coolant. A moderator is a substance which neutrons collide with but 'bounce off' without absorbing too much energy and without itself being split. It controls the amount of neutrons escaping from the core before they have hit another U235 atom. A good moderator is one which absorbs the least energy and does not absorb the neutrons before they split another uranium atom. Graphite is an excellent moderator; ordinary water is a poorer moderator but is much cheaper. If water is used, the U235 content must be increased (enrichment) to about 3 per cent to allow a chain reaction to take place. A rare isotope of hydrogen (deuterium) can be used to make so-called heavy water (deuterium is twice the weight of normal hydrogen) and this is also an excellent moderator.

· In so-called fast (breeder) reactors (as opposed to the thermal reactors described above), no moderator is used and some of the neutrons escape the core and strike a 'jacket' of uranium where they convert the unused part of the uranium, U238, to fissile material, plutonium, which can be used as a reactor fuel. The jacket is processed to isolate the plutonium for use in more fast reactors. The attraction of this design is obvious, it can use almost 100 per cent of naturally occurring uranium instead of the 0.7 per cent thermal reactors achieve. The disadvantage is equally obvious: it requires the separation, transport and widespread use of the material used to make nearly all nuclear weapons and is regarded as a serious proliferation risk. The technical attractions of the design have lead to huge amounts of public money being spent on this technology. However, in practice, all prototype plants have proved most unreliable and the technology is now all but abandoned.

· In order to produce electricity, the heat in the core has to be transferred to a fluid (a liquid or a gas), the coolant. The heat will expand the fluid (boil it if it is water) and the force of the expanding gas can be used to drive a turbine generator to produce electricity. This principle of transferring heat from a 'boiler' to a turbine generator is the same for all types of thermal power station whether it uses nuclear or fossil fuel. The coolant can go directly from the core to the turbine generator or there can be an intermediate stage where the coolant goes through a heat exchanger to produce steam in a second circuit. Liquids are much denser than gases and so a given volume of liquid can cool much more efficiently than the same volume of gas, so if the coolant circuit with a liquid cooled reactor breaks, the plant will only be cooled by gases, that is, steam and air, and the plant could over-heat catastrophically.

· Ordinary water is a common, cheap coolant for power plants of all types, including nuclear power. Its primary safety disadvantage in a nuclear power plant is that if it escapes, the reactor will not be properly cooled (loss of coolant accident, or LOCA). Water can also be corrosive and will require expensive materials to prevent damage to the coolant pipes. However, water coolant requires much less volume of materials because of its greater efficiency in cooling than gas. So pressurised water reactors (PWRs) of the type built at Koeberg in South Africa, which use water as the coolant, are much more compact than, for example, the British designs of gas-cooled reactor. Of the gas coolants possible, carbon dioxide was used in the British power plant designs, but while this is cheap, it is somewhat corrosive. Helium is entirely inert, but is expensive so leakage has to be avoided.

· Of the many possible technologies, two are of particular relevance to South Africa, the two existing civil nuclear power reactors at Koeberg and the PBMR. The Koeberg plants are each 900 MW (1 megawatt (MW) is 1 million kilowatts (kW)). They are known as pressurised water reactors (PWRs) because the coolant is maintained as liquid despite being at about 300°C by keeping it at very high pressures. This coolant is passed through a heat exchanger in which the energy is transferred to a second circuit in which water is boiled and drives the steam turbine generator. Ordinary water is used as the moderator and as a result, uranium enriched to about 3 per cent is required.

· The PWR is the most widely used design of nuclear reactor in the world and just under half the 430 nuclear power plants in the world are of this design. The main supplier is Westinghouse and its design has been adopted by Framatome (the Koeberg supplier), Siemens and Mitsubishi.

· The PWR is a direct descendant of submarine propulsion units and, as a result, its operating schedule is planned around annual stoppages when the plant is refuelled and maintenance is carried out. Typically, a quarter of the fuel rods are replaced each year, because the concentration of U235 is no longer great enough to maintain full power operation.

· The PBMR uses helium as the coolant and graphite as the moderator and is one of a number of designs that come under the general classification of High Temperature (Gas-Cooled) Reactors, HTGRs or HTRs. The use of helium and graphite gives it several intrinsic safety and technical advantages over, say, the PWR. As noted above, the use of a gaseous coolant reduces the risk from loss of coolant accidents. Being inert, helium can be used at very high temperatures without concerns about corrosion.

· The use of a good moderator like graphite increases the efficiency with which the uranium is used. With HTRs, fuel is made in ceramic pellets (or pebbles) which can also withstand very high temperatures, compared to a PWR where the fuel is in the form of rods of uranium oxide contained in a metal cladding. With HTRs, the moderator is in the form of a coating for the fuel and is an integral part of it, unlike the PWR where the water flows past the fuel. This gives some safety advantages as the moderator which controls the reactor cannot be separated from the fuel.

· This combination of helium coolant, graphite moderator and ceramic fuel allows the reactor to operate at very high temperatures, 750ºC compared to 300ºC in a PWR. This in turn means that a much higher proportion of the energy from the core can be turned into electricity (the thermal efficiency), 40 per cent compared to 34 per cent for a PWR. It also means that a much higher proportion of the U235 can be split, giving high fuel 'burn-up'. This means that the reactors are more economical in their use of uranium and create a much lower volume of used, or 'spent' fuel.

· All high temperature reactors built to date have used highly enriched uranium (HEU) - more than 90 per cent U235. While this may lead to good uranium utilisation, such material is a serious weapons proliferation risk. South Africa's nuclear bombs were built using HEU. The use of such a material as a basis for nuclear power plants to be exported round the world would raise huge concern on proliferation grounds and it is unlikely that the international community would allow South Africa to go ahead using such material. For its PBMR, Eskom plans to use 7-8 per cent enriched uranium, very different to the type of fuel used in HTRs so far.

