40 Comments

  1. Does this news effectively kill the AP1000’s future in the US? If so, what a monumental waste of the DOE and private sector resources. Guess the SMR is our only hope now, because there’s no way any other design is being built here anytime soon.

    1. This doesn’t necessarily kill future AP1000s in the US, but it doesn’t help. What has to happen is that 1) The AP1000s in China get completed soon and startup testing does not reveal any significant problems, 2) The Vogtle/Summer schedules are maintained and there are no more financial surprises and 3) The Toshiba announcement on February 14 isn’t much worse than expected.

      Item 1 would likely impact item 2. Both could result in financial penalties for Toshiba that are beyond Toshiba’s strained financial resources. In that case, it gets ugly.

      While the major components have been delivered, Vogtle and Summer might have to get additional financing to complete construction. I don’t know about Vogtle, but I think Summer has made it’s last trip to the PSC/legislature. Let’s hope interest rates stay low!

      I am assuming that Fluor remains in charge of construction. If Fluor is dropped or quits (very unlikely in my judgement), I think the delay in transitioning to a new contractor could be fatal.

      Another factor is the supply chain for future AP1000s. The problems with module fabrication at the Lake Charles facility was a contributing cause of construction delays. Problems with an Italian supplier resulted in Westinghouse buying them as it did with CBI. Ultimately, alternative sources were found for some components. But it seems that this is an issue that has to be addressed before a future AP1000 is ordered. Presumably, all the engineering design will be completed!

      1. From Construction Milestones at New Chinese Units, WNA 5 Jan 2017:

        “Sanmen unit 1, construction of which began in April 2009, is expected to be the first AP1000 to begin operating. First concrete for Sanmen 2 was poured in December 2009. All four Chinese AP1000s are scheduled to be in operation by the end of this year.”

        Sanmen unit 1 was scheduled to go online late last year, now early this.

        Real Soon.

  2. Perhaps this shows how fragile these large projects really are (or rather, this is yet another example). There are simply too many variables that can ruin a large project – ranging from problems with investors, problems with regulations / the regulator, problems with production, problems with labor, problems with quality assurance, problems with intervenors, etc.

    I do no think that we need any more / deeper soul searching. We know the problems: inappropriate level of regulation driving up costs, extending the schedule, and causing uncertainty all the way around. We need to cut our costs and de-risk our requirements (e,g,, not allow the regulator to change the rules part way through the game) dramatically if we are going to ever be relevant in the US again. In my view, the drive to SMRs / advanced reactors should accomplish the former, but, as we have seen from the NRC / DOE output to date, will not accomplish the latter (even though they should since they are inherently much safer designs).

    We need a whole sale change in how we regulate our industry if we want to be relevant in the future.

      1. that’s the thing the small reactors have a faster and less risky learning curve. Large reactors can be very competitive on price if you can get a standard design and build enough to get over the initial learning curve for both the on-site construction and for your suppliers. Also with firm plans for future work so investment is made in supply chains to reduce cost.

        This however requires 10’s of billions of investment decisions to be made with similar order of risk. In today’s world there just isn’t that apatite for long term planning, investment and risk carrying of such magnitude.

    1. ” that can ruin a large project”

      Some of the AP1000 projects have been delayed and have seen cost increases. This not the same as being ruined.

  3. Westinghouse and Areva experience significant problems with executing projects. Does it mean that to be successful in the reactor business one needs a protected domestic market where projects are allowed to mature?

    1. The take away is that a firm that specializes in design should probably let the professionals do the construction. Perhaps construction was started too early before the supply chain was ready.

  4. The way CW makes mistakes and discovers them much later when rectifying them is massively expensive almost makes me think of sabotage. Given how much gas money is riding on stopping nuclear, how difficult could it be to infiltrate a subcontractor of a subcontractor?

    1. It is far less risky to let lawyers do what they do best without the risk of a James Bond covert action.

      1. Not much covert action required. A substantial part of the population really believe that nuclear energy is pure evil, that chernobyl killed millions and that every npp is a ticking timebomb. Finding such people in a non-nuclear company that suddenly has a key nuclear contract, and then convincing them to “do the right thing” would be quite easy.

    2. I am dismayed that this was either not caught at fabrication or if it was, was deemed good to ship.

    3. “Given how much gas money is riding on stopping nuclear …”

      A conspiracy argument is open to reversing the bad actors; that is, see how much nuclear money is riding in stopping frac’d gas production which might be impeded by some intentional ground water contamination.

      1. @Falstaff

        There is plenty of money from Saudi Arabia, Russia, Qatar, Iran and Israel that is interested in making fracking U.S. shales more expensive.

