At 3:45 PM on September 5, 1945, history was made at Chalk River Laboratories in Ontario as the Zero Energy Experimental Pile or ZEEP achieved criticality for the first time. In that moment, Canada entered the nuclear age – only the second country after the United States to do so. While rarely thought of as a leader in nuclear technology, Canada has long punched above its weight. During the Second World War, uranium from Canadian mines fuelled the first atomic bombs, while many Canadian scientists like Dr. Louis Slotin were intimately involved in the Manhattan Project. After the war, Dr. Harold Johns of the University of Saskatchewan developed cobalt therapy for the treatment of cancer, making Canada a world leader in the production and use of medical isotopes. But perhaps Canada’s greatest achievement in the nuclear field is the CANDU, a highly-innovative nuclear reactor that ranks among the safest and most economical in the world. When first introduced in the late 1960s, CANDU seemed poised to revolutionize the nuclear power industry, but due to various political and economic factors only a handful of units were ever built. In this era of looming climate catastrophe, where nuclear power looks to be the only viable means of producing clean, reliable electricity, it is worth asking: what made the CANDU so special, and why did it fail to catch on? This is the fascinating story of Canada’s forgotten super-reactor.

CANDU, short for CANadian Deuterium Uranium, was developed by Atomic Energy of Canada Limited or AECL, a crown corporation formed in 1952 to develop peaceful uses for nuclear power. The innovative design of this reactor grew out of Canada’s unique limitations as an industrial nation. While Canada had vast reserves of uranium ore in its northern territories, it lacked the capacity to enrich said uranium into conventional nuclear reactor fuel. And for those unfamiliar with the process of enrichment and why it is so important, it is worth taking a slight detour to examine the inner workings of a typical nuclear reactor.

At the heart of every atom lies a small, dense structure called a nucleus, composed of two types of subatomic particles or nucleons: protons, which have a positive electron charge; and neutrons, which have no charge. This is in turn surrounded by a cloud of negatively-charged electrons which determine the atom’s overall electric charge and reactivity – though this is not relevant to the subject at hand. The number of protons in the nucleus – i.e. its atomic number – determines the type of element; for example, hydrogen has one proton, helium two, lithium 3 and so on. The number of neutrons, meanwhile, determines the isotope. For example, if an atom of carbon, which always has 6 protons, also has 6 neutrons, then it is the isotope carbon-12; but if, instead, it has 8 neutrons, then it is the isotope carbon-14 – and to learn more about that particular isotope and how it is used to determine the age of archaeological finds, please check out our previous video How Do We Know How Old Things Like Dinosaur Bones Are?

When atomic nuclei become too large or suffer an imbalance between the number of protons and neutrons, the the weak nuclear force – one of the four fundamental forces of the universe along with the strong nuclear force, electromagnetism, and gravity – is no longer able to hold the nucleons together, and the nucleus becomes unstable and begins to break apart. This is known as radioactive decay or simply radioactivity. There are many different types of radioactive decay, the four most common being alpha decay, beta decay, gamma decay, and spontaneous fission. In alpha decay, the nucleus emits an alpha particle composed of two protons and two neutrons; in beta decay it emits a beta particle (identical to an electron), while in gamma decay it emits photons of high-energy, short-wave electromagnetic radiation known as gamma rays. Of these forms of ionizing radiation, gamma rays are the most highly-penetrating and dangerous to living cells; while alpha particles can be stopped by paper, human skin, or even a few centimetres of air and beta particles by a thin sheet of aluminium, it takes a thick layer of lead or concrete to block gamma rays.

In each of these forms of radioactive decay the nucleus transforms or transmutes into a new element and isotope, which itself may also be radioactive and undergo further decay. This goes on until a stable isotope is finally reached. For example, the most common isotope of uranium, uranium-238, goes through a decay chain of up to 14 steps, transmuting into Thorium-234, Protactinium-234, Uranium-234, Thorium-230, Radium-226 and so on before finally becoming the stable isotope Lead-206.

The fourth major form of radioactive decay is spontaneous fission, in which the nucleus splits or fissions, producing two or more smaller nuclear fragments known as fission products and one or more free neutrons. If these neutrons are directed towards other heavy nuclei, they can strike and split them in a process known as induced fission, releasing more free neutrons which go on to fission yet more nuclei and so on in a nuclear chain reaction. Such reactions can release enormous amounts of energy; for instance, a single kilogram of uranium, burned in a nuclear reactor, can produce the energy equivalent of 500 tons or 5 standard rail cars of coal.