· Like most purpose-designed reactor types, but unlike the submarine-derived PWR, the PBMR would avoid the need for an annual shut-down for re-fuelling, by re-fuelling while the plant is operating, 'on-line'. In theory, this should mean that extra power can be produced. In practice, on-line refuelling has not always worked out well because the machines for doing it are complex, expensive and prone to break-down. Also, the time required for maintenance, which is carried out at the same time as refuelling, usually exceeds the time required for re-fuelling so on-line refuelling would not reduce the amount of time the plant is off-line.

· For example, in Britain, the Advanced Gas-Cooled Reactor (AGR) was designed to refuel on-line, at full power. But more than 20 years after the first plant went into service, the regulatory authorities still do not allow refuelling at full power because of safety concerns. Ironically, in 1965 when the AGR was chosen, it was the extra output that was expected to be produced because of on-line refuelling, that swung the economic case in favour of the AGR over US designs. This reduced the overall generation cost of the AGR by a small fraction of a penny. This experience will not necessarily be repeated in South Africa but it does demonstrate that refuelling on-line can be a difficult process and that any projected economic advantages to on-line refuelling should be treated with some scepticism.

The Technology

The Track Record of High Temperature Reactors

In nuclear power, as with any other field of technology, design concepts that look good on paper cannot necessarily be turned into viable and economic technologies. It is therefore important to examine attempts by other countries to turn this apparently attractive concept into a commercial technology. The clear intrinsic advantages of the HTR, namely (a) high thermal efficiency, (b) economical use of uranium and (c) better safety, have meant that from the earliest days of civil nuclear power, this class of reactors has been examined carefully by almost every nation that has tried to design nuclear power plants. The first prototype plants of this type were ordered in the late 1950s. The USA and Germany have gone as far as building prototype plants of a commercial size, about 300 MW (a third the size of each Koeberg unit and three times the size of the proposed South African PBMR). German experience is particularly relevant to South Africa because it is German technology which has been sold to South Africa and forms the basis of the PBMR. The UK and Japan have built small-scale prototype reactors for research purposes which do not produce electricity. France seriously considered developing its own commercial scale design of HTR in the late 1960s as an alternative to importing PWR technology. Of the countries which can claim to have nuclear design capability, only Russia and Canada have shown little or no interest in the HTR.

Today, the USA, Germany, the UK and France have now abandoned all interest in HTRs, while Japan's development programme is very slow and there are no plans to build commercial power plants.

The USA: The USA was the first country to build a HTR power plant, the Peach Bottom 1 plant, ordered in 1958 and completed in 1967, which produced about 40 MW of electricity. Like all plants of this design in the USA, it was built by General Atomic (a company owned by Gulf Oil) and operated until 1974. The operating record of the plant seems to have been fairly good and the plant has now been completely decommissioned. None of the US plants is of the pebble bed design.

Confidence in nuclear technology of all types was then so high that even before this plant had been completed, a successor, about 8 times as large was ordered. Fort St Vrain was ordered in 1965 and designed to produce 330 MW. It was owned by a utility, Public Service of Oklahoma but about half the construction cost was paid by the US government. It went critical in January 1974, but did not generate its first power until December 1976 and was only declared commercial (handed over from the supplier to the owner) in 1979, a good indication that all was not going to plan. For a commercial nuclear power plant, the time from first criticality to commercial operation should be less than 6 months (it was four months at both Koeberg units). However, confidence in nuclear technology was undiminished and at the time, the USA was undergoing a huge surge of nuclear orders. In the peak year for orders, 1974, 41 units were ordered. Ironically, only 9 of these plants were completed and all subsequent orders in the USA (a further 41 plants) were cancelled. The plants were cancelled because the costs were too high or electricity demand was not sufficient to justify them.

Orders for full-size plants of the HTR design, without any government subsidy, were first placed in 1971 and by 1974, eight orders had been placed, four for units of 770 MW and four for units of 1160 MW. Little or no progress on these plants was made and with problems at Fort St Vrain becoming apparent, all were cancelled in 1974-75.

Fort St Vrain continued in service from 1976 until August 1989 when its high costs and appalling reliability finally persuaded the owner to give up the struggle and retire the plant, which has now been largely decommissioned. Over its 10 years of commercial service, its average load factor (power produced as a percentage the power the plant would have produced had it operated uninterrupted at full power) was 15 per cent. Typically a plant owner would expect a load factor of about 80 per cent from a nuclear power plant. There was no single overwhelming factor that led to its failure, more a series of different equipment problems.

Despite this bad experience, in 1991, when the US government decided it needed to put money into new reactor development, it looked at three or four technologies, one of which was the Gas Turbine Modular High Temperature Reactor (GT-MHTR). The design was close to the PBMR because it used a gas turbine rather than a steam turbine and was planned in modules, but used fuel rods rather than pellets. This would have been developed partly to consume plutonium taken from dismantled bombs and partly as a civil reactor. The technology was developed until 1995, although it was close to losing funding on several occasions, and in August 1995, the US government finally withdrew support. It used the few resources it was prepared to spend on nuclear technology to support advanced PWRs and BWRs (Boiling Water Reactors, a close relative of the PWR).

At the time, a National Academy of Sciences review revealed that HTR technology had received US$ 900m of government money over 30 years. It claimed that the GT-MHTR would take a long time to get a safety licence. It identified fuel as a particular problem because of the lack of any fuel production facilities. New fuel facilities would have to be licensed and built adding to the delay and cost.

Germany: Germany also has a long history of HTR development dating back to the ordering of the Jülich plant, at the government research centre there, in 1959. This 15 MW plant, financed by the government, was ordered from a group led by Brown Boveri and Krupp and went critical in 1966, generating electricity a year later and continuing in service until 1989. Its reliability seems to have been good for a prototype and in 1970, its successor, sometimes known as THTR-300, Uentrop or Schmehausen was ordered. This too was subsidised by the government but also involved utility funding. The industrial grouping behind it, HRB, again centred on Brown Boveri but with General Atomic support. Subsequently Siemens produced modular designs involving pebble bed reactors but none were built.

