Is nuclear power flexible, does it have load following capability?

Some Notes on the Flexibility of Nuclear Units

 What is “Flexibility”?

 What does “flexibility” mean for a generating technology?  I guess that most people would think about the ability to increase or decrease active power output (Megawatts) either on instruction, or automatically, in order to help balance generation and demand on the system.  But the issues are “How much?”; “How quickly?”; “With how much notice?”; and “How often?”.  There are also other kinds of flexibility.

 Of course all nuclear units are able to cycle between full load, no load and full load again since they do that every time they shut down for maintenance or refuelling.  But that amount of flexibility may not be of any practical use to the grid operator in controlling the system if it can only be done very slowly and infrequently.

 The sorts of megawatt-flexibility that we could consider, each of use in their own way include:

  • The ability to reduce and increase output in a planned way – i.e. planned by the operator of the nuclear plant. (e.g. gradually ramping output down in the late evening and ramping up again in the early morning, sometimes called 12-3-6-3 operation).  The operator of a nuclear unit may choose to operate in this way.
  • The ability to reduce or increase output on instruction/request from the grid operator.  In Great Britain, under the Grid Code, this could mean starting to respond within 2 minutes of an instruction, and achieving a significant change in output within 10-20 minutes.  Sometimes greater notice is given.  Changing output on instruction from the grid operator in this way is likely to be a paid-for service.  In GB the payment would be based on the bid/offer prices declared by the operator of the generating unit.
  • The ability to reduce or increase output automatically in response to a control signal from the national or regional grid control centre (known as “automatic generation control”).   This is not used in GB, although it is used in France and Germany.  Changing output in this way is likely to be a paid-for service.
  • The ability to reduce or increase output automatically in response to changes in system frequency – sometimes called automatic frequency control (“AFC”), or automatic frequency responsive operation, (“AFRO”).  In Great Britain, National Grid says a critical issue is how much change in output can be achieved within 10 seconds and then sustained.  In larger networks the critical time is longer – in mainland Europe or North America, it is 30 seconds.  In smaller networks the critical time could be shorter (e.g. about 5 seconds in Ireland).   In deregulated markets, generators are usually paid when they provide AFRO.

 But in addition to the above there is the ability to control voltage, which means to be able to change reactive power output either on instruction or automatically.  In this area, the nuclear units in Great Britain are at least as flexible as all the large fossil units, and there is no significant problem in ensuring similar or better capability on new nuclear units. And to the grid operator nuclear units are probably more valuable than fossil units in this respect, because they are operated base-load, and hence “always there”, outside of reactor outages, to help control voltage.

 There is also the related ability of generating units to remain in operation at reasonably steady full power during events of abnormal frequency and voltage, or during the disturbances caused by faults on the grid system (i.e. not reducing output or tripping off if there is a transient fall in frequency or voltage).  The nuclear units in Great Britain have proved quite good in this respect; by contrast some early CCGTs and wind turbines were quite poor.   The nuclear units demonstrated their value in this area during the unusual frequency dip in Great Britain on 27 May 2008 – by staying at approximately constant load throughout the frequency dip, they helped stabilise the system by effectively providing a couple of hundred megawatts more generation than National Grid was expecting.  (The Grid Code rules allow output to fall pro rata with falling frequency below 49.5 Hz, but the output of the nuclear units did not fall.)  During the same transient quite a lot of small generators (including some wind farms) tripped off on low frequency.

 Can/Should Nuclear Units Offer Flexibility?

 Returning to “flexibility” for active power, I suggest that there are 4 questions that we could pose in this area:

 1.   Is it possible? I.e. is the nuclear plant physically capable of operating in the desired flexible mode?

       Or perhaps a better question might be

1A. Would it be possible at a price?  I.e. would the nuclear plant be able to operate in the desired flexible mode if we spent more money on it and changed its design a bit?

 2.    Is it permissible?  That is, even if the nuclear plant has the physical capability to change output etc, can the plant operator make a suitably robust safety case to support this mode of operation, and will the nuclear regulator in the country concerned permit this mode of operation?  (Don’t assume this is easy, or that all nuclear regulators would adopt the same position.  For example the NRC in the USA does not permit operation in AFRO mode or automatic load following. Neither does the Japanese regulator.  The French and Belgian regulators do permit AFRO mode, and the French and German regulators permit automatic load following – see below.)