However, a chain reaction can only become self-sustaining under very specific conditions. For example, a minimum quantity of nuclear fuel configured in a particular geometry is needed to ensure that more neutrons remain in the reactor to trigger fissions than escape into the outside environment. This is known as the critical mass. Furthermore, the probability of a neutron fissioning a nucleus is highly dependent on its energy, with the highest probability or neutron cross-section for uranium being achieved by low-energy or thermal neutrons at around 0.025 electron-volts (as an aside, neutron cross-section is measured in units called barns, equivalent to 100 square femtometres. This was coined by physicists working on the Manhattan Project, who quipped that to a neutron, a uranium nucleus was “as big as a barn”. A related unit created around the same time is the shake, equivalent to 10 nanoseconds and derived – and we’re not making this up – from the expression “two shakes of a lamb’s tail”. As if you needed any more proof that physicists are very, very odd people…).

Unfortunately, the free neutrons given off during atomic fission are high-energy or fast neutrons, with a relatively low probability of inducing further fissions. This is why most reactors place a material called a moderator between the fuel elements to slow the free neutrons down to the correct energy level and sustain the reaction. The earliest reactors, including the historic Chicago Pile 1, used ultra-pure carbon in the form of pyrolytic graphite as the moderator. However, this material had numerous shortcomings – namely the annoying habit of spontaneously bursting into flames. Indeed, this dangerous property played a major role in two of history’s greatest nuclear reactor disasters: the 1957 Windscale Fire in Sellafield, England; and the 1986 Chernobyl explosion in Ukraine – and to learn more about the extraordinary – and somewhat insane – process of building the world’s first nuclear reactor, please check out our previous video That Time Scientists Built the World’s First Nuclear Reactor Under a Chicago Football Stadium.

As a result, most subsequent reactors have instead used ordinary distilled water as the moderator, which conveniently doubles as a coolant for carrying heat away from the core. But water has its own downsides, particularly its ability to absorb large numbers of neutrons. The only fissile isotope found in nature – that is, the only one capable of sustaining a nuclear chain reaction – is uranium-235. However, this isotope accounts for only 0.72% of natural uranium, the rest being non-fissile uranium-238. This results in a very low rate of spontaneous fissions, meaning that if natural uranium is placed in a reactor moderated with ordinary water, that water will absorb too many neutrons to sustain a chain reaction. The conventional solution is to enrich the fuel to contain a higher proportion of uranium-235, producing a greater neutron flux that can overpower the neutron-absorbing properties of the water moderator. Most reactor fuel for light water moderated reactors is made of low-enriched uranium or LEU, which contains around 3-5% uranium-235. Unfortunately, uranium enrichment is an extraordinarily resource-intensive process, requiring giant facilities full of centrifuges and other equipment and vast amounts of electrical power – resources post-war Canada simply did not have.
Thankfully for the physicists at AECL, there was another solution: heavy water. This is a form of water containing deuterium, a heavy isotope of hydrogen with an extra neutron in the nucleus. As its name suggests, heavy water is 10% denser than ordinary light water. It also has a neutron capture cross-section 508 times smaller than that of ordinary light water, meaning that a heavy-water moderated nuclear reactor can run on unenriched natural uranium fuel. However, the scattering cross section of heavy water is also lower than that of light water, meaning that more of it is needed to slow neutrons to the correct energy levels. Furthermore, it is very expensive, currently trading at nearly $600 per litre. This is due to the extremely resource-intensive process used to produce it, which involves using large amounts of hydroelectricity to electrolyze ordinary water. Indeed, during the Second World War, the Nazi effort to develop a nuclear bomb was focused on heavy water-moderated reactors. However, the only major source of heavy water in Europe was the Norsk Hydro plant in telemark, Norway, where it was produced as a byproduct of fertilizer synthesis. Knowing this, in 1943 British-trained Norwegian commandoes conducted a series of sabotage raids which succeeded in destroying the heavy water plant and severely crippling the nascent German nuclear program – and to learn more about this extraordinary story, please check out our previous video How Close Did the Nazis Actually Come to Building an Atomic Bomb?