        They have far more resources than the nuclear industry. Saudi Arabia, for example, waged a 20 month price war on U. S. fracking (mainly oil) at the cost of tens of billions for both themselves and their OPEC partners.

        That not just my own supposition; you can find plenty of articles in the business press about the assumed reasons why the Saudis suddenly began producing as much oil as they could to flood the market.

        Here’s one example – http://www.latimes.com/opinion/topoftheticket/la-na-tt-saudis-price-war-20150115-story.html

        1. @Rod

          True, though I think the nuclear industry has more at stake, life and death so to speak. One could argue, given the nuclear shutdown in Germany (pending) and Japan, that the entire US nuclear industry could be destroyed by similar government fiat with sufficient FUD against nuclear in an environment of cheap gas alternatives. By contrast the OPEC players would lose money with either a US nuclear resurgence or expanding US gas production, but they won’t be put out of business in the coming decades.

        2. I suspect that the US wants low oil/natural prices to deny Russia of foreign earnings from sales of its oil/natural gas exports. Further, if our prices are low enough, it makes an export market that can wean Europe from Russian energy supplies more feasible. The Saudis cooperated with a similar strategy in the 1980’s.

          I can understand Iran’s interest in seeing our energy prices higher. Israel’s motive is not so clear other than to maintain continuing US interest in their biblical theme park and to maintain a possible export market for their recently discovered offshore gas finds.

  5. There is wider impact than just the US here in the UK just up the road from me is the proposed Moorside power plant which comprise 3 AP1000. The AP1000 was to complete the GDA process this year with investment decision start of next. The hope now is for KEPCO to step in and take up substantial part of the investment and lead the consortia to build the reactors.

    On that note KEPCO have shown that in south korea these type of projects canbe delivered on time and to schedule. Even in the UAE their first AP1400 reactor is due to start operating this year ahead of schedule. Having an experienced supply chain and building a number of these plants makes managing construction schedules and cost much easier.

    1. *APR1400

      And yes this is a different reactor design to the AP1000 but has its basis in much of the same technology

      1. ? No it doesn’t. The APR1400 is the offspring of the System 80 design. It has none of the passive features of the AP1000. The only link is that Westinghouse acquired the C-E nuclear business.

  6. Comment on the “Update”
    The actual NRC Event Report is here: https://www.nrc.gov/reading-rm/doc-collections/event-status/event/2017/20170117en.html#en52487

    The purpose of QA is to prove the ‘as-built’ meets the ‘as designed’, with the inherent assumption that the design is OK. If the approved design is not OK, QA won’t make it OK. Especially on FOAK stuff, other additional processes actually just ‘increase the confidence level’ the design is OK, e.g. building to Codes and Standards, actual testing, etc.
    These FOAK pumps are supposed to run 60 years without removal from the system for maintenance. The only actual way to ‘prove’ that is do it. Compared to 60 years the short test runs to date can only be used to increase confidence (no real way to do otherwise).
    But here’s the QA hook with CW-EMO. Their first pump problem was with the pump Thrust Bearing, which was discovered during pump test runs. That problem was traced to a ‘materials’ problem with the Subcontractor that supplied the Thrust Bearing. In other words the as-built bearing material did not meet the as-designed material requirement. So how did it ever get put in those pumps in the first place? It was only the ‘after-the-fact’ investigation that found the defect by reviewing QA manufacturing records after-the-fact. This is exactly the type of problem a functioning QA program is intended to prevent before the ‘defect’ ever gets installed for use.

    The current Part 21 issue:
    “The casings have an excess material condition in the transition region between the cast bowl and the suction nozzle. The condition is a discontinuous (non-tangential) axisymmetric feature between the cast bowl region and machined suction nozzle OD. The as-built transition feature on the casings has not been analyzed and is not included in CW-EMO Engineering Memorandum 7242, Revision 1, Volume 1, which is the applicable generic design report for the AP1000 RCP casing and main closure. ”

    Sounds like something where the suction nozzle attaches to the pump bowl doesn’t meet the design requirement. That doesn’t mean ‘it’s not OK’ it means it’s not analyzed.

    The question in my mind is how did umpteen of these pumps get manufactured and shipped with a defect where the ‘as-built’ does not meet the design requirements, when a functioning QA program will prevent this exact thing? Seems to be a repeat of the Thrust Bearing type of problem.

    1. You would be surprised at how much even an excellently run QA/QC program can miss because it is usually not possible for a QA/QC program to check / verify everything (and even if something was checked it matters who checked it and when they checked it).

      Without knowing any more about this new Part 21 issue than what you stated in your comment, since many pumps have this same issue my initial reaction is that either that specific area was not inspected or not inspected in a manner that would have revealed the issue.

