The cost of the new reactors for Canada

What will the new reactors proposed for Canada cost? That’s very much of a “how long is a string” type of question but I’ll try to give an answer.

The joke used to be that one should multiply by pi any cost estimate given by project proponents.  This just reflects the time honoured tactic to low ball projects to get them approved.

Increases in the prices of raw materials such as steel (40% higher than last year), concrete (50% increase in last two years) and copper (a factor of four in five years) are driving up the costs of all power plants.  

Reactor vendors in sales mode prefer to talk about two numbers namely the price of the electricity their machine will produce and the capital cost per kW (electrical) both are considered more palatable to buyers than the multi-billion dollar price tags that come up at contract signing. 

Personally I find the projected electricity cost approach to be particularly misleading since it depends on a great many factors including the capital cost and others outside the control of the reactor vendors.  Concepts such as Levelized Unit Electricity Cost (LEUC) may be useful for comparing various energy options but for a nuclear station the capital cost dominates.

The unit of dollars per kW (e) can be somewhat more helpful since they it is a way of expressing capital costs especially in this case since the proposed AREVA unit has higher power than its two rivals.  However, there are fixed costs in reactor construction such as environmental assessments, site preparation, licensing, fuel storage, security arrangements and others that don’t necessarily scale with reactor power. It’s better just to look at the prices for reactors already sold.   

AREVA is already constructing two EPR reactors. The EPR reactor at Olkiluoto, Finland is already 25-50% ($1.4-$2.8 billion) over budget – its initial cost was $5.5 billion (3.7 billion euro).  This over run was severe enough to cause serious financial problems for AREVA particularly in 2006. The price for the second EPR at Flamanville, Normandy was $5 billion (3.3 billion). AREVA has also sold two EPR’s to be installed at Taishan, China for $12 billion (8 billion euro). Therefore, the cost of an EPR is in the $5 to 6 billion range, probably nearer the upper number in the range.     

Westinghouse just sold two AP1000 reactors to South Carolina for $9.8 billion ($4.9 billion each). Two more have been sold to Georgia Power at (I assume) a similar price. Westinghouse has also sold AP1000 technology to China to build four reactors for $5.3 billion total but this is not a contract to construct complete reactors. The AP1000 price looks to be about $5 billion.     

In contrast to AREVA and Westinghouse who are constructing their reactor types, the design of the ACR-1000 is not yet completed and the earliest construction start is estimated to be 2012 even if a firm sale were made tomorrow. Thus, not only will AECL be six years behind AREVA and four years behind Westinghouse before starting construction of its first ACR-1000 but also its price can only be estimated on an incomplete design basis. AECL may have some leeway in setting a price, for example, in the price assigned to the heavy water that’s been in storage at LaPrade for up to two decades.  One can argue about how good their estimates might be but the price of an ACR-1000 is determined by a sale with the real cost only known at the end of construction.

Generally, I feel it’s unlikely that the ACR will cost significantly less than its competitors especially since the balance of plant will be roughly similar for all three reactor types. Therefore, I would guess around $5 billion would be a reasonable ballpark estimate. 

To be realistic we also need to consider the first-of-a-kind costs involved in solving the teething problems of new reactor types, including the first environmental assessment, licensing and construction. AREVA has been going through this process with its first two EPR’s and has experienced 25-50% cost overruns particularly on the Finland project. Therefore, it’s fair to estimate similar “learning” costs for the first AP1000 and the first ACR-1000, meaning costs for the first one of either type in the order of $6 to $7 billion or more can be expected.   The learning process for the AP1000 will occur in the four Chinese and four US reactors Westinghouse has already sold.  However, the first reactor of the ACR type will be constructed in Canada. Perhaps the first, even the first two, will be built in New Brunswick. This is a small province which I hope will not have to bear all the learning costs on its own. 

As I’ve stated before, I’m strongly in favour of nuclear power but hyping it by using unreasonably low costs is asking for future trouble.  We need nuclear electricity but we should reconcile ourselves to accepting that it won’t be cheap.

Void Reactivity – a key issue for the ACR-1000

An important issue bearing on the safety of today’s generation of CANDU reactors concerns the void reactivity as expressed by the CVR (Coefficient of Void Reactivity).

In light water reactors (LWR’s), most of the world’s operating reactors, the formation of coolant bubbles (voids) around the fuel can occur for example when the coolant gets too hot. This causes the nuclear chain reaction to shut down very rapidly.  In effect, this is a negative feedback mechanism which is good because it achieves the desired effect of stopping the reaction in an anomalous situation. Negative is good in this context. This negative void reactivity is an inherent  safety feature of light water reactors (LWR’s) because it comes from the basic physics of the reactor in contrast to an engineered system.

