(A shorter version of this article appeared in the Canadian Nuclear Society’s BULLETIN magazine, 2009 June edition)
This article is intended to show how Ontario’s nuclear power plants interact with the grid and how they will be affected by wind generation. Hopefully it will get readers to raise questions about the risk wind poses to the availability of the nuclear units and to the reliability of the grid. There are presently 16 CANDU reactors operating on the Ontario grid with a capacity of around 11,400 megawatts on a grid of about 34,000 megawatts, generating over 50 percent of Ontario’s electricity. Generation of over half of the energy from only about one third of the capacity clearly indicates that nuclear generation is a low cost, preferred component in the Ontario system. In 2008 nuclear and hydro combined met 77 percent of Ontario’s electricity demand cleanly, with no greenhouse gas emissions, with the balance coming from coal and natural gas. The total hydro-electric capacity, including base, intermediate and peak, is about 7,800 megawatts. There are about 950 megawatts of nameplate wind generation connected to the grid at the moment but this is expected to increase to 5,000 megawatts over the next 20 years, or sooner, together with around 12,000 megawatts of gas-fired generation which, with oil, is presently at around 7,500 megawatts. The Ontario government has arbitrarily capped the installed nuclear capacity at 14,000 megawatts. Installed coal-fired capacity is about 6,400 megawatts but coal is to be phased-out by 2014.
In 2008 the four CANDU units at Bruce B had an average capacity factor of 86.5 percent. Capacity factor is the ratio of actual energy generated over a time period to the amount that could be generated at continuous full power operation over the same time period, usually one year. The four units at Darlington had an average capacity factor of 93.9 percent and the four units at Pickering B, 71.1 percent. Provided the demand is there even better nuclear performance can be expected in the future. Darlington’s four units have now started on a three year outage cycle and AECL’s new CANFLEX fuel will result in improved operating and safety margins in all reactors, which becomes more important as pressure tubes age. The CANDU potential is even greater. The four CANDU-6 units in Korea had an average capacity factor of 93.2 percent in 2008 and have a lifetime average also of 93.2 percent up to the end of 2008. However if periods of over generation increase (see later in article) then nuclear performance in Ontario might have to include availability factors as well as capacity factors. Unless otherwise stated this article only refers to the CANDU units in Ontario.
Wind generator output depends on the vagaries of the wind. The Independent Electricity System Operator (IESO) assumes for planning purposes that 10 percent of the installed wind capacity is available at the time of the weekday peak. Of course, it could be zero percent. If the annual capacity factor was 20 percent wind would need 5,000 megawatts of nameplate capacity to give 1,000 megawatts of “firm” generation.
Nuclear plant operating modes
When supplying power to the Ontario grid the CANDU units have two operating modes.
(a) Reactor-following-turbine plant operating mode
If CANDU and the new ACR-1000 (proposed for Ontario new build) units are operating in reactor-following-turbine mode they can contribute to grid frequency stability. In the reactor-following-turbine mode of plant operation the steam generator pressure, which will change due to differences in reactor output and turbine-generator output, is kept at its setpoint by changing the reactor power setpoint, using the reactor regulating system, to accommodate changing turbine steam demands in response to grid conditions. Any difference between supply (generation) and demand (load) on the grid shows up as a grid frequency deviation from the nominal 60 Hz. If a unit is operating at 97.5 percent of full power it can provide +/- 2.5 percent power variation automatically by turbine governor action. For an ACR-1000 this translates to around plus or minus 25 megawatts and means that if grid frequency departs nominal 60 Hz a unit can increase or decrease power up to 25 megawatts to help resist the frequency change in concert with other nuclear, hydro, coal and natural gas-fired units on the grid. The more units contributing to this grid stabilization the less the power variation will be on each unit. The designated hydro or coal plant(s), normally hydro, supplying automatic generation control (AGC) service will then return the grid frequency to nominal by removing the offset. Fast acting AGC corrects the minute to minute differences in generation and load to balance the grid. This is called regulation. The current AGC regulation service requirement from the IESO is for at least plus or minus 100 megawatts at a ramp rate of 50 megawatts per minute but this may be changed to allow other generators to supply this service. The CANDU units are not presently used for AGC. Wind powered generators could result in more perturbations to the grid causing larger and more frequent automatic correcting action by all the units on the grid, including nuclear, depending on the amount of wind generation and the load on the grid, as well as putting more work on the AGC units that are trying to maintain grid frequency.