 3.    Is it necessary for the nuclear plant to have the capability and to operate in this flexible mode? 

 4.    Is it commercially sensible for the nuclear plant to have the capability and to operate flexibly?

 Dealing with these in reverse order, most people understand that nuclear units have high fixed costs and low variable costs (low fuel costs) when compared with fossil units (which have lower fixed costs but much higher variable costs).  Hence in any system that has a mix of nuclear and fossil units, the commercially sensible arrangement, which minimises fuel costs, is for the nuclear units to run at steady full output as much as they can, and for the fossil units to provide AFRO, do load following, and reduce output at night, at weekends, and during lower demand seasons in the year.  This would be true even if the nuclear units have the capability to be very flexible.

 In systems that have a large percentage of hydroelectric plant (high capital cost but very low variable costs) such as Sweden, it generally makes sense for the stored-water hydro units to do the load following and provide AFRO, since they usually have a limited amount of water and can’t run at full output all the time.  Hence the operators of the hydro units will prefer to use their limited supply of water to operate mainly at periods of higher demand (higher price), and reduce output at lower demand (lower price) periods.  In addition, many hydro units are more efficient at 80-85% load than at full load, so it is a benefit for them, when they are generating, to operate at less than full load and provide AFRO if they are paid for it.  Hence it is a win-win for the hydro units to provide AFRO and do load following while nuclear units provide base load. 

 Moving on to question 3 above, if the percentage of nuclear units on the system is not too high, then the other units can provide all the flexibility that is needed, so nuclear units don’t need to, and this makes commercial sense too. (See previous paragraph). 

 Is it “Fair” for Nuclear Units to be Inflexible?

 I have heard it said that it is “unfair” to other generating units for nuclear units to be “allowed” to operate at steady base load and not provide load following or frequency response services.  Whether this is unfair or not rather depends on the commercial arrangements in the electricity market concerned, and whether generating units operating flexibly are paid suitably for this service, to compensate them for any increased costs.

 The market arrangement we had in England and Wales from 1990-2001 (the Electricity Pool) did not specifically reward power stations for load following, and the payment for providing frequency control was decided and administered by National Grid, so did not necessarily properly reward the generators who provided this service.  Under those circumstances, there is an argument that the inflexible nuclear units may have had a slight commercial advantage, although in fact any generating unit was permitted to be inflexible if its operators wished, which was achieved by setting an “inflexibility marker”.   In practice I think the operators of flexible generating units (coal, oil, and later, gas) were able via appropriate bidding strategies to obtain the income that they wanted to profitably cover their costs of operating flexibly.

 Under the present GB market arrangements (NETA/BETTA), generating units have the right to operate as they wish, and can bid whatever price they wish to deviate from their plans (i.e. to provide load following on instruction), and can nominate their dynamic parameters (e.g. how fast they can change output).  Similarly, the provision of frequency control is now a commercially competitive service, where generators bid the price they want for providing it.  National Grid, acting as system operator, attempts to minimise operating costs by choosing the most economic providers of load following and frequency response services.  Under this arrangement, any generating unit (whether nuclear or not) that cannot operate flexibly, or chooses not to, is foregoing a potential income opportunity.  Hence it can no longer be argued “unfair” for nuclear units to operate inflexibly.

 I don’t know, so I can’t comment on the market arrangements in other countries.

 Limitations to Flexible Operation

 Because in many countries it is neither necessary nor commercially sensible for nuclear units to operate flexibly, not much effort has been expended in those countries on proving or improving the flexibility of existing nuclear units, or on convincing the nuclear regulators that operating flexibly is acceptable.  In any case there are number of physical limitations to flexible operation:

(i)      Xenon Poisoning: Some of the fission products produced during operation of the reactor are neutron absorbers, the main one being an isotope of xenon.  In operation at steady power the concentration of the xenon reaches an equilibrium, but if the reactor power is reduced this equilibrium is disturbed and the xenon concentration increases, increasing the absorption of neutrons, and hence tending to further reduce the power.  The build-up of xenon has to be balanced by adjustment of the control rods or other neutron control mechanisms, but depends on there being sufficient spare reactivity in the reactor core and sufficient control range on the control rods etc.  Reactors fuelled with natural uranium, like the Magnox reactors in Great Britain, have very little spare reactivity, so a rapid and prolonged load reduction could eventually lead to the reactor shutting itself down by build-up of the xenon. Hence for a Magnox reactor, load following like the 12-3-6-3 cycle described above would be almost impossible.  This limitation does not apply to anything like the same extent to reactors using enriched fuel such as pressurised water reactors (PWRs), except perhaps at the very end of their fuel cycle (i.e. shortly before they shut down for refuelling).