This brings us back at last to the Canadian nuclear program, which due to their economic and logistical advantages used natural uranium fuel and heavy water moderator in most of its earliest reactors, including ZEEP in 1945, the National Research Experimental Reactor or NRX in 1947, and the National Research Universal Reactor or NRU in 1957. Interestingly, on December 12, 1952 NRX became the site of one of the world’s first major nuclear accidents when a malfunction in the cooling system resulted in a partial meltdown of the core. Cleanup and repair of the reactor involved 150 personnel loaned from the U.S. Navy nuclear submarine program, including one Lieutenant Jimmy Carter – later President of the United States. But while these relatively small research reactors continued to operate for decades, producing mountains of valuable data and life-saving medical isotopes, when in 1955 AECL began work on a full-scale civilian power reactor they ran into another logistical barrier. Most power-generating reactors use a primary coolant loop of either light or heavy water to carry heat from the core to a steam generator, where the heat is transferred to a secondary loop and used to generate steam. This steam is then passed through a turbine, turning a generator and producing electricity. To prevent the water in the primary coolant loop from boiling, it is maintained at high pressure – typically around 150 atmospheres. However, Canadian industry in the 1950s and 60s lacked the ability to produce sufficiently large pressure vessels to contain a reactor core. Indeed, when construction began in 1961 on the pilot-scale Nuclear Power Demonstration or NPD reactor near Chalk River Laboratories, the containment vessel had to be contracted out to a firm in Scotland. As this design could not be economically scaled up, halfway through construction AECL engineers developed a radically new core design that would become the cornerstone of the full-scale CANDU reactor system.

In this design, the heavy water moderator is contained in a large, horizontal cylindrical tank called a calandria. As this moderator is unpressurized, the calandria does not need to be particularly robust and can be more easily manufactured in Canada. Instead, the fuel elements are contained in multiple smaller pressure tubes running horizontally through the calandria, which can be produced using conventional pipe-fitting technology. These tubes, which carry pressurized coolant past the fuel elements, are made of zircaloy – an alloy of zirconium, tin, and niobium all but transparent to neutrons – and are contained within larger calandria tubes to prevent radioactive coolant from leaking into the moderator. The gap between the pressure and calandria tubes is filled with carbon dioxide gas, which acts as an insulator to prevent heat from leaking into the calandria.

The CANDU is fuelled by natural, unenriched uranium in the form of uranium dioxide, a black ceramic-like material that is compressed into small cylindrical fuel pellets. These are stacked inside hollow zircaloy fuel pins which are then sealed and assembled into fuel bundles measuring 500 millimetres long, 100 millimetres in diameter, and weighing 22 kilograms. Early fuel bundles contained 37 fuel pins of equal diameter, but in the mid-1980s AECL introduced improved CANFLEX bundles with 43 pins of different diameters to improve burnup homogeneity and heat transfer to the coolant.

These bundles, along with the horizontal pressure tubes, give the CANDU a capability almost unique among reactor designs: on-line refuelling. After a certain period of operation, reactor fuel becomes so depleted in fissile U-235 that it can no longer sustain a chain reaction and must be replaced. In most nuclear reactors, where the fuel elements are arranged vertically, this involves shutting down the reactor, allowing the fuel to cool down, removing the top cap of the reactor vessel, and swapping out the fuel elements. This process can take weeks or even months, severely cutting into the reactor’s production capacity. By contrast, CANDU reactors can be refuelled continuously while they are still running. A pair of robotic refuelling machines on either side of the calandria insert fresh fuel bundles into one end of the pressure tubes, causing spent bundles to fall out the other side. This unique capability gives CANDU an extremely high capacity factor – the percentage of a power plant’s total lifespan actually spent generating electricity. When CANDU was commercially introduced in the late 1960s, most regular light water reactors had capacity factors of around 50%, spending nearly half their lifespans being refuelled or undergoing maintenance. CANDU, by contrast, offered capacity factors of 70% or greater, which – along with the use of cheaper natural uranium fuel – theoretically made the technology much more economical in the long term. Indeed, Unit 7 reactor at the Pickering B nuclear power plant in Ontario currently holds the world record for continuous operation at 894 days – nearly two and a half years.

On-line refuelling also makes CANDU more flexible to operate than most other reactor designs. For example, if a certain section of the core is running unusually “hot” or “cold”, then special fuel bundles containing more or less-enriched uranium can be inserted into that section to correct the imbalance and promote more homogenous neutron flux and fuel burnup. This process, along with the ability to divert steam away from the generator turbines, also allows CANDU power plants to perform deep load following, adjusting their output to match the ever-changing demands of the electrical grid. Indeed, more advanced designs like the Enhanced CANDU 6 and Advanced CANDU Reactors or ACR-1000 can quickly cycle down to as low as 60% maximum power and back with no ill effects.