(b) Turbine-following-reactor plant operating mode
If the nuclear unit is operating in turbine-following-reactor mode (not reactor-following-turbine mode) it makes no contribution to grid stability. In the turbine-following-reactor mode of operation the steam generator pressure is controlled at its setpoint by operation of the turbine governor valve when the reactor power setpoint is changed for any reason. In Ontario the CANDU units are operating in this turbine-following-reactor mode, preferred by operators Bruce Power and Ontario Power Generation, and at the maximum allowable power, normally 100%. The operators say this mode gives more stable reactor operation and increases the probability of the unit remaining connected to the grid during major disturbances, as well, incidentally, as generating more electricity and more income. Grid perturbations brought on by the wind powered generators would not affect the nuclear units in this mode of operation. However, in a future with coal and eventually unsustainable natural gas-fired generation phased out (for many reasons – greenhouse gas emissions pre and post combustion, air pollution, future price increases, price volatility, declining conventional reserves, lost gas legacy for future generations, questionable security of foreign supplies, demands on gas for other uses, for example, in the Alberta tar sands, in the petrochemical industry, for home heating and for export to the United States), the nuclear stations would have to operate in reactor-following-turbine mode to help stabilize the grid, so in this case wind would be a hindrance.
In order to keep the designated plant(s) that is on AGC service in its desired operating range, particularly during the difficult morning ramp-up and the evening ramp-down, other selected units on the grid will be dispatched at frequent intervals to power up or power down so as to allow the plant on AGC service to do its work of balancing the grid. The dispatched units would be load-following hydro (very flexible, very low minimum load level), coal-fired (flexible, low minimum load level) or gas-fired (less flexible, high minimum load level combined cycle gas turbine units) and not normally the nuclear units which would be operating at full power. If wind is ramping down in opposition to a demand which is ramping up this will make the job of balancing the grid more difficult.
Dispatches take into account the technical constraints and economics of the unit being dispatched and are sent at five minute intervals but not necessarily to the same generator unit or load. Lower priced generation will be used ahead of higher priced generation subject to transmission restraints or other reliability related considerations. Although the nuclear units operate baseload they are dispatchable and in the future, with coal and eventually natural gas phased out, load-following nuclear and hydro would have to respond to these dispatches. For example, the new ACR-1000 uses steam bypass in combination with control of reactor power to provide flexible load-following operation and very likely AGC. The steam bypass system on the current CANDU units would not have been designed for daily dispatchable load-following, but can be used less frequently, so they would respond slower to dispatch. However, if necessary, changes to the reactor power setpoint could be made that would follow up on the fast initial response of the hydro units to the dispatch. The downside is that the hydro units could be subjected to dispatch volatility, that is, dispatch instructions and dispatch reversals. For current nuclear units with aged pressure tubes all this is contingent on there being enough safety margin available for a shallow power reduction to be made at high power without exceeding fuel bundle power limits when adjusters are removed. Raising reactor power in a CANDU takes more time than reducing reactor power if it were necessary to withdraw adjusters after the power reduction and this may affect dispatch response.
When a nuclear unit is operated as a load-follower the reactor and its fuel are subjected to frequent changes in power depending on grid requirements and the fatigue life of the fuel may limit the number of power cycles. Nuclear fuel that has been subjected to daily power changes after a couple of years in the reactor must still survive a design basis loss-of-coolant accident without failures even with aged pressure tubes. Too much dispatch volatility, a significant current concern of the grid operator, reduces operating efficiencies and increases maintenance costs, no matter the type of unit. Dispatchable fossil units, and even the hydro-electric units, have a minimum output capability and must remain loaded above this minimum output level or shut down. In addition there may be other constraints on the fossil units like minimum run-time, ramping capability, maximum number of starts per day, and minimum turnaround time, which could mean some generators would have to continue operation overnight.