(ii)    Temperature and Pressure limits.  The safety cases and operating rules for all nuclear reactors (which have to be approved by the nuclear regulator) generally have strict limits for temperatures and pressures, their ranges of variation and their rates of change, and the number of acceptable cycles.  These limits potentially place limits on how much, how quickly and how often the power output of a reactor can be changed.   The limits are different for different reactor designs  (It should be noted that there are limits on temperatures and rates of change for other forms of generation such as gas-fired gas turbines that also limit performance, but these are generally manufacturers’ recommendations rather than regulatory rules.). 

(iii)   New Fuel Limits: PWRs and BWRs are refuelled during planned shutdowns (typically once every 18 months) during which time around on third of the fuel in the core is replaced with new fuel.  This new fuel has to be “conditioned” when the reactor returns to power, which means that in the first few days after return to service, the reactor power has to be kept steady and increased slowly, so the reactor cannot operate flexibly during this period of a few days at the beginning of the fuel cycle.

(iv)  End of Cycle limits:  For PWRs and BWRs which are refuelled during reactor shutdowns, the reactivity margin (the amount of spare reactivity in the reactor core) is reduced at the end of the fuel cycle  (i.e. in the last couple of weeks before reactor shutdown, also known as the coast-down period.), so the margin for control is reduced, and the reactor is less able to operate flexibly.  This relates partly to the phenomenon of Xenon poisoning mentioned above.

(v)    Other Limits: Some reactor types require certain on-line instrument checks or recalibrations etc from time to time during normal operations.  Such checks require the reactor to be operated at steady load for a few hours, so the reactor operator would not want to operate flexibly during those short periods.    

 Do Nuclear Units Do it?

 Here in Great Britain, the Magnox and AGR units were built with the assumption of base-load operation (i.e. steady full load operation, and not load following etc).  We have not done anything to change this for the Magnox units.  BE did some work on the AGR units, and have an agreed safety case that allows them to do a limited number of load reductions each year on instruction from the grid operator.  Some years ago, BE tried to make a safety case for operation of AGR units in AFRO mode, but gave up.  As an alternative, BE started to install a controlled steam dumping system on one unit at Heysham 2, which could have provided very rapid AFRO if needed.  In the end it was not commissioned and put in service because the threat that it might be needed went away.  When Sizewell B was commissioned (post privatisation) it was required under the Connection Conditions in the Grid Code to demonstrate the ability to operate in AFRO mode, although the amount required was not specified in the Grid Code at that time, and Sizewell B’s capability is quite limited.  (The amount of capability seems to be commercially confidential, but someone in National Grid told me it was greater than some early CCGTs!).  In practice Sizewell B has rarely been asked to operate in this way. (See reference [2])

 When the percentage of nuclear units on the system is large, then it will be necessary for the nuclear units to do load following and to provide AFRO.  In both France and Belgium, which have a large percentage of nuclear generation and are required by UCTE/ENTSO-E rules to provide a certain level of frequency response nationally, they both do this by operating most of their nuclear units in AFRO mode, pulled back from full output by a couple of percent and each providing a small amount of response. (For some evidence see the footnote on page 53 of Reference [1], and Reference [2]).  So, to provide frequency response from their nuclear units, the French and Belgians need to sacrifice at most a couple of percent of load factor, which by itself is not a show-stopper commercially.

 For the French and Belgian units, load following to reduce output at night, at weekends or on public holidays, is primarily a commercial decision whether to reduce output instead of exporting the excess from the country at a low price, or using it for pumped storage.  Load following in France is discussed in Reference [2] and in the section on “Load Following” in the page about nuclear power in France on the web-site of the World Nuclear Association (Reference [3])

 Around 2004, I met the plant manager at one of the German nuclear units, who was proud of the fact that his company had negotiated a commercially beneficial contract with the grid operator to reduce output on instruction occasionally at night or weekends if it was very windy at the time.  (This is because there are lots of wind turbines in certain parts of Germany, and the wind turbines have “must run” rights.  “Commercially beneficial” means his plant earned more money by doing such load drops, than by remaining at full output.  I was not able to find out from him how much they reduced output or how often or how quickly.). 

 In August 2010, I saw a presentation at an IAEA meeting in Vienna by two gentlemen working for the German nuclear regulator, which described a system now in operation on a number of German nuclear units, where at times they load follow automatically in response to a signal from the grid control centre, mainly to reduce output at times when there is a lot of wind at the same time as low demand (Reference [4]).  There was a similar presentation at a workshop in October 2010 organised by WANO (World Association of Nuclear Operators), by a lady engineer from one of the German nuclear utilities (Reference [5]).   I was not able to find out if (or how much) the nuclear units are paid or compensated for providing this service.