Furthermore, CANDU’s ability to burn natural uranium means it can easily burn all sorts of alternative fuels, including the spent fuel from regular light water reactors – a practice known as Direct Use of PWR Fuel in CANDU or DUPIC. Containing around 0.9% uranium-235, spent PWR fuel is rich compared to regular CANDU fuel, allowing 30-40% more energy to be extracted which would otherwise go to waste. This capability is currently being studied by South Korea, which operates a fleet of both PWR and CANDU reactors. CANDUs are also being studied for converting the long-lived actinide elements like plutonium and americium found in spent nuclear fuel into shorter-lived radioactive elements, reducing the amount of time radioactive waste must be stored – and to learn more about the complex science of nuclear waste disposal, please check out our aptly-named video How Does Nuclear Waste Disposal Work?

CANDU can also burn so-called mixed oxide or MOX fuel, composed of uranium mixed with plutonium. This fuel cycle theoretically allows for the peaceful and productive disposal of plutonium from decommissioned nuclear warheads, reducing the risk of nuclear proliferation. However, some critics have argued that the construction of the fuel reprocessing facilities required to support this cycle would actually increase the risk of proliferation by giving civilian agencies the ability to extract and purify plutonium.

And if that weren’t enough, CANDUs are easily adapted into breeder reactors, which convert normally non-fissile or fertile isotopes into fissile ones and can theoretically breed more fuel than they consume. For example, when the fertile isotope Thorium-232 is bombarded with neutrons in a reactor core, it can absorb a neutron to become Thorium 233. This then decays twice via beta decay to become Uranium-233, which, like Uranium-235, is fissile and can be burned in the reactor. This fuel cycle is being actively investigated by several nations including India, which boasts large deposits of thorium ore. However, breeder reactors can also be used to breed plutonium-239 from uranium-238, significantly increasing the risk of nuclear proliferation. Indeed, CANDU-derived reactors have already been used for this purpose – but more on that later.

Further innovation is to be found in the CANDU’s control system. In most nuclear reactors, power output is adjusted using so-called control rods made of neutron-absorbing materials like cadmium or boron, which are moved in and out of the core to control the neutron flux and the rate of the chain reaction. CANDU uses a similar system, only the control rods – which move vertically in the calandria at right angles to the pressure tubes – are known as adjuster rod units or ARUs and are made of stainless steel. However, to allow for even finer control, the CANDU core is divided into 14 zones, each fitted with multiple ion chambers and self-powered vanadium flux detectors to monitor neutron density as well as zone control unit or ZCUs. These are vertical metal tubes which can be filled with varying levels of neutron-absorbing light water, allowing neutron flux and fuel burnup in each zone to be adjusted with a high degree of precision.

But perhaps the greatest strength of the CANDU design is its high degree of safety – among the best of any nuclear reactor design. This is accomplished by a combination of both passive and active safety systems. Actually sustaining a nuclear reaction in natural uranium fuel requires a great deal of precision in the core design, meaning that any disturbance tends to reduce rather than increase the rate of the reaction. This is yet another case in which the CANDU’s unique horizontal pressure tube design proves its ingenuity; if these tubes start to overheat, they will sag under the force of gravity, throwing them out of alignment and causing the nuclear reaction to slow down. The CANDU’s use of heavy water as a moderator is also a major safety feature. Nuclear fission produces large numbers of smaller nuclei known as fission products, most of which are highly radioactive. The decay of these fission products generates a large amount of heat, which is why reactors must be actively cooled for a period of time even after the chain reaction has been shut down. If a reactor suffers a loss of coolant accident, or LOCA, this decay heat can rapidly build up, potentially leading to a core meltdown. The typical response to such an event is to pump water into the core to cool it down, but in a light water reactor this risks making the core critical again and re-starting the chain reaction. However, as the CANDU core is only critical in heavy water, in an emergency it can be cooled with ordinary water without fear of re-starting the reaction.

The CANDU’s unique calandria-and-pressure tube design also allows it to get around a common problem that plagues many other reactor designs: a positive void coefficient. The void coefficient indicates how a reactor’s reactivity is affected by the formation of voids or steam bubbles in the coolant and moderator. If the coefficient is positive, this means that the reactivity will increase, generating more heat and more steam and potentially triggering a runaway chain reaction. Indeed, this was one of the major factors in the 1986 Chernobyl disaster: the control rods, when lowered into the core, created voids in the coolant, causing the reactivity to spike and the coolant to flash-boil, causing a steam explosion that ripped the core apart. In the CANDU core, however, coolant and moderator are separated, with the former having such a small volume compared to the latter that the effect of void formation in the coolant has a relatively minor impact on the overall reactivity of the reactor. Furthermore, the large mass of moderator in the calandria acts as a giant heat sink that can easily absorb any excess heat. The core thus reacts very slowly to power excursions, giving operators ample time to get the reaction back under control.