THTR-300 went critical in September 1983, but was not connected to the electricity grid until November 1985 and was only declared commercial in June 1987. From June until October of that year, it operated at about two thirds full power, suffering a range of problems including difficulties with the fuel circulation system. It restarted in January 1988 for a couple of months, again running at about two thirds of its full power rating, until more repairs were necessary to the fuel circulation and collection system. It ran for another five months and was shut down due to damage in the gas ducts. Repairs were completed by February 1989. But the plant remained closed on the orders of the safety regulator because of concerns about safety and the unwillingness of the various owners of the plant, including the federal government, to continue to provide subsidies to operate the plant. In 1990, the plant was permanently closed and is being decommissioned.

Siemens and ABB (the new name for Brown Boveri) pooled their expertise on HTRs to form a new company called HTR Gmbh. Their strategy appears to have been to license the technology to countries such as the then Soviet Union, China, Japan and South Africa.

The UK: The UK was a pioneer of nuclear technology. Its first nuclear power plants were scaled-up versions of the plants built to make plutonium for bombs. This used graphite as the moderator and carbon dioxide gas as the coolant. Nine power stations were built using this technology, but the technology was only seen as a stop-gap. Three new technologies were developed to working prototype scale, including the Dragon HTR. This was ordered in 1957 and completed in 1964. It was a research reactor with no electricity generation facilities and ran until 1974. Anecdotally, it was known as a plant that leaked radiation and another design was chosen in 1964 to form the basis of the civil nuclear power programme in Britain. Since then, HTRs have not been seriously considered in Britain.

France: France followed a very similar route to Britain, developing its first civil nuclear power plants from plutonium producing reactors. Like Britain, it too had to choose a new technology route by the mid to late 1960s. The French nuclear research establishment strongly favoured HTRs, but strongly influenced by the utility, American PWR technology was chosen and, as in Britain, HTR technology was abandoned

Japan: Japan has persisted with a wide range of nuclear technologies for much longer than other countries. It imported British technology for one commercial plant in the 1960s, but since then, all commercial orders have been for US designs, PWRs and BWRs. Nevertheless, it has built a medium size plant of its own design (165 MW) using heavy water as moderator. This was completed in 1979 and for many years there was talk about building a plant of 600 MW of this design. This technology line has now been abandoned.

A prototype fast reactor, Monju (280 MW), was completed in 1995, but an incident at the plant in December of that year drained public and regulatory confidence in the plant and it is highly unlikely the plant will run again.

A third line of reactor development using HTRs of a Japanese design has been underway at a slow pace since about 1990. A prototype reactor producing about 30 MW thermal power but no electricity was completed in 1998, some 3 years later than scheduled.

China: For more than 20 years, China has had ambitious plans to launch a programme of civil nuclear power plants and from 1980 onwards, forecasted that about 20 nuclear power plants would be in service within 10-15 years in China. There is still little to show for their efforts. Two imported power plants were completed in 1993-94 (the same design and supplier as Koeberg) and one plant of a Chinese design was completed in 1992. The potential size of the Chinese market and the dearth of nuclear orders in the West mean that nuclear vendors continue to pursue orders in China despite the political, economic and commercial problems that arise. In 1989, China signed a licensing deal with HTR Gmbh to develop HTRs in China. There is little to show for these efforts yet.

Development of Nuclear Technologies

The history of nuclear power development has been one of unfulfilled promises and unexpected technical difficulties. The ringing promise from 1955, of 'power too cheap to meter' is one that has come back to haunt the nuclear industry.

With most successful new technologies, people confidently expect that successive designs become cheaper and offer better performance. This has not been the experience with nuclear power: costs have consistently gone up in real terms and processes which were expected to prove easy to master continue to throw up technical difficulties. The issues surrounding waste processing and disposal which at first were assumed to be easily dealt with, remain neglected.

Despite this history of unfulfilled expectations, two factors have meant that nuclear power continues to be discussed as a major potential energy source. First, the promise of unlimited power independent of natural resource limitations and second, the attraction to engineers and scientists of meeting the technological challenges that are posed. However, in the developed world, patience with nuclear technology is running out. Governments are no longer willing to invest more tax-payers' money in a technology which has provided such a poor rate of return. Electric utilities cannot simply pass on development costs to consumers. Equipment supply companies, which have generally made little or no money from nuclear technology, are unwilling to risk more money on developing technologies which might not work well and which might not have a market.

There is still talk about new nuclear technologies, but a critical look at the real resources going into them shows that little money is now being spent.

Other Technological Aspects

In this first section, the track record of the HTR has been examined and it is clear from this that the world's leading nuclear countries have all examined HTR technology in some depth, especially Germany and USA, arguably the two leading nuclear nations, and none has been able to make a success of it. It is not impossible that South Africa could succeed where so many others have failed, but it seems inappropriate that public money should be gambled on such a risky technology. However, the technological risk does not end with the reactor.

No facilities exist to manufacture the nuclear fuel and these would have to be set up in South Africa. The German reactor of this basic design experienced a number of fuel problems in its short life, so it cannot be assumed that manufacturing fuel pellets will be simple.

Even the conventional part of the plant, the gas turbine, would be a new product developed at Eskom's expense. Eskom's publicity describes this part of the plant as using the 'standard Brayton cycle' implying a well-proven standard product. No gas turbine using helium has ever been operated and a number of its features are substantially novel. Eskom did request the major manufacturers to tender for a full product with guarantees but it appears that none of them responded. One supplier suggested that research, funded by Eskom would be needed before a commercial product could be designed and produced.

Summary

· The HTR has major intrinsic safety advantages which have led most countries pursuing nuclear technology to investigate the HTR.

· Today, the USA, Germany, the UK and France have now abandoned all interest in HTRs, while Japan's development programme is very slow and there are no plans to build commercial power plants.