      1. But this is a visually obvious discontinuity. It did not require an electron microscope or radiography to detect.

    2. The answer to the above question is in my experience in the engineering field as follows:

      Supplier company manager is under pressure to deliver casting prior to the end of quarter financial report.

      Supplier QA department notices the part doesn’t match the drawing and requirements and raises a defect report.

      Decision is made that rework will be too expensive, or take too much time, therefore Instruction is to send the part to the customer meeting shipment dates and carrying a risk that the customer will notice and reject the part.

      Once the part is delivered to the customer they notice the part does not conform. They inform the supplier and the supplier asks for a concession.

      Manager of the customers company can’t afford to wait for the rework as this would make him late costing time and money and have large programme risks. The decision is therefore taken to approve the concession and the customer to carry the risk that during testing the defect causes a performance or qualification issue.

      The steps described keep going on and on up the chain until ultimately the final product fails to match performance. The cost then to fix the final product is very large and results in long delays and huge programme impacts (aka AP1000 pumps).

      The cause behind it is an aversion to completing rework to ensure each part is to spec. The aversion to rework is driven by the cost of the reworks and time lost which the supplier may not be able to afford whilst the relative minimal risk of passing the problem on to the customer. Effectively once the customer has the part what are they going to do? sending it back is going to cost too much time, they can’t get someone else to fix someone else’s work, and getting an all new supplier is very expensive for them. They are left in a catch 22 position

      Interestingly the additional paper work asked for nuclear build often makes the above situation more likely as the cost of rework is substantially greater than for non nuclear work. In non nuclear work you just fix the part so it meets spec, the cost for doing so is far smaller than the cost of the bad reputation. For nuclear work you have to right chapter and verse saying what you have done and why and get every man and his dog to sign it off. This makes the rework task often more expensive than making the part in the first place as you have to pay engineering time sign off from regulators and so on. There are few suppliers in the nuclear industry due to the relatively small market and therefore less risk of loosing trade.

      Ultimately to solve the above situation especially for bespoke FOAK parts the customer should take all the risk of the rework by telling their supplier that they will cover rework costs (there are numerous models for this). This removes the risk of problems being ignored until it is too late. Unfortunately the financial people often see risk budgets as lost money and covering supplier rework as costs they shouldn’t incur so are not too keen on the idea.

      1. @Jeremy Owston says February 7, 2017 at 10:57
        And
        @RTK42C says February 7, 2017 at 10:50 AM

        I 100% concur with both assessments. The takeaway for me is in a start to finish project each major player in the process can only control the risk in his responsible area piece part. But they all have ‘skin-in-the-game’ financially (except NRC has no financial risk), and if any piece part of the project fails they all fail. I think Toshiba is saying they no longer will take that risk in the Construction part because they can’t control the Toshiba risk if one of the other area piece part fails. Based on the build record of these large units that is a prudent decision. When you add FOAK to a large build, the risk increases drastically (for a lot of reasons, one of which is relying on non-QA experience Subs for vital equipment).

        For me the answer is quit trying the large builds, the failure risk is huge and uncontrollable, or the final cost and schedule is outrageous. Both of which cause overwhelming permanent PR damage for nukes of any kind.

        And I have to wonder, W has decades of successful track record building both small canned RCPs and large shaft-sealed RCPs for their own NSSS designs, yet they choose not to attempt the large canned pump build for their own new NSSS design.

      2. “Decision is made that rework will be too expensive, or take too much time, therefore Instruction is to send the part to the customer meeting shipment dates and carrying a risk that the customer will notice and reject the part.”

        I don’t know if you are exaggerating to make a point or if this type of thing actually occurs. I grew up under the Rickover system and this is NOT my value system, and if such occurs the supplier has a severe ethics problem. They should be banned from nuclear Q work until all the responsible parties have been fired and replaced, everybody retrained, and then go through a probation period. Some of your other examples are similar. If this is actually typical and commonplace, we should not be building ANY nuke plants. When asked how he knew his plants were safe Rickover said because he approved the design, so he knew it was safe, and then his QA program insured the as-built matched the design. QA was the essential ‘proof.’
        But he followed up with later Ops Reports and if Ops experienced a problem they darn well better report it to Rickover for further evaluation by his experts. The point is HIS system worked, several more advanced designs were developed and put into operation under his system. Until the commercial system is run like Rickover’s system it will be plagued with the types of problems the commercial system has suffered for several decades. Duh (slaps forehead), that means no room for an NRC as currently structured.

        1. Unfortunately it does exist in both nuclear and non nuclear world. But to be fair the error from the suppliers view point can often appear small and not important, as they don’t have visibility how it will impact the overall part. They may honestly think it isn’t an issue. Equally I have been involved in companies where machines have been shipped unfinished, out of spec, just to meet an arbitrary date. At the end of the day if you only get rewarded for meeting a shipping date then that is what you do at the expence of anything else.