Another passive safety feature is the mounting of CANDU’s steam generators above the calandria. In the event of a cooling pump failure, this allows decay heat from the fuel to circulate and be carried away via natural convection, preventing the core from overheating and melting down. Each of the coolant loops feeding the steam generators is also divided in two, meaning that any loss of coolant accident likely to effect only one-quarter of the total core.

In addition to these passive features, CANDU reactors are also fitted with a number of active safety features. For example, the Adjuster Rod Units as well as a set of four cadmium Mechanical Control Absorbers can be rapidly lowered into the core to reduce its reactivity. There are also two independent shutdown systems: SDS-1 comprises 28 cadmium shutdown rods suspended over the calandria by electromagnets, such that in the event of a power failure, the rods will automatically drop into the core and shut down the reaction. As these rods are inserted into the low-pressure calandria and not the pressure tubes, the formation of high-pressure steam by a runaway reaction cannot cause them to be ejected from the core – and for more on a now largely-forgotten nuclear disaster where this happened – with gruesome consequences – please check out our previous video The SL-1 Accident – America’s First Fatal Reactor Mishap. Meanwhile, SDS-2 injects neutron-absorbing gadolinium nitrate into the moderator, stopping the reaction dead.

The last major safety system installed at most CANDU power plants is a vacuum building, a large reinforced concrete structure connected to the reactor containment building and maintained at lower than ambient pressure. In the case of a steam explosion, the steam vents into the vacuum building and immediately condenses, preventing it from breaching containment and releasing radioactivity into the environment. Working in combination, all of these various safety systems make a disastrous criticality excursion or core meltdown exceedingly unlikely.

The first pilot-scale CANDU demonstration reactor, the Nuclear Power Demonstration or NPD, first achieved criticality on April 11, 1962, and achieved its full power generation capacity of 22 megawatts on June 28 of that year. This was the first nuclear power to be generated in Canada. NDP proved the viability of the CANDU design concept, and remained in operation as an engineering research reactor for 25 years before being shut down and decommissioned starting in 1987. The first commercial-scale CANDU plant was the Douglas Point Nuclear Generating Station, built on the shores of Lake Huron near Kincardine, Ontario. Douglas Point first achieved criticality on November 15, 1966 and officially entered service on September 26, 1968 with a maximum generating capacity of 206 megawatts of electricity. This was soon followed by the Pickering Nuclear Generating Station on the shores of Lake Ontario, whose first generating unit entered service on July 29, 1971. Seven more units would be added between 1971 and 1986 for a maximum capacity of 14 gigawatts. Next came the Bruce Nuclear Generating Station near Kincardine on Lake Huron, whose 8 units were commissioned between 1977 and 1987 for a total capacity of 22.6 gigawatts; followed on February 1, 1983 by the Point Lepreau Nuclear Generating Station in New Brunswick, with a single 660 megawatt reactor.

Meanwhile,in November 1970 the Gentilly-1 reactor in Bécancour, Quebec first achieved criticality . This was an experimental version of the CANDU intended to reduce the original design’s complexity and cost and make it more attractive for international export. Unique among CANDU reactors, Gentilly-1 had vertically-oriented pressure tubes, allowing the use of a single refuelling machine mounted beneath the calandria, and used regular light water as a moderator and coolant – a design choice which required the use of low-enriched uranium or LEU fuel. The unit was also configured as a Boiling Water Reactor or BWR. As previously mentioned, in a Pressurized Water Reactor or PWR the cooling system is divided into two separate loops: a primary loop that extracts heat from the core and carries it to a heat exchanger or steam generator, and a secondary loop that uses heat transferred from the primary loop to generate steam, which is passed through a turbine connected to a generator to produce electricity. By contrast, a BWR has only a single coolant loop, in which coolant is fed directly into core and turns immediately to steam, which is then fed through a turbine. While simpler and cheaper to build and operate than PWRs, BWRs come with their own set of issues – for example, radioactive contamination of the turbines by fission products from the core. Instrumentation and control of the core is also more complex, resulting in a higher risk of accidents – indeed, BWRs were involved in both the 1979 Three Mile Island and 2011 Fukushima nuclear accidents. Unfortunately, Gentilly-1 was plagued with technical problems and never worked as intended, logging only 180 days of normal operation over 7 years in service. The reactor was shut down in 1979 and a regular 925 megawatt CANDU-6 unit installed beside it, entering service on October 1, 1983. A third unit was planned for the Gentilly site, but this was cancelled when it was decided that Quebec’s future energy needs could be met using hydroelectricity. In 2012, operator Hydro-Quebec decided to shut down Gentilly-2 for economic reasons and initiated a 50-year, $1.8 billion decommissioning process.