During periods of surplus baseload generation (SBG), which up to early this year had occurred only a few times a year on the Ontario grid, nuclear plants are dispatched to reduce power. SBG is an over generation condition that occurs when Ontario’s electricity production from baseload facilities such as nuclear, must-run hydro-electric units (must run and not spill for regulatory reasons) and non-dispatchable wind is greater than market demand. In the future, with 14,000 megawatts of nuclear online, more conservation and more efficient use of energy, more embedded generation, and much more wind powered generation, the frequency of SBG events will dramatically increase and so will the number of times that nuclear plants will be asked to load-follow or to reduce power sufficiently enough that hydro can respond to load-following dispatches. This will increase wear and tear on the nuclear units. At the present time the grid operator, the IESO, only considers the curtailment of wind generation when all market mechanisms are exhausted, including nuclear shutdown. Now the grid operator is proposing that curtailment of wind generation is considered if the nuclear units can mitigate the SBG situation only by taking the risk of not being available in future hours when they will be needed, for example, a deep reactor power reduction or a shutdown. This dispatch priority for SBG events is a continuing major concern of the IESO and it is likely the future will see a juggling of dispatch between nuclear manoeuvring/shutdown, wind spilling and water spilling depending on circumstances.
Supplying dispatchable power will be the most challenging duty faced by the present and new nuclear units, as it would be for any thermal plant. Increasing amounts of wind generation will make the job of balancing the grid even more difficult especially during the daily periods of major load changes, and if the wind generation displaces gas-fired generation in periods of low demand the nuclear units would have to respond to load-following dispatches. This will be exacerbated in the future when coal and eventually unsustainable natural gas-fired generation is phased out leaving load-following to nuclear and hydro. Future climate change may affect the capacity of hydro-electric facilities putting more pressure on the nuclear plants. Even now some hydro plants may not be available all the time, there are seasonal fluctuations in water supply, there may be local, provincial or international agreements on water management, or water is being kept in storage for load following or operating reserve. The load-following and operating reserve capability of the hydro plants will become even more valuable after the phase-out of flexible coal-fired generation and should not be squandered to support wind, which would be the job for gas.
Some generators on the grid have to accommodate daily load-cycling since demands are low during the night and high during the day. These are the hydro, coal and gas fuelled plants and they supply this so called intermediate load generation. Overnight some of the coal and gas-fired units would have to turndown to their minimum load or shut down completely depending on system requirements. The CANDU plants operate baseload at full power for economic reasons although in the past some domestic units and off-shore units (CANDU 6) did accumulate considerable good experience with load-cycling, with some deep power reductions, but not on a continuous daily basis. For example back in the 1980s several of the Bruce B units experienced nine months of load-cycling including deep (down to 60 percent full power, or lower) and shallow power reductions. Analytical studies based on results of in-reactor testing at the Chalk River Laboratories showed that the reactor fuel could withstand daily and weekly load-cycling. However, this confidence may not necessarily apply to frequent dispatchable load-following duty even though the new CANFLEX fuel will provide improved margin to failure over the present fuel. It should be remembered that all CANDUs have failed fuel detection and location systems and failed fuel can be replaced while the reactor is at full power. On-line refuelling also results in consistent manoeuvring performance since the amount of excess reactivity does not change much with time.
All CANDUs were designed to be capable of quickly reducing power to 60 percent of full power, holding at reduced power, and then returning more slowly to full power, using their adjuster rods (Bruce A excepted). The specification for the new ACR-1000s states that they are designed to rapidly reduce power from 100 percent steady state down to 75 percent full power overnight and periodically down to 60 percent or even 50 percent on weekends. Use of low-enriched fuel and light water coolant in the ACR-1000 has resulted in a lower xenon load following a power reduction compared to CANDU and this simplifies reactor operation, making the ACR-1000 inherently more responsive so the adjuster rods found in current CANDUs are not necessary.
Bruce B had three weeks of day/night load-cycling in the early spring of this year to help the grid operator cope with an extended SBG situation. Each unit saw 300 megawatt load changes occurring in less than two hours with output reduced to about 64 percent of current maximum electrical output, using steam bypass. The ACR-1000 and the CANDU electrical output can be reduced even more, to around 6 percent full power, just enough to supply the plant’s auxiliary services load, with the reactor held at around 60% full power and steam bypassed around the turbine to the condenser. This mode is normally used if the grid goes down (blackout) since the reactor can remain at 60 percent full power indefinitely until the grid is re-established. In this so called “poison prevent” mode the already hot turbine can be quickly brought up to 60 percent power by gradually closing the steam bypass valves to load the turbine and then the slower return to 100 percent power output can begin. During the 2003 August blackout in Ontario and the north-eastern U.S.some Ontario units at Bruce and Darlington were put in this mode. The Pickering A and B units do not have turbine steam bypass to the condensers and reject steam to atmosphere which restricts their time in the poison prevent mode because of limited demineralized water make-up.