In CANDU reactors, the void reactivity is positive. In other words the chain reaction increases when bubbles form and the potential for a reactor runaway or Loss of Regulation Accident is present.  A positive CVR is often associated with pressure tube reactors including the Soviet RBMK design that failed catastrophically in the Chernobyl accident of 1986. Combustion of the carbon moderator of that reactor was a major factor in the severity of the accident, whereas existing CANDU reactors have heavy water coolant.

The calculation of the void reactivity can be very complicated since for instance the CVR can change as the fuel is consumed. For a full technical explanation of the complex issues involved in the void reactivity and its role in CANDU and in the Chernobyl accident please refer to the book by Daniel Rozon, Introduction to Nuclear Kinetics, Polytechnique International Press, Montreal, 1998.

A positive void reactivity has always been a concern for CANDU’s. In today’s CANDU reactors the positive void reactivity is compensated by having two completely independent shut down systems so that the probability of a reactor runaway is acceptably low. This has so far proven to be an effective solution to the problem. In fact the CNSC’s current draft regulations state that reactors built in Canada must have the two shutdown systems. This incidentally would be a great difficulty for licensing LWR’s in Canada since they have only one shutdown system. However, their negative CVR might be counted as a second “shutdown system”. More about regulatory problems in future posts.

AECL’s design for the ACR-1000 predicts a negative CVR and AECL must convincingly demonstrate that this will in fact be achieved. The sign of the CVR for its MAPLE isotope production reactors was a continual bone of contention between AECL and the CNSC. AECL had designed MAPLE for a negative void reactivity but the reactors as built showed a slightly positive one. Intense efforts at AECL and at two US laboratories have not identified how this came about in the design computer codes.

Unless the void reactivity of MAPLE is fully and satisfactorily explained, AECL may have great difficulty in convincing the CNSC that the calculated CVR of its ACR-1000 is indeed negative. The MAPLE reactors are now abandoned but I would hope that AECL would continue to research the problems calculating the MAPLE void reactivity. To do otherwise would be to incur a large credibility deficit for the ACR-1000.

Where will Ontario’s new Reactors be built – Darlington or Bruce?

Location is not only important in real estate but also in the reactor business.

To start by pointing out the obvious, the Bruce site is leased by Bruce Power, a private company and the Darlington site is operated by Ontario Power Generation OPG), a provincial government organization. Both sites are owned by OPG.

It’s clear that the best location for the first new reactors would be at an existing nuclear site. Pickering is probably too small to comfortably accommodate new reactors and that leaves Bruce or Darlington. Both have vigorous local boosters promoting the large economic benefits sure to flow to the surrounding communities. Darlington is located closer to the GTA and there are potential problems building more transmission lines from Bruce to Toronto. However, these should not be the decisive issues.

The more fundamental question is where would the construction of the new reactors be more successful?

On the basis of their recent performance in refurbishing existing CANDU reactors, Bruce Power has been the more successful organization. The cost overruns in returning Pickering units 4 and 1 to service and the consequent permanent shut down of units 2 and 3 did not inspire confidence in the project management capability of OPG. On the other hand, Bruce Power was able to return Bruce units 3 and 4 to service relatively quickly although without re-tubing and its multi-billion dollar project to bring back units 1 and 2 appears to be more or less on track with so far only moderate cost overruns.

The peremptory shutdown of seven CANDU units (Bruce unit 2 was shut down earlier) by Ontario Hydro in 1997 was the definitive end point in what had started out as a technological love affair between Hydro and AECL’s CANDU reactor. It was the culmination of a long period of increasing recriminations between the two companies. The reason for the abysmal performance of these reactors is still not fully explained but I’ll take a shot at it in a forthcoming post. Suffice it to say for the purposes of this discussion that the culture at OPG still tends to blame what they regard as the high maintenance CANDU design as the root of their problems whereas AECL types vehemently defend their reactor and attribute the problems to managerial incompetence at Ontario Hydro.

The point of this much over-simplified history is that Darlington is unlikely to be a hospitable site for building ACR’s in spite of the all-is-forgiven noises OPG and AECL executives emit from time to time. This would not be the case at Bruce Power where my guess is that an attractive deal would trump any history.

Let’s suppose for the moment that Darlington was selected as site for the new build reactors. Would this mean that Ontario had essentially chosen to build Areva EPR’s or Westinghouse AP1000’s in preference to ACR’s? Certainly Areva or Westinghouse could come in and build turn-key light water reactors and then train the personnel to operate them which would probably be OPG’s preference. However, if the business terms were right, Bruce Power could lease part of the Darlington site and collaborate with AECL in building and operating ACR’s or other CANDU’s.  

We’ll just have to wait to learn how Ontario views the relationship between site selection and reactor selection.  To date their policy has been to deny any relationship between the two as evidenced by their concept of the generic environmental assessment as a way to speed up reactor approvals. 

As H.L. Menken said “For every complex problem, there is a solution that is simple, neat and wrong.”