The last CANDU plant to be built in Canada was the Darlington Nuclear Generating Station near Clarington, Ontario, whose four reactors, with a combined capacity of 11 gigawatts, were commissioned between 1992 and 1993. Together, the Darlington, Bruce, and Pickering plants generate around 58% of Ontario’s electrical power.

Throughout this period, AECL continued to refine the CANDU design, producing a number of different versions. The original design introduced in 1968 had a rated capacity of 500 megawatts and was designed for use in large multi-unit installations. This was followed in the early 1980s by the CANDU 6, a 600-megawatt design intended for small single or double-unit installations. An even larger 900-megawatt version known as CANDU 9 was later developed, though as of this recording none have actually been constructed. In 1962, construction of an innovative CANDU variant known as Whiteshell Reactor or WR-1 began at AECL’s Whiteshell Laboratories in Pinawa, Manitoba. First achieving criticality on November 1, 1965, WR-1 used a special mixture of organic terphenyl oils called HB-40 instead of heavy water as its primary coolant. This had numerous advantages over the conventional CANDU design; most notably, oil has a much higher boiling point and heat capacity than water, allowing the coolant system to operate at higher pressures and temperatures. This, in turn, increases the thermal efficiency of the entire system, allowing for more energy to be produced using a smaller core. Use of oil also reduced corrosion of the core components, extending the life of the reactor. But while WR-1 ran for 20 years and promised to significantly improve the cost and efficiency of the CANDU system, the design was never successfully commercialized and the experiment was shut down in 1985 for economic reasons.

At the same time as it was building reactors across Canada, AECL was also making significant efforts to market the CANDU overseas, with its first foreign sales success being India. AECL had previously sold India the Canada India Reactor Utility Services or CIRUS research reactor. Based on the NRX reactor at Chalk River Laboratories, CIRUS was installed at the Bhabha Atomic Research Center near Mumbai in 1956. A similar reactor was later sold to Taiwan in 1969, but the country broke off nuclear relations with Canada after the latter officially recognized the People’s Republic of China. In 1966, AECL negotiated a deal to build two reactors in India based on the first Canadian CANDU installation at Douglas Point. The first of these, the Rajasthan Atomic Power Project or RAPP-1, began operation in 1972. However, on May 18, 1974, while RAPP-2 was still under construction, India detonated its first atomic bomb, rather surreally nicknamed “Smiling Buddha”. It soon emerged that the plutonium for the bomb had been produced using the CIRUS research reactor, violating an agreement that the reactor only be used for peaceful purposes. Absurdly, India tried to claim that it had not, in fact, violated the agreement, since Smiling Buddha was a “peaceful nuclear explosion.” Suuure, India… AECL, unconvinced by this argument, pulled out of the RAPP project and ceased all further transfers of nuclear technology to India. Then, in 1976, Canada updated its nuclear export policy, limiting transfers of nuclear technology to nations which had signed the 1968 Non-Proliferation Treaty. As a result of the Canadian withdrawal, completion of RAPP-2 was delayed until 1981. However, India went on to copy the CANDU design to produce its own indigenously-manufactured version called the Indian Pressurized Heavy Water Reactor or IPHWR. As of this recording there are 18 IPHWR reactors in operation across India – and to learn more about the real-life effort to use nuclear weapons peacefully, please check out our previous video That Time the Soviets Tried to Extinguish a Fire With a Nuke for…Reasons.