Today CANDUs do not normally cycle down overnight and would not need to cycle down in the future if there were a market for the low cost electricity. This could be export, thermal storage, or supplying the demand for battery charging of Ontario’s electric vehicles. Without coal and gas-fired electricity in the future, hydrogen fuelled gas turbines and fuel cells could help meet the afternoon peak load and provide some of the load following to balance the grid enabling the nuclear units to run at near full power. Hydrogen and compressed air produced overnight could be fired up in power turbines (no air compressor needed) with or without heat recovery. Thus even if practical wind energy storage were available wind still would not be needed on a future all hydro/nuclear grid. Making use of the nuclear plant’s low cost scheduled output overnight makes more sense than using variable, intermittent and expensive wind generation.
Much has been written about the perceived poor flexibility of nuclear and this may be the result of the good job the nuclear industry has done in promoting nuclear as a dependable baseload supply and the not so good job it has done in promoting nuclear load-cycling. The 2005 December 9 Ontario Power Authority (OPA) Supply Mix Advice Report to the Minister of Energy revealed many references to perceived nuclear operational inflexibility.
The IESO still has this pessimistic outlook on nuclear flexibility but at the same time is enthusiastically supporting the Ontario government plan for more wind which will cause even more periods of excess supply during periods of low demand causing the low cost nuclear units to reduce power or load-follow to accommodate high cost wind. However, as described later, Bruce B has demonstrated flexibility by helping the IESO respond to an extended SBG period earlier this year.
Much more recently the IESO has again complained about nuclear manoeuvrability. In its draft “Dispatch Priority” of 2009 Feb. 4 and referring to SBG it states, “Nuclear generation is limited in its manoeuvrability. Often times when asked to move, the nuclear generator will agree for technical reasons, to a specific MW amount which may be much greater than what they were asked to move. For example, the IESO may request a nuclear unit to move 80 MW down. Due to equipment limitations, the nuclear unit may agree to move down but will have to move by 300 MW.” It is not clear if the IESO is saying that this is a fault of Ontario’s “nuclear” in general or of a specific nuclear station that has manoeuvring restrictions. It reads like it could be the result of pressure tube ageing eating into the safety margins so that fuel bundle power limits would be exceeded if adjusters were pulled at relatively high power after a shallow power reduction in response to a SBG dispatch. If so, this would be alleviated by the new fuel and/or refurbishment.
To avoid changing reactor power Bruce Power’s nuclear units currently respond to occasional SBG dispatches by implementing a tightly choreographed turbine steam bypass procedure to reduce unit electrical output. This is appropriate for Bruce A since it does not have adjusters because it originally had boosters that were later removed. There may be other restrictions on Bruce B, either licensing, age related, or both. Of course using steam bypass to reduce electrical output is not as efficient as reducing reactor power and it also causes wear and tear to the bypass system and eats into thermal fatigue life. It increases the risk of condenser tube leaks, possible vibration, and bypass valve trim damage, which must be a concern to the plant operator. There are also thermal emission constraints with the temperature of the cooling water returned to the lake. Darlington, as designed, should be able to respond to a SBG dispatch by reducing reactor power, with no need for steam bypass unless a deep electrical power reduction is required and a return to full power could be expected within a short time period.
CANDU nuclear steam plant is designed to manoeuvre in the upward direction at 4 percent of actual power per second when at powers of between zero and 25 percent of full power and at 1 percent of full power per second when at powers of between 25 and 80 percent of full power. Above 80 percent of full power core boiling restricts the upward manoeuvring rate to 0.15 percent of full power per second. The plant can also quickly respond to a 5 percent step increase in output demand, provided unit is operating at least 5 percent below full power. All this assumes no operational constraints that would limit power increases. The plant power manoeuvring rate is limited by the turbine rather than by the reactor to between 5 and 10 percent of full power per minute but in practice much lower manoeuvring rates are used. Although this information applies specifically to the CANDU 6 it should also apply to Ontario’s latest CANDU units as well.