 Some More Sources of Information

 There are a couple of elderly documents published by the IAEA (the nuclear agency of the UN) on the interaction of nuclear units and the electrical grid, that discuss flexibility a little (References [6] and [7])

 CIGRE (the International Council on Large Electric Systems) carried out a survey of flexible operation of nuclear plants in the 1980’s.  The report on this (reference [8]) was written by my then line-manager in the CEGB in 1986.

 There is a good summary in a paper entitled “Can Nuclear Power Be Flexible?” written by Pouret and Nuttall of the Energy Policy Research Group in Cambridge  (Reference [9]), partly based on Pouret’s MSc dissertation.  There is a later paper with the same title by the same authors in the Journal of the Nuclear Institute in Great Britain (Reference [10])

 The flexibility of the proposed new EPR reactor design for Great Britain is summarised in section 5.1 of subchapter 1.2 of the pre-construction safety report (reference [11])

It seems it can be load-cycled between 25% and 100% of nominal full power.

 Similarly the flexibility of the proposed AP1000 reactor design for Great Britain is summarised in paragraph of section 1.2 of Chapter 1 of the AP1000 European Design Control Document (EDCD)  (Reference [12]).  It seems the AP1000 can also be load cycled between 25% and 100%, and can do so on a daily basis for 90% of the fuel cycle.

 Hence it appears that the proposed EPR and AP1000 designs have quite a good capability for flexible operation as part of the basic design and the issues/problems discussed in the older documents seem to have been largely overcome.


 AFC           Automatic Frequency Control

AFRO         Automatic Frequency Responsive Operation (same as AFC)

AGC           Automatic Generation Control

AGR           Advanced Gas-cooled Reactor (unique to GB, operated by British Energy/EdF)

BWR          Boiling Water Reactor  (Second most common reactor type world-wide)

IAEA          International Atomic Energy Agency of the U, based in Vienna

Magnox       The reactor type of the first family of nuclear power stations in the UK (mostly shut down now)

NEA           Nuclear Energy Agency of the OECD, based in Paris

PWR           Pressurised Water Reactor (the most common reactor type world-wide)

WANO       World Association of Nuclear Operators (a voluntary association of all the operating commercial nuclear plants in the world, devoted to “excellence in operations”)

WNA          World Nuclear Association (the commercial association of nuclear operators, fuel suppliers and contractors to the nuclear industry)

 References etc

  1. “Defence in Depth of Electrical Systems and Grid Interaction” final report of the working group on “Defence in Depth of Electrical Systems”, Nuclear Energy Agency of the OECD, Reference NEA/CSNI/R (2009) 10, November 2009. [Available free of charge from]
  2. “Load following Capability of New Nuclear Power Plants”, John Morris, EDF Energy Presentation to the Nuclear Institute, Glasgow, 21 September 2010.  [This is supposed to be available soon from the Nuclear Institute website.]
  3. “France” page in World Nuclear Association Website []
  4. “Load-follow Operation of German Nuclear Power Plants – Survey of Plant Operating Restrictions and Experience”, F. Seidel, S. A. Meiss, J. Hutter, M. Krauss, M. Schneider, Federal Office for Radiation Protection, Germany,  Presentation at IAEA Technical Meeting on “Interfacing Nuclear Power Plants with the Electric Grid: the Need of Reliability amid Complexity”, Vienna, August 2010. [Not currently available on the Internet]
  5. “Wind Power Impact on the E.ON Nuclear Fleet in Germany” Adriana Cortés, Eon Kernkraft, Presentation at WANO Workshop “Living with Wind power” Birmingham, October 2010. [Available for WANO members from WANO private website; probably not publicly available]
  6. “Interaction of Grid Characteristics with Design and Performance of Nuclear Power Plants – A Guidebook”, IAEA Technical Reports Series TRS-224 (1983) [available free of charge from]
  7. “Introducing Nuclear Power Plants into Electrical Power Systems of Limited Capacity: Problems and Remedial Measures” IAEA Technical Reports Series TRS-271 (1987) [available free of charge from]
  8. “Nuclear power plant performance in power system control”, F L Carvalho, CIGRE Reference 39-14_1986 (1986) [Available from,  for 50 Euros.]
  9.   “Can Nuclear Power Be Flexible?”  L Pouret and W J Nuttall, Cambridge University Energy Policy Research Group, Paper EPRG0710, [Available free of charge from:]
  10. “Can Nuclear Power Be Flexible?” L Pouret and W J Nuttall,  Nuclear Future (journal of the Nuclear Institute in Great Britain), Volume 5, Number 6, November/December 2009. [I think this journal paper is not available on the internet.]
  11. EPR reactor pre-construction safety report, Subchapter 1.2  [Available free of charge from  1 – Introduction and General Description/Sub-Chapter 1.2 – General Description of the Unit.pdf]
  12. AP1000 European Design Control Document (EDCD) , Chapter 1, Section 1.2. Westinghouse Inc  [Available free of charge from DCD EPS-GW-GL-700 Rev 1_Public/EPS-GW-GL-700 Rev 1 Chapter 1/EPS-GW-GL-700-Rev 1 Chapter 1 Section 1-2.pdf]