· The USA and Germany both built a commercial scale plant subsidised by tax-payers. Neither of these plants worked satisfactorily and were closed because of economic, technical and safety problems.

· The history of nuclear power development has been one of unfulfilled promises and unexpected technical difficulties.

· With most successful new technologies, people confidently expect that successive designs become cheaper and offer better performance. This has not been the experience with nuclear power: costs have consistently gone up in real terms.

· Governments are no longer willing to invest more tax-payers' money, electric utilities cannot simply pass on development costs to consumers and equipment supply companies are unwilling to risk more money on developing technologies.

· There is still talk about new nuclear technologies, but a critical look at the real resources going into them shows that little money is now being spent.

· The technological risk is not confined to the design of the PBMR. No facilities exist to manufacture the nuclear fuel and these would have to be set up in South Africa. Even the conventional part of the plant, the gas turbine, would be a new product developed at Eskom's expense.

Why Electricity Liberalisation and Nuclear Power do not Mix

Electricity liberalisation, sometimes called privatisation or re-regulation, is a complicated subject which would not be appropriate to discuss in detail here. However, there is one common feature to liberalisation processes of crucial importance to this debate. In a liberalised system, the activity of generating electricity ceases to be a monopoly, new generating companies are allowed in and power stations are operated on competitive principles. This transforms electricity generation from being amongst the safest investments available to amongst the most risky.

The momentum for liberalisation now seems unstoppable and, sooner or later, even well run monopoly utilities are going to have to give up their monopoly status and run their business under competitive pressures. For South Africa, this may mean that Eskom will be broken up into several competing companies and privatised. Even if Eskom is not broken up and sold, it will have to accept the loss of its monopoly and will have to compete with new companies to supply electricity.

In a monopoly situation, the risk of building new power plants falls on the consumer. If plant construction costs over-run, if the plant does not work well or power stations that are not needed are built, the costs are passed on to consumers. The greatest risk is that there will be insufficient power stations to meet demand leading to power cuts and adverse publicity for the utility. There will therefore be a tendency to over-invest in plant. Thus, in the 1980s when Eskom so over-estimated electricity demand that new coal-fired plant had to be moth-balled on completion of construction, the extra costs inevitably fell on consumers or taxpayers.

In a competitive situation, if utilities make mistakes, they will either lose market share because their plant is too expensive, or they will have to sell at a loss and the costs will fall on share-holders. Utilities choose proven technology for which construction time and costs can be easily controlled and even guaranteed, and for which performance can also be guaranteed. Since the British electricity market was liberalised in 1990, a large quantity of new plant has been built, all of it using combined cycle gas turbines (CCGTs). Nuclear power had to be placed into a separate company which could not be privatised until six years after the reforms had taken place, when it had completed the one plant it had under construction and had abandoned all plans to build more nuclear plants. The British history of nuclear power is a complex one which cannot be fully covered here. However, it is clear that investors regarded a company building, or planning to build nuclear power plants as too risky to invest in. In the British context, the economics of new nuclear power plants appeared very poor, but even if they had been good, or subsidies had been available, the perception of economic risk would have made privatisation impossible.

The situation with existing plants is rather different. Many nuclear power plants, if operated efficiently and not requiring major repairs, can generate enough income from power sales to cover their running costs. Those that cannot will either be retired, as has happened with a number of US plants, or will have their losses met by subsidy, as has been the case with the oldest British nuclear power plants. However, those that can cover their operating costs seldom make a proper return on the investment that was made. Repaying the loans and paying the interest is invariably the largest cost in any assessment of the cost per kilowatt hour of electricity generated from nuclear power.

Privately owned plants which cannot meet their full costs, including capital, are known as 'stranded assets'. The owners argue they built the plants in good faith to meet all demands, they were subject to regulatory approval and under the old monopoly system, the owners were allowed to recover the full cost from consumers. If by changing to competitive markets, plant owners are no longer able to recover all their costs, they claim they should be compensated for the income lost through consumer subsidies. This process of compensating owners of stranded assets is happening at most nuclear power plants in the USA and the plants will continue in operation.

For publicly owned utilities which are privatised, stranded assets are seldom identified and electricity consumers and tax-payers unknowingly bear these costs. For example, in Britain, Nuclear Electric completed the Sizewell B PWR in 1995 for a cost in excess of £3bn. The company was privatised a year later with the Sizewell B plant and eight other relatively new nuclear power plants of the same size as Sizewell B for about £1.7bn, little more than a half the cost of building just one of the nine plants sold. Consumers who paid for these plants, footed the bill of more than £10bn, which was lost during liberalisation.

Summary

· The momentum for liberalisation throughout the world now seems unstoppable and, sooner or later, Eskom is going to have to give up its monopoly status and run its business under competitive pressures.

· In a monopoly situation, the risk of building new power plants falls on the consumer.

· In a competitive situation, if utilities make mistakes, they will either lose market share because their plant is too expensive, or they will have to sell at a loss and the costs will fall on share-holders.

· In a competitive situation, utilities choose technologies with guaranteed reliable performance and no utility operating in a competitive environment will choose nuclear power.

· If Eskom invests in PBMRs now, the money invested will either be lost if Eskom is privatised, because the sale value of Eskom will fall far short of the money spent on its assets. If it is not privatised, tax-payers, the owners of Eskom, will make little or no return on the investment because the market price of electricity will be too low.

The Economics of Nuclear Power

The economics of nuclear power is a highly contentious area. It is often difficult to establish independently verified estimates of the basic construction costs and the operating cost. In addition, the results are crucially dependent on the accounting and investment appraisal assumptions such as the rate of return on capital that is sought (the discount rate) and the life-time of the plant.