In the downward direction there is no limit to the rate of station electrical power reduction. The combination of steam bypass and reactor power reduction allows a sudden electrical output reduction of any magnitude, even 100 percent load rejection caused by loss-of-line, to be tolerated without a turbine or reactor trip. The reduced reactor power level is normally limited to 60 percent full power or more (i.e. a 40 percent power reduction), so called “poison prevent” level, to prevent excessive xenon transients. A return to full power can be achieved in less than 3 hours depending on the amount and duration of the power reduction and other conditions, which allows for load-cycling.
The refurbished nuclear units will be operating for at least 25 years and the new nuclear units are going to be operating for at least 60 years after start-up. In the future the operation of the nuclear units will have to change. For example, without coal, and eventually without unsustainable natural gas, nuclear units would have to run at less than full power (even without wind, wind just makes things worse) to provide grid stability, have dispatch capability to load follow to balance the grid, provide AGC service when required, perform daily load cycling, and provide operating reserve. The IESO wants this of all new generation. With nuclear limited to 14,000 megawatts it is preferable that a large proportion of these megawatts be supplied by the new more flexible ACR-1000s. The North American Electric Reliability Corporation (NERC) released it’s Special Report, “Accommodating High Levels of Variable Generation” on 2009 April 16, describing what needs to be done in the future to integrate wind into the grid, including the provision of sufficient flexible support generation.
Without gas wind will be a hindrance, and not a help, to nuclear operation as well as to grid reliability on a nuclear/hydro grid. Running nuclear and baseload hydro units that have high fixed cost and low operating cost at reduced power and varying outputs to support expensive, intermittent and varying wind power makes little economic sense and there are no environmental benefits to having wind on a clean nuclear/hydro grid. In the future the grid may have to contend with less hydro-electric generation because of climate change so intermittent and varying wind generation will have to be balanced even more by the nuclear plants which will negatively effect their performance and economics. High wind generation coupled with low grid demand, a major concern of the grid operator, will put more unnecessary manoeuvring demands on the nuclear units and put the grid at risk.
The 2006 October “Ontario Wind Integration Study” by General Electric International Inc. for the IESO and the OPA, like all similar studies, looked at a grid with significant flexible generation from hydro and from coal and natural gas-fired units to support wind, when it is blowing and when it is not. In the future the Ontario grid may not have coal or gas and wind could impact on the reliable and economical operation of the nuclear stations and hence the reliability of the grid. Even now, with gas-fired generation and a small amount of wind, periods of SBG result in nuclear units being powered down so even higher penetrations of wind, with nameplate capacity of around 5,000 megawatts, would make these events more frequent. As the wind generation displaces the more flexible generation the nuclear units would have to be able to respond to load-following dispatches including those caused by wind variability, resulting in wear and tear. Alternatively nuclear could attempt to reduce electrical output substantially using turbine steam bypass if necessary to allow hydro, or gas if water storage is low or there are other restrictions, to load-follow. Neither approach is good for the nuclear units and there are limits to both. The 5,000 megawatts of wind may come sooner than expected since the OPA, at time of writing, had not then responded to Energy Minister Smitherman’s request of last September to accelerate and increase its renewable and conservation goals in its Integrated Power System Plan.
The future of nuclear load-cycling and even shutdown arrived in the early spring of this year when Bruce B had three weeks of day/night load-cycling with mainly 300 megawatt power reductions on the units to help the IESO cope with a very extended SBG situation. Spring is normally a low demand season but this year it coincided with the downturn in the economy. To avoid having the nuclear units load-follow they took big power reductions overnight so that the hydro units could take care of the overnight load-following dispatches since this particular spring season provided lots of water. Coal was mostly shutdown overnight and some gas units operated at steady power overnight for either technical or contractual reasons. Pickering units 4 and 5 were down for maintenance while Pickering units, 1, 6, 7 and 8 operated throughout at full power. Darlington units 1 and 4 went through one 24 hour load-cycle with output dropping to around 52 percent of full power, a 400 megawatt plus power reduction, and right after unit 3 was shutdown. The shutdown may or may not have been related to the SBG since several days later the other three units were shutdown for scheduled testing and inspection of the vacuum building as required by the nuclear regulator. By the time Bruce B had finished load-cycling unit 5 had accumulated 16 consecutive days of load-cycling out of a total of 22 days of load-cycling and unit 6 had 15 consecutive days out of a total of 16 days of load-cycling. Units 7 and unit 8 had 17 days and 11 days of load-cycling respectively with many of them consecutive before unit 8 took a two month maintenance outage. All together Bruce B had 66 day/night load-cycles over a three week period. Wind was averaging several hundred megawatts over the SBG period. A few thousand more megawatts of wind into the grid (if it were there and injected) might have put other nuclear units into low power operation overnight or even shutdown making them unavailable for up to three days, which would have resulted in more coal being burned during the higher day time demand. Even so, for long SBG periods in the future CANDU unit shutdowns may be a possibility.