In 1971, AECL built a 137 megawatt CANDU unit at the Karachi Nuclear Power Plant in Pakistan – the first operational reactor in the Muslim World. This, too, proved controversial, especially following the 1974 Indian nuclear test, since it was feared that Pakistan would use this reactor to develop its own atomic bomb. The Karachi installation was followed in 1983 by a 635 megawatt installation at the Embalse Nuclear Power Station in Argentina. That same year, the first 657 megawatt unit at the Wolsong Nuclear Power Plant in South Korea came online; this would be followed by three more units between 1997 and 1999. In 1996 and 2007, two 706 megawatt units came online at the Cernavodă Nuclear Power Plant in Romania; while between 1994 and 2003 seven units with a total capacity of 4.1 gigawatts were installed at the Quinshan Nuclear Power Plant in China. Many of these deals were highly controversial at the time due to AECL’s questionable business practices. Being a relative newcomer to the global nuclear energy market, the crown corporation faced intense competition from established players like American firms Westinghouse and General Electric, forcing it to court smaller nations lacking the industrial capacity to build more conventional reactor designs – many of which were government by authoritarian regimes. For instance, construction of the Cernavodă plant was begun under the auspices of communist dictator Nicolae Ceaușescu, while the Embalse project was undertaken by the right-wing regime of Isabel Perón and – after the 1976 coup d’état – the military junta led by Lieutenant General Jorge Rafael Videla. It also later emerged that AECL had paid bribes to both the Argentine and South Korean governments, prompting an official investigation by the Royal Canadian Mounted Police. Finally, when construction began on the Quinshan plant, critics questioned the ability of the Chinese to safely operate nuclear reactors due to the Communist Party of China’s lack of transparency and accountability. The sale also occurred shortly after the infamous 1989 Tianamen Square Massacre, leading activists to protest AECL’s dealings with the Chinese government.

But such controversies were the least of AECL’s problems, for despite decades of intensive sales efforts to countries as varied as Australia, Chile, Egypt, France, Greece, Hungary, Italy, Japan, Mexico, the Netherlands, the Philippines, Russia, Thailand, and Yugoslavia, demand for CANDU reactors both at home and abroad quickly fizzled out, with only 41 units were built worldwide between 1968 and 2003. Of these, 24 were installed in Canada – all but two (Gentilly 2 and Point Lepreau) outside of Ontario. Other Canadian provinces showed little interest in the technology, opting instead to use either hydroelectric power or fossil fuels like coal and oil.

So, what happened? Why, despite its numerous advantages in efficiency, flexibility, and safety, did the innovative CANDU fail to catch on more widely both domestically and abroad? While the specific reasons are complex, the overall answer boils down to three main factors: politics, economics, and the inexorable march of technology. The economy of running a power plant is typically measured in terms of Levelled Unit Energy Cost or LUEC, the average cost to produce a kilowatt-hour of electricity over the plant’s lifetime. This figure includes all costs including initial design and construction, maintenance, refuelling, administration, and final decommissioning. Thanks to its on-line refuelling capability and use of cheap natural uranium fuel, CANDU is able to achieve an impressively low LUEC of 3.2 cents per kilowatt hour, compared to 4.1 cents for a coal plant and 6.6 cents for natural gas. However, this advantage is greatly offset by CANDU’s enormous up-front construction costs. This problem is common to all nuclear power plants – with construction costs typically accounting for 65% of overall lifetime cost and refuelling less than one cent per kilowatt in generating costs – but particularly pronounced in CANDU. The low scattering cross section of heavy water plus the low reactivity of natural uranium fuel means more of both are required to achieve criticality. This in turn results in a larger core, calandria, containment building etc. compared to conventional reactor designs, pushing up initial construction costs. This is made even worse by construction delays, which have occurred on numerous occasions. For example, while plans for the Darlington Nuclear Generating Station were approved in 1973, numerous delays caused by – among other things – budget shortfalls, changes in government, the 1973 oil crisis and the 1979 Three Mile Island accident meant that construction did not begin until 1981 and the fourth and final reactor not commissioned until 1993. This caused the total capital cost of the plant to balloon to an eye-watering $11.9 billion – more than double the original estimated budget.

While such high up-front costs are theoretically offset by CANDU’s low long-term operating costs, most politicians only think in terms of four-year election cycles, making the technology unattractive to many potential buyers. Worse still, CANDU becomes most economical when operated at capacity factors above 60%. As previously mentioned, when first introduced, CANDU reactors offered capacity factors far superior to conventional light water reactors. However, over the following decades dramatic improvements in reactor management have allowed LWRs to achieve capacity factors as high as 90%. Meanwhile, the increasing maintenance demands of the aging CANDU fleet plus the relatively poor performance of refurbished units have brought the reactors’ capacity factors down to around 80-90% and their LUECs up to 10 cents per kilowatt-hour, all but eliminating their original economic advantages. Lower up-front costs plus comparable capacity factors and LUEC have thus led most clients to choose conventional LWRs over CANDUs.