These latter factors are of particular relevance to nuclear power because the main element in the cost for each unit of electricity generated is that associated with building the plant, the capital cost. The shorter the expected life-time and the higher the discount rate, the higher these fixed costs will be. In a monopoly system, the assumed life of the plant can be the expected physical life-time because there will be nothing to stop the owner running the plant until it is worn out. In a competitive system, the plant may have to be retired much earlier if it cannot compete with new plants.

The running costs of nuclear power plants are difficult to establish because most electric utilities regard this data as commercially confidential. However, in the USA, utilities are required to publish fully authenticated running costs. In 1997, the cheapest to run nuclear plants cost about 1c/kWh (0.6p/kWh), while the average was about 2.4c/kWh (1.5p/kWh). Of this, about 0.4-0.6c/kWh was fuel cost while the rest, 0.5-1.8c/kWh, represented the non-fuel cost of operation and maintenance (wages, spare parts etc.)

Government owned utilities have usually been able to invest money at very low rates of return on capital partly because new power stations were seen as a safe investment and partly because, for a variety of reasons, governments have tended to require a lower rate of return on capital than private industry. Thus, in Britain before privatisation, the national utility, the CEGB, could invest at a 5 per cent real (net of inflation) rate of return and recover the costs over 35 years. After privatisation, it is known that private investors are looking for about 12-15 per cent real return and recover the capital over 15-20 years.

A simplified scheme can be used to estimate the fixed cost of electricity from nuclear power stations. We can assume that the capital is repaid in equal annual payments over the life-time of the plant. For the interest payments, we can assume that the average amount owed over the life-time of the plant is half the total construction cost. If we do some simple arithmetic based on the cost of Sizewell B, the consequences of the change in lifetime and discount rate are clear.

· Each kilowatt of capacity at Sizewell cost about £3000 to build and will generate about 6000 kilowatt hour (kWh) per year.

· If we recover the costs over 35 years and charge 5 per cent interest, the cost in pence per kWh simply to repay fixed costs and taking no account of running costs, will be:

(Interest paid based + capital repayment) / units of = fixed cost

on the average output per kWh

amount owed per year

(1500 x 100 x 0.05 + 3000 x 100 / 35) / 6000 = 2.7p/kWh

During the process of getting public approval for Sizewell B, the government, realising that its discount rate was well below commercial rates, raised the level to 8 per cent. This change alone raised the fixed cost to 3.4 pence.

If we do the same calculation with an interest rate of 12 per cent and recover the cost over 20 years, generous assumptions in a competitive market, the cost per kWh is 5.5p/kWh. With a 15 per cent discount rate and a 15 year life, the fixed cost is 7.1p/kWh

To put these figures in context, the total cost (fixed and running) of a new coal plant when Sizewell B was first planned was about 3.5p/kWh (British coal was then about four times as expensive as South African coal). So, if the running costs of nuclear were as low as the best US plants, using the original assumptions (5 per cent discount), Sizewell B might have been economic. By the time of privatisation, new gas-fired plants were being bought and these were expected to generate at about 2.9p/kWh and so, with an 8 per cent discount rate, the total cost of power from Sizewell B was perhaps 50 per cent more expensive than gas-fired generation. By 1996, the cost of gas-fired plants and of gas had come down and their efficiency had gone up such that the total generation cost was now about 2.2p/kWh, a quarter of the cost of nuclear power using the same assumptions on life-time and discount rate.

The importance of operating performance should also be clear from these examples. If instead of 6000 kWh per year, the plant had only produced 3000 kWh, the fixed costs would double. Over its life, Fort St Vrain averaged about 1300 kWh per year.

It can easily be seen that nuclear power is so far from being economic in Britain, it is not a serious option for any utility. In France where large numbers of nuclear power plants have been built, construction costs appear to be much lower (they are not independently authenticated). If plants could be built for half the cost of the British plant and generate 7500 kWh per year, the cost per kWh would still be 75 per cent higher than gas-fired plant. So even in the most successful nuclear countries, nuclear power appears to be uneconomic in a competitive market.

The key economic assumptions that have gone into Eskom's estimate for the PBMR are, (a) the construction cost is assumed to be about US$1000 (£625) per kW, (b) the plant life is 40 years, (c) the discount rate is 6 per cent and (d) the assumed availability is 95 per cent (8300 kWh per year). The expected running cost is not fully documented, only the fuel cost which is estimated to be about 0.4c/kWh, equal to the cheapest US nuclear power plants, is included. The total running cost is therefore likely to be about 1c/kWh (0.6p/kWh).

For comparison, this means Eskom expects the PBMR to be built for about 20% of the cost of the most recent British nuclear power plant and they expect it to be able to achieve a reliability better than any nuclear plant in the world has ever achieved over several years. At £1=$1.6, this gives a fixed cost, using these assumptions, of about 0.4p/kWh. If we accept these remarkable construction costs and availability, but put in commercial discount rates and life-times, but at the low end of the likely values, 12 per cent and 20 years, the fixed cost doubles to 0.82p/kWh. If we use the values for discount rate and plant life-time generally used in Britain now, 15 per cent and 15 years, the fixed cost increases to 1.1p/kWh. Simply by changing the investment appraisal parameters to ones more appropriate, much of the cost advantage of the PBMR over CCGTs has largely disappeared.

The importance of the life-time is clear, but the discount rate may be seen as a rather esoteric debate which it is hard to relate to. However, the reality is that the choice of discount rate is at the heart of the debate about how national resources are allocated. The amount of investment capital available to a country is not unlimited. If money is spent on low-return projects, money will not be available to higher return projects and the economic growth of the country will suffer. The discount rate is as high as it is in Britain because that is the rate of return that the projects can achieve. If the government (and Eskom is owned by the South African government) spends money on low-return projects, there could be two effects: first, money will not be available to the private sector to invest in projects that will generate more wealth; and second, public sector projects, perhaps even within Eskom, such as urban and rural electrification, with a much better rate of return will not be funded.