Bruce Power has told the IESO that SBG is its “number one operational concern” and that “manoeuvring nuclear units represents a significant reliability risk to the province”. Yet despite Bruce Power admitting that wind will increase SBG events its CEO earlier this year said, when promoting the wind farm it owns, that “wind and nuclear generation naturally complement one another”. Bruce Power owns Huron Wind, a 9 megawatt wind farm in Bruce County.
The design of the CANDUs at each of Ontario’s five nuclear generating stations evolved over time from Pickering A to Darlington and they have different operating characteristics and limitations. They are solid baseload performers and although they were not designed to quickly respond to frequent load-following dispatches, like the new ACR-1000, they should be able to respond to less urgent and less frequent dispatches. CANDUs have the capability to load-cycle by quickly reducing power to 60 percent of full power and then returning more slowly to full power, without using steam bypass. Past operating experience and fuel studies show that this is achievable on a daily basis, though possibly less frequently if steam bypass is involved. For load-following and load-cycling the operating limits are set by fuel safety margin, fuel fatigue life, and the thermal fatigue life of the steam bypass system. Note the recent Bruce B and Darlington load-cycling experiences earlier this year.
Wind will increase the times the nuclear units will be exposed to load-cycling, or even load-following and shutdown, causing them unnecessary wear and tear and decreasing grid reliability. Wind is a negative influence to the Ontario grid and its nuclear units. Natural gas-fired plants are supposed to support the present installed wind generation but even today periods of low demand and negative pricing on the grid have resulted in the manoeuvring of critical nuclear units and this will become more frequent with more wind and even more so in the future when the gas buffer disappears. When gas goes so will all the small gas-fired embedded/distributed/decentralized generation units, putting more load on the centralized nuclear plants. In SBG periods when hydro generation is restricted, gas-fired generation will replace powered-down/shutdown clean nuclear generation that is unwilling to load-follow so in this case more wind would mean more gas will be burned.
The 2005 OPA Supply Mix Summary, that resulted in the expansion of gas and wind to replace coal, said, with the blessing of the nuclear industry, that no single energy source can meet the needs of base, intermediate and peak loads. This premise is flawed. Nuclear is capable of supplying base, intermediate and peak loads. Nuclear hydrogen produced overnight, or any time there is excess generation, would power turbines/fuel cells to help satisfy the day time load including peak loads. This would also enable load-following to balance the grid to be done more by manoeuvring the hydrogen facilities than by manoeuvring the critical nuclear units. Wind and gas would not be needed.
The nuclear industry and the IESO/OPA should take a close look at the co-existence of nuclear, hydro and wind on the future Ontario power grid. More wind, even with natural gas available, will result in unnecessary wear and tear on the valuable nuclear units giving lower unit availability and consequently lower grid reliability. The generation of hydrogen by nuclear units during periods of low demand could smooth out the load on them and provide fuel for the afternoon peak demands, and should be pursued.
Plant and grid operators should get together to clarify the capabilities and limitations of specific current/refurbished CANDUs to load-follow and load-cycle.
The Ontario Energy Board (OEB) should take a critical look at the viability of grid wind power in Ontario together with its reliance on unsustainable natural gas-fired electrical generation to replace coal. This recommendation may be in direct conflict with the recent mandate given to the OEB by the government under its new Green Energy and Green Economy Act, 2009, which, amongst other things, directs the OEB to speed up the installation of wind generators.
(A shorter version of this article appeared in the Canadian Nuclear Society’s BULLETIN magazine, 2009 June edition)