Dave Ward,

dmward (the synbol that says “at” removed to stop spammers)

Last updated 5 February 2011

Yes a good article. But I thought that although the EPR could ramp-up from 25% to 100% capacity, at 5% per minute, of its maximum output (i.e 80 MW per minute) e.g. from 400 to 1,600 MW in 15 minutes, it could only do that 100 times per year e.g. once every three days. Not much use for balancing regular wind variations, but maybe OK for long lulls in wind, if the nuclear operators will accept running at lower power (and loosing money) when there is wind available.  Note also that nuclear plants, like all power plants, have to be backed up: for example a 300 MW  light oil fired gas turbine plant is being built in Finland to back up the new, much delayed,  Olkiluoto nuclear power plant.
Dave E

5 comments on “Is nuclear power flexible, does it have load following capability?

  1. Dave – thank you for a very informative and comprehensive article about a technically complex subject. You did a great job explaining why flexibility is a matter of design choices and the market structure. If there is a financial reward for being flexible, nuclear plants can be designed that way. If they are already built, the choices are more limited, but can still be implemented in some circumstances.

    Whenever people tell me that nuclear fission is inherently inflexible, I usually smile and ask them if they think that aircraft carriers and submarines operate with power plants that are either at 100% or zero. Unfortunately, I am not allowed to describe the technical details of just how flexible they are or how they achieve that flexibility. The fact remains that fission itself is actually more responsive than combustion.

  2. The real motive for designing nuclear plants to be flexible is that as deployment goes up towards 75% of maximum requirements so the plants must follow the nightly drops of 30-40% and seasonal variations. This is especially necessary in countries where 90% of requirements can be met by 3-4 reactors. Daily load fluctuations would be much reduced in an all electric economy when charging of batteries and storage heaters and more intensive night operations of electric freight trains and other night shift operations would happen, when users wanted the power, and thereby reduce the net fluctuations. The flexibility of EPR and AP1000 designs is clearly sufficient for all these situations. A requirement that all other forms of power should leap up and down to suit an erratic source is the wrong approach for Wind power.

  3. Indeed, a good article, setting out fairly and clearly issues that few understand, and many wish to avoid understanding.
    I understand that in Finland the body that is paying for the nuke (or at least buying its electricity) is an industrial consortium who operate a range of power intensive factories, such as paper making. Part of the deal is that, if the nuke fails, so do these factories, and quickly, with high speed automatic trips being set up. Otherwise the Finnish part of the Nordic Grid would have to have a lot more short term reserves.
    With paper making, the stopping of a mill is expensive, as much of the paper passing through the mill would be harmed and it can take quite a lot of paper on startup before the quality is restored. Again, like nukes, the impact can be mitigated by design and control systems, and I am sure they expect tripping of the nuke to be rare.
    The “backup” is part of the deal that is rarely discussed.

  4. Thanks to Rod for a very sensible comment – I would just comment that nuclear submarines are subject to very different regulatory arrangements from commercial nuclear power plants, so they can perhaps do things that would not be permitted on dry land!

    An Update:

    John Morris’s presentation (Reference 2 above) is now available from

    French nuclear units have actually been load following and providing frequency control since the early 1980’s, when their percentage of nuclear already exceeded 50%. A good paper summarising their early experience is “The adaptation of nuclear reactors to the needs of networks” , by G B Giesdon, C Martin, & C Miossec, (In French),in Revue de l’Energie No 372, March 1985, [I think not available via internet].

    There is now a public paper on the German’s use of nuclear to balance wind variations – “Load cycling capabilities of German Nuclear Power Plants”, by Holger Ludwig, Tatiana Salnikova, Andrew Stockman and Ulrich Waas, in International Journal for Nuclear Power, Volume 55 (2010), Issue 8/9 August/September [available from


    David Ward

  5. Dave – thank you for a very informative article about a very complex subject. You did a great job explaining why choices and the market direction is so important in today’s market. If we can create cleaner and safer sources of energy, we can have safer cheaper power at a lower cost for our nation.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.