It is not clear how fully the PBMR has been costed and whether equipment suppliers have been identified. However, even if suppliers are known and costs have been quoted, all the history of nuclear power suggests that these costs will not be an accurate reflection of the actual costs. Two main factors, uncertainty about the features that the safety regulator will demand and the risk that, with an unproven design, unforeseen difficulties will arise, mean that no credible supplier would quote a guaranteed fixed cost. Even if such guarantees were given, there must be some doubt about whether they were worth the paper they were printed on. Even a small nuclear power plant such as the PBMR would produce electrical output worth about £20m per year. Eskom plans these plants in clusters of ten so any design fault would probably be repeated ten times over before it was discovered. If this resulted in a delay of only a year to construction, the value of the lost power would be £200m which the supplier would be liable for. Few companies have the resources to back such a guarantee and even fewer would choose to do so.

The HTR has undeniable intrinsic safety advantages which probably make a catastrophic accident such as occurred at Chernobyl impossible. However, these intrinsic safety advantages are not sufficient to guarantee the safety of the plant. A competent safety regulator would not be prepared to give approval for the design until the full detailed design was available and the plant could not get an operating licence until it was built. There is ample experience in the West of plants of similar basic design to those already in operation, running into construction cost and time overruns because detailed design points were not acceptable. The German experience with the THTR-300 plant, the fore-runner of the PBMR which had the same intrinsic safety features is relevant here. This plant was licensed and in service for a year when problems at the plant led to the withdrawal of the operating licence, a factor instrumental in its closure soon after.

The British experience with the AGR is particularly salutary in this respect. When the Dungeness B plant was ordered in 1965, a prototype plant of this design was operating, apparently successfully. The plant was ordered under fixed cost terms from a British supplier. The detailed design proved to contain serious errors which resulted in constant redesigns throughout the construction period. The supplier and two successor companies went bankrupt, so cost guarantees proved worthless. The plant was finally declared commercial in 1988 after 23 years of continuous construction and huge cost overruns, all of which were paid for by electricity consumers. The lengthy construction period (some of the equipment was obsolete before the plant entered service) and the numerous design errors mean that the plant will never operate as designed and in 1998, one of its better years, the load factor was only 42 per cent.

The reliability levels projected by Eskom are also hard to justify based on Eskom's track record with the Koeberg plant. In 1996, the latest year for which there is full data, the average load factor for the world's nuclear power plants was 77 per cent. Over the 12 years that Koeberg had been in service, the plants averaged a load factor of 58 per cent. In 1997 and 1998, the plants did rather better, but neither was in the world's top 50 plants. There is therefore nothing in Eskom's record to suggest that it is capable of world-beating performance with nuclear power plants, especially with a new and unproven design.

If we assume that Eskom's construction cost estimate is half what costs would really be - this would still make the PBMR the cheapest nuclear plant in the world to build - and we assume the load factor achieved is a little above the average of plants in the rest of the world (7000 kWh per kW per year) and we recalculate the fixed costs, the equation is as follows, using a 12 per cent discount rate and a 20 year life-time

625 x 100 x 0.12 + 1250 x 100 / 20 / 7000 = 2.0p/kWh

or, using a higher discount rate (15 per cent) and shorter life-time (15 years),

625 x 100 x 0.15 + 1250 x 100 / 15 / 7000 = 2.5p/kWh

We can compare this with the full cost new gas-fired plant in Britain of about 2.2p/kWh. It is clear that even if South Africa could build plants at less than half the cost of Britain, if it could operate them at above the world average level of reliability, and if running costs were as low as the best US plants, gas-fired plants would be much cheaper.

Summary

· Eskom's cost estimates for the PBMR are unrealistic in a number of respects.

· The rate of return on assets, 6 per cent, is far too low, and if money is invested in projects with such a poor rate of return, there will be insufficient capital to go ahead with some private and public sector projects offering a much better rate of return.

· The assumed life-time of the plant is too long and does not reflect the fact that facilities are generally retired, not when they wear out, but when new plants are available with better economics.

· Using real data from Britain, it is possible to show that by putting in more appropriate estimates for these factors, nuclear power went from being competitive to costing about three times that of the cheapest alternative.

· Eskom's estimates of construction cost and operating performance for the PBMR seem hopelessly out of line with experience of nuclear technology in the rest of the world. The PBMR could prove to be a world-beater in terms of capital costs, operating performance and running costs, but it could still turn out to be more expensive than new gas-fired plants.

The World Market for Nuclear Power Plants

Eskom's evaluation of the PBMR is based on projections of an annual market of 30 units, 10 for installation in South Africa and 20 in the rest of the world. It is therefore important to establish what the world market for nuclear power plants is and what share South Africa might hope to gain from it.

If we start with Europe, 10 countries have built nuclear power plants. Austria closed its plant without operating it after a referendum. Italy closed its three plants after a referendum. Sweden is committed to closing its plant early after a referendum. The newly elected German government has committed itself to phasing out nuclear power. The Netherlands and Switzerland are also likely to phase out nuclear power, while the Spanish government ordered the abandonment of work on several unfinished plants in the 1980s. As argued above, new nuclear orders in Britain are not feasible, leaving only Finland and France as the only countries where new orders are possible, although not inevitable. France has spent huge amounts of money developing its own nuclear capability and it is inconceivable that, if orders were placed, it would not use French companies.

For more than 20 years, Turkey has talked about placing nuclear orders and frequently, deals are said to have been imminent. So far, these have all come to nothing and it seems unlikely that Turkey will be a major market for nuclear power in the next decade.

In North America, no orders not subsequently cancelled have been placed since 1974. Canada has developed its own technology which is now running into severe problems on the economics and safety side with several units shut down for several years as a result. It is barely conceivable that any new orders would be placed. In the USA, more than 100 nuclear orders were cancelled, losing consumers billions of dollars. As in Canada, the electricity industry is being liberalised and many existing nuclear plants are being categorised as stranded assets. The two Mexican units took more than 20 years to build and cost over-runs were huge. Given this poor record, new orders for nuclear power in any of these countries are not feasible.