CANDU reactors have also historically suffered from numerous design defects – most notably the decision to use zircaloy in the pressure tubes. This alloy is highly susceptible to a type of corrosion called hydrogen embrittlement, which can cause the tubes to crack and rupture over time. Indeed,
such a rupture in the Pickering-2 reactor resulted in a major loss of cooling accident in 1983. Other common problems include corroded feed water pipes, defective steam generators and poor maintenance and safety practices, which have resulted in many CANDUs spending increasing amounts of time offline undergoing repairs and upgrades. Others have been decommissioned as the cost of refurbishment was deemed too high. Indeed, of the 41 CANDUs built around the world, as of this recording 31 are still in operation – 19 in Canada.

Finally, major nuclear accidents like Three Mile Island in 1979 and Chernobyl in 1986 led to major slump in the global nuclear energy market, which severely impacted CANDU’s already limited sales prospects. Nonetheless, AECL continued to upgrade the CANDU design in order to make it more economical and attractive. For example, the 740-megawatt Enhanced CANDU 6 or EC6 maintains around 95% of the original CANDU 6’s design features while incorporating simplified and cheaper construction techniques, advanced control systems, enhanced safety margins, and improved deep load following capabilities.

An even greater departure was the 700 megawatt Advanced CANDU Reactor or ACR-700, which introduced more modularized construction and abandoned the original design’s natural uranium and heavy water system in favour of low-enriched uranium fuel containing 1-2% uranium-235 and pressurized light water coolant. This design not only reduced the cost of heavy water by nearly one-third, but also allowed for a more compact and energy-dense core and greater thermal efficiency., potentially reducing up-front construction costs by up to 40%, increasing overall power output by up to 50%, and significantly reducing nuclear waste production. Yet despite the promise shown by ACR-700 and its 1200 megawatt follow-up, ACR-1000, the design ultimately failed to find any buyers, prompting AECL to shelve the design in 2009. Then, in 2011, AECL sold its reactor division to Candu Energy, a wholly-owned subsidiary of controversial Quebec-based construction firm SNC Lavalin. In the wake of this deal, Candu Energy took over maintenance and refurbishment of existing CANDU plants in Canada, pursued the completion of long-dormant projects in Romania and Argentina, and sought new sales opportunities for the technology in countries like China and Britain, focusing on the ECR-6 design. Meanwhile, back in Canada, great interest has been expressed in using CANDU reactors as steam plants for extracting oil from Alberta’s bitumen sands – a system which promises significant long-term cost savings and reduced carbon dioxide emissions compared to conventional natural gas-fired units. However, the revelation that it would take up to 20 such reactors to meet Alberta’s projected oil production growth, combined with general anti-nuclear sentiment among the Canadian public, resulted in this project being postponed indefinitely.

In today’s climate-conscious world, where nations are increasingly turning to alternative energy sources to help curb greenhouse gas emissions, one would assume that a technology like CANDU would be poised to make a comeback. After all, it has been estimated that over the past 60 years, Canadian nuclear power has prevented the release of more than 900 million tonnes of carbon dioxide and other pollutants into the atmosphere – saving up to 20,000 lives. However, plans for Canadas nuclear future have instead focused on a newer, even more flexible technology: Small Modular Reactors or SMRs. In 2017, in response to an “SMR Roadmap” report cprepared by Natural Resources Canada, Candu Energy developed a compact 300 megawatt reactor called the CANDU SMR. Easily transportable by truck, rail, or even aircraft and more easily and cheaply installed than larger units, these reactors are ideal for serving communities too small and remote to warrant conventional power plants – such as those in Canada’s north. They can also be used to make up for smaller shortfalls in baseload power generation and for various non-standard applications such as bitumen sands extraction, industrial heating, and seawater desalination. Many other firms around the world are also actively developing SMRs, with the global market for this technology being projected at nearly $150 billion between 2025 and 2040. The future of nuclear power, it seems, lies in thinking small.

Despite its lacklustre commercial performance, the CANDU remains one of the safest and most innovative reactor designs in history – an ingenious adaptation to the limited industrial resources available to the fledgling Canadian nuclear industry in the 1950s and 1960s. Yet despite this, due to an unfortunate combination of shifting politics, economics, and technological advances, CANDU failed to achieve its considerable potential, winding up as a dead end in the history of nuclear power. In other words, it was the right technology at the wrong time.