          The saying that I was taught early in my career was “get what you ask not what you expect”. If your only measure of performance is delivery and that determines your bonus performance reviews etc that is what you will meet, don’t expect it to be high quality too.

    3. ” The only actual way to ‘prove’ that is do it. Compared to 60 years the short test runs to date can only be used to increase confidence (no real way to do otherwise).”

      High confidence is the best that can be done by any reliability system. Even a a successful 60 year run of device one off the line does not prove device two is also be similarly sucessful the day it’s installed. QC in the fab of device two may not check every parameter relevant to success, or the run parameters or environment of device two may differ from installation to installation. See, for example, the SRB O-ring temperature variation of the 25th US Space Shuttle flight.

      In the end, the depth of understanding of how the device performs it’s function and under what conditions determines confidence, and the parameters established by this understanding as key to success can most always be checked by QA acceptance testing.

  7. Toshiba Corp. will cease taking orders related to the building of nuclear power stations, sources said Saturday, in a move that would effectively mark its withdrawal from the nuclear plant construction business.

  8. I work for a company that makes control valves for industry, including Nuclear power, and I submit that not every surface is controlled (by drawing or spec) in a casting, and that there are specifications that allow conditions that are less than perfect. We will have to wait for the final answer to know the true root cause of the issue.

    With FOAK, there is no history of what is REALLY critical, that only comes with experience with that specific design/product. That is why engineers design specific, intense testing regimens- to identify and flush out failure modes at the earliest possible time.

    If this SNAFU had occurred at a coal or gas site, no one would ever hear about it.

    1. At least this article doesn’t say radiation readings are rising. Other articles do and even show maps from 2011.

    1. @David B Benson

      I’d advise NOT contaminating this proposal with government spending programs of any kind. The topic of a carbon tax is already a big bite for many to accept; in order to get it to the finish line proponents need as many allies as they can find.

      Social Security has shown the way. If you want a LOT of allies, you have to make sure that the money generated by the program actually flows out to almost as many voters as there are sources of money into the program. The beneficiaries and often their families and friends will become the political constituency that keep the program in place long enough to do the good that it needs to do.

      The beauty of a real CO2 disposal fee that is assessed on all fuel sources that produce CO2 is that it is focused on addressing the market imbalance between fuels with a very cheap, almost free, waste disposal process provided by foisting the issue on everyone else and those energy sources that do not dump their waste into our common atmosphere.

      The fees will be paid by everyone who uses fuel based entirely on the amount of fuel that they use, but the payments will be equal to everyone. That levels up the competition and gives people the opportunity to choose to use fuels that have low or zero fees.

  9. SHANGHAI, Feb 9 (Reuters) – A 120 billion yuan ($17.5 billion) nuclear power plant in China’s northern Hebei province is expected to come online by 2020 and will use a Toshiba-owned Westinghouse reactor, the state-run China Daily said on Thursday. China’s nuclear agency said last year that it would build a nuclear reactor in heavily polluted Hebei province, an area near the capital Beijing, in a bid to cut smog and promote cleaner energy. The Hebei plant will use a Westinghouse AP1000 reactor, China Daily said, citing CNNP CHD Hebei Nuclear Power, a unit of China National Nuclear Power Co Ltd
    …Construction of the Haixing nuclear plant, which will be Hebei’s first nuclear facility, began last year in the coastal city of Cangzhou, China Daily said.

    Read more: http://www.nasdaq.com/article/nuclear-power-plant-in-chinas-hebei-to-be-online-by-2020–china-daily-20170208-01575#ixzz4YEIr6UKx

    [This is the first I’ve heard of this project.]

    1. Actually, at the World Nuclear Assn. website, one finds about 22 AP1000s planned including these two at Haixing; 64 more proposed; 62 more less-definitely-proposed; and two CAP 1400 under construction at Shandong (although one is a bit delayed.) Of course, 4 AP1000s are nearing completion this year, two at Sanmen and two at Haiyang. Whew.

  10. Interesting discussion on safety related components on the other thread. Regarding the AP1000: Yes it can safely remove decay heat without ac power. But the normal decay heat removal is much preferrable to having steam and dripping. condensation in the containment. So does it still have diesel generators and are they safety related because they would mitigate consequences of an accident?

    1. Yes, the AP1000 design has diesel generators. No, they are not safety related because they are not required to mitigate the consequences of accidents.

Comments are closed.

Recent Comments from our Readers

  1. Avatar
  2. Avatar
  3. Avatar
  4. Avatar
  5. Avatar

Similar Posts