In South America, Brazil and Argentina have built nuclear power plants. Argentina has two operating plants and has been struggling to finance completion of a third plant, of Canadian design for more than 20 years. Brazil has one operating nuclear plant which, over a 20 year life, has an average availability of about 20 per cent. It may complete a second plant of German design which started construction in 1975 and will cost about US$9bn, making it about the most expensive nuclear plant built. These countries are unlikely to want to repeat their sad experience with nuclear power, nor are their neighbours likely to launch new programmes.

In Africa, only South Africa is actively pursuing nuclear power and the chances of nuclear sales outside South Africa are minimal.

This leaves only Asia as a possible market for nuclear power. The two giants of the continent are India and China, both with nuclear power programmes. India and Pakistan both acquired nuclear power plants in the 1960s but after India exploded a nuclear bomb in 1975, all international nuclear contacts were cut. As a result it has tried to develop its own designs based on the plant it bought from Canada. It now has about 10 small (200 MW) plants in service. All have seriously overrun their construction time and cost forecasts and have been hopelessly unreliable. India is now trying to buy a plant from Russia, but it is unlikely that either side has the cash to carry out this project. Pakistan has recently bought a small plant from China of Chinese design. Like India, its poor record on nuclear proliferation makes it largely impossible for Western countries to do business there with nuclear technology.

China has, for the past 20 years, had ambitious plans to build nuclear power plants of imported design and of its own design. These have resulted in few orders so far: two plants are in service of French design, two more French plants are on order and two Canadian plants are on order. One plant of Chinese design, a 300 MW PWR, is in service, but is currently off-line with serious equipment problems. One plant of this design was sold to Pakistan and China is planning to build further units of this basic design, but double the size. All nuclear vendors are active in China because of the potential size of the market, but it is doubtful whether China can finance a significant nuclear power programme.

As noted previously, Japan has developed a number of its own nuclear technologies, but none of these has been ordered for commercial use. All its operating plants are of US design and Japanese companies such as Mitsubishi, Hitachi and Toshiba have spent large sums of money over the past 30 years developing an understanding of these technologies as well as manufacturing facilities for them. While Japan now has a large number of operating plants (53 at the beginning of 1999), public opposition and problems in finding sites due to seismic activity mean that further orders are now very difficult. There is no room on established sites for further plants and, now, only two plants are under construction. If Japan does order further plants, they will almost certainly be more units of US design or units using a new Japanese design.

Of the other Asian countries, South Korea and Taiwan have nuclear power plants in service. Korea has 14 plants in service and another 3 under construction. It has expended a large amount of effort transferring US technology in and has built up full manufacture facilities. It is highly unlikely that future nuclear orders would not be supplied using these facilities. Taiwan has six plants in service and two on order. When these two plants are complete, there will be little scope for further nuclear plants. Other Asian countries, such as Thailand and Indonesia have, for 20 years or more, discussed the possibility of ordering nuclear plants. However, there is little to suggest that these discussions will soon be turned into nuclear orders.

The Market for South African Nuclear Power Plants

It seems likely that the world market for nuclear power plants may be no more than one or two units a year. It is not clear whether South African designed plant could be expected to win any of this market mainly because of the conservatism of the market.

The accidents at Three Mile Island (USA) and Chernobyl (Ukraine) have alerted nuclear buyers to the economic risk arising from such accidents. Following any serious accident, all plants throughout the world have to demonstrate (if that is possible) that they are not vulnerable to such a set of events. This can be expensive and time-consuming. If modifications are required, there is some comfort in owning a type of plant widely installed elsewhere whose owners will pool resources to solve the problem quickly and efficiently.

The record of rivals to the established designs, the PWR and the BWR, is poor especially for the HTR and the breeder reactor, designs with many theoretical attractions but which do not seem able to be translated into a working commercial design. Buyers therefore have a strong incentive to stick with tried and tested designs. Buying a new design from a country with no track record in nuclear reactor technology appears an enormous risk.

Summary

· Eskom's evaluation of the PBMR is based on projections of an annual market of 30 units, 10 for installation in South Africa and 20 in the rest of the world.

· It seems likely that the world market for nuclear power plants may be no more than one or two units a year.

· Buyers have a strong incentive to stick with tried and tested designs. Buying a new design from a country with no track record in nuclear reactor technology appears an unjustifiable risk.

Waste Disposal

When nuclear power plants were first planned and built, there was little consideration of how waste would be dealt with and worn-out plants removed. It was assumed that new technologies would emerge and costs would be small.

In most countries, waste is divided into three categories. Low-level waste (LLW) is not strongly radioactive and humans would require significant exposure to suffer any health consequences. After a few decades, the radioactivity has generally decayed sufficiently that the material presents little hazard. Intermediate level waste (ILW) is much more strongly radioactive, it remains radioactive for much longer and must be dealt with much more carefully. High level waste (HLW) is not only strongly radioactive but it also generates large quantities of heat. While activity does decay to some extent, HLW must be kept away from human contact indefinitely.

Most countries have had some limited means of dealing with LLW for several decades. Medical and scientific uses result in small quantities of LLW, the isotopes themselves, but also everything they come into contact with, such as gloves and lab coats. At first, this material was simply bull-dozed into holes in the ground and covered. Now, greater care is taken and it is placed in sealed concrete containers and usually buried in shallow ground. It is assumed that by the time the concrete containers have failed, the radioactivity is no longer a hazard. These original dumps are now becoming full: their capacity can be eked out by compaction techniques, but most countries are now searching for new sites. This is invariably politically contentious and few countries have had any success in the last couple of decades in siting new dumps.

In Britain, it was decided in the mid-80s that all LLW would be disposed of in a new deep engineered facility, which would also take all ILW, when the existing facility at Drigg was full. This would clearly raise the costs by a large amount, probably an order of magnitude. However, proving that the geology of such a facility would be stable over a long enough period that it could be assumed there would be no risk that radioactive material would get into the ground water, is a difficult task. It was planned that a test hole be drilled and the geology observed over a decade before the facility was built. A public inquiry rejected the case in 1997 for the one site selected in Britain. There is now no investigation for alternative sites. If the process started tomorrow, an optimistic time-table might require 5 years to identify another potential site, a couple of years for public consultations (the siting would be bitterly resisted), 15 years to build and observe a test drilling, 5 years to build a commercial facility. Britain therefore cannot have a new LLW facility until 2025, by which time LLW will be piling up in temporary stores.

As the standards for LLW disposal have been raised, the costs have gone up. In the last 10-15 years, LLW disposal costs in the USA have been rising at about 6-7 per cent per year in real terms, that is, doubling every 10 years. There is little sign that this price escalation is falling away and, while waste disposal is still quite a small part of nuclear generation costs, if this process is not checked, it could become significant.

ILW is typically material that has been in close contact nuclear fuel, for example, steel vessels. There are no facilities for final ILW disposal in Britain or in most other countries - the only modern facility is a deep repository in Sweden. The material is presently stored in temporary containers on the surface awaiting the construction of the facility described above. Most such material was temporarily packed in containers designed to last a decade or two. The late completion of the disposal facility will mean that this material will have to be unpacked and re-packed at significant expense and will be a hazard over that period.

HLW represents the most intractable technical problem, although the volumes of material are much lower than for the other categories. Essentially, HLW is either spent fuel or the product of the reprocessing of spent fuel. Disposal facilities must be designed such that for thousands of years, there can be no risk that the material can get out of its containers and get into the ground water where it would come into contact with humans. There is a difficult philosophical debate about whether the material should be retrievable or not. If the material is retrievable, if anything goes wrong with the storage facility, it can be retrieved and made safe, but the material is accessible and can be misdirected. If the material is not retrievable, the pros and cons are reversed. There is no clear winner to this debate yet.

At present, no country in the world has identified a site for the disposal of HLW and all material is stored in temporary surface facilities. The technical rationale for this is that the spent fuel is still generating too much heat for it to be disposed of - any containers would come under intense strain because of this heat and would not be able to last the thousands of years required. Thus, in Britain, a decision was taken in about 1980 not to even look for sites for 50 years. However, until sites are identified, the geology proven and the methods of containment subjected to proper public scrutiny, the costs cannot be predicted with any confidence, nor can it even be certain that the process will be politically feasible.

Of particular relevance to the waste debate is the process of decommissioning plants at the end of their life and removing all radioactive material for disposal in proper waste disposal sites so that the land can eventually be released for unrestricted use ('green-field' status). Until this has been done, there is a risk that radioactivity from the plant will leak into the environment damaging the ecology. Decommissioning does generate large quantities of LLW and some ILW.

There is almost no experience in the world of decommissioning a commercial scale plant that has operated over a full life-time to green-field status. As with waste disposal, estimated costs are escalating rapidly. If the costs are accounted for properly from the beginning of operation of the plant, they do not have a large impact on the economics of nuclear power. Under the 'polluter pays' principle, this can only be done by setting up a 'segregated' fund (one that cannot be drawn upon by the plant owner for other purposes) and placing the funds in low risk investments so that when decommissioning is required, there is little risk that the funds will have been lost or used for another purpose.

A possible source of confusion with the spent nuclear fuel is the role of reprocessing. The rationale for reprocessing was mainly that it separated out from the spent fuel plutonium, which could be used to make bombs, or used in fast reactors. It does not destroy radioactivity, it merely separates out the fuel into its constituent parts, some of which might have a use, e.g. plutonium, but most of which still has to be disposed of as HLW. Given that weapons production from civil nuclear power plants is not politically acceptable and that fast reactors have now been abandoned, all the material still has to be disposed of. Reprocessing creates large quantities of LLW as all the material involved in reprocessing becomes LLW. It is a very expensive process which has occasionally resulted in leakage of radioactivity into the environment. Most countries now acknowledge that the cheapest and safest way of dealing with spent fuel is to dispose of it as HLW without any processing.

Overall, the political, technical and economic feasibility of disposal of all types of waste and of decommissioning plants has yet to be proven anywhere in the world. A responsible policy would appear to be to carry out investigations into these processes so that there is confidence that when these processes are required, they are technically proven and the resources to carry them out are available.

Summary

· The issue of waste disposal has been neglected throughout the world. Few modern facilities exist for even the most easily handled waste and for the most difficult waste, plans remain tentative.

· Until modern working facilities for disposal of all types of waste are demonstrated, it will not be clear whether waste disposal, and hence nuclear power is a sustainable technology.

Conclusions

The development of the PBMR by Eskom would represent a highly risky venture which would be underwritten by tax-payers and electricity consumers. Despite the investment of millions of dollars of tax-payers' money, countries with the technological capability of Japan, Germany, France and the UK have failed to develop their own independent nuclear technologies. HTR technology similar to the PBMR has been investigated in depth by most of the world's major nuclear power design nations, including the USA and Germany. Despite the technological capabilities of these countries, commercial scale prototype plants proved failures costing tax-payers many millions of dollars and Germany and the USA have abandoned any interest in this technology.

Eskom's justification for the PBMR is based on achieving a substantial number of export sales (20 units per year). However, a detailed examination of the world market for nuclear power shows that very few nations are likely to order new nuclear power plants. The poor market prospects for nuclear power are the result of: (a) the poor economics of nuclear power; and (b) the pervasive moves to liberalise electricity markets, which transfers risk in building new plants from consumers to share-holders. Those countries that do order plants are likely to choose well-proven options which have been developed by the world's leading nuclear power companies. The issues of waste management have not been fully addressed anywhere in the world and until it can be proven that viable methods are available for dealing with waste, pursuing any nuclear power technology cannot be regarded as a sustainable policy.