On February 16, 2015, an AP article by Ray Henry titled Nuclear plants delayed in China, watched closely by US firms contained a short paragraph that has contributed to a number of sleepless nights. I’m pretty sure there are plenty of other people affected in the same manner who have far more at stake than I do. Here are the worry-simulating words.
Officials at China’s State Nuclear Power Technology Corp. blame the delays on the late delivery of equipment from the United States. Westinghouse Electric Co. and project manufacturers are working to redesign a coolant pump for the plant.
The pump being redesigned is a reactor coolant pump manufactured by the EMD division of the Curtiss-Wright company. There are four of these pumps in each Westinghouse AP1000® pressurized water reactor. There are four AP1000 reactors under construction in China and four in the United States. Each of those reactors is a major construction effort with a budget of somewhere between $5 and $10 billion and a project schedule that includes a minimum of four years on site with peak employment of roughly 2,500 – 3,000 workers per unit.
Aside: If you need information about what reactor coolant pumps do in a pressurized water reactor, please refer to a detailed article titled The world’s largest canned motor pump, published in Nuclear Engineering International on January 1, 2013. End aside.
Each of the pumps is vital to the operation of the plant it will serve; there are no installed spares and no easy way to replace a pump. A pair of pumps is bolted and welded to the casing at the bottom of each of two steam generators in the plant; there are no isolation valves. It is not possible to operate a plant without all four pumps operating properly. Therefore the plant owners expect that the pumps will operate reliably for the full 60-year lifetime of the reactor plant that they have purchased.
A couple of decades ago, when I first started learning about the advanced passive nuclear power plant designs that Westinghouse was developing, I was pleased to find out that the systems would include an “innovation” called a canned-motor reactor coolant pump. I’ve operated remarkably reliable machinery in the same family; the ones with which I was familiar had seen 25 years of hard service without any hints of mechanical failures.
They were so reliable, in fact, that the uninstalled spare pump had been removed and redeployed to newly built facilities. We retained the installed spares that provided the ability to accept electrical or mechanical failure while retaining overall system operability.
At the time I learned about commercial interest in canned-motor pumps, I had recently learned that only the first commercial PWR used them; all subsequent plants had pumps with rotary seals. I didn’t think much about the basis for that choice, but I remember wondering why my acquaintances who operated commercial plants expressed occasional concerns about reactor coolant pump seal failures. As I learned more, I found out that improving seals was an important area of research and development.
It seemed logical and reassuring that modern commercial plants would take advantage of a well-proven technology, but I neglected to consider the risks associated with a vast increase in the size of the pumps that would be required. The largest canned-motor pumps with extensive operational history have perhaps 1/10th of the nameplate capacity of the pumps proposed for the AP1000. The AP1000 RCPs are 6.9 m tall, about 2 m in diameter at the widest point and weigh more than 91,000 kg.
Apparently, the key decision makers were not very worried about the challenges associated with scaling up the pump either. Again referring to The world’s largest canned motor pump:
The AP1000 pump is designed with EMD’s strong design and analysis capabilities, which are based on more than 50 years of canned motor pump experience. Advancements in analytical software capabilities have been maintained and improved for all aspects of the design including thermal-hydraulics, structural, dynamics, and electrical design. Specific features targeted for design analysis utilize computerized fluid dynamics software for the hydraulic design, and internal circulation features to support the flywheel and thrust bearing designs. In addition, EMD has modeled defined regions of the motor electrical design and performed electromagnetic finite element analysis to optimize flux field performance.
Careful reading of that quote by skeptical, experienced operators might initiate some serious questions like the following. “Are you telling me that you’ve never actually built one of these machines? Your confidence is based on software modeling, right? What kind of guarantees are you willing to provide that your pumps will work well enough so that they do not put several multibillion dollar investments at risk of never producing revenue?”
I have not been able to obtain any responses from the pump manufacturer, despite several repeated emails and phone calls to a variety of different points of contact listed both on the EMD division web site and on the parent company web site. Messages left with the public relations point of contact have also not been returned.
Therefore I am left with reporting the design issues that I was able to find in published sources. Here they are in date order.
On May 19, 2010 — nearly five years ago — Nuclear Engineering International published an article titled Successful testing of AP1000 reactor coolant pumps that included the following quote.
Curtiss-Wright Flow Control – EMD and Westinghouse Electric Company LLC say they are continuing testing of the world’s largest canned-motor reactor coolant pump (RCP) for the AP1000.
During testing on 13 May, testing of the AP1000 pressurized water reactor RCP was successfully completed in ambient operating conditions. On Saturday, 15 May testing was successfully performed at normal operating temperatures and pressures. All test objectives for both tests were met. The tests were completed ahead of schedule, and witnessed by representatives of China’s State Nuclear Power Technology Corporation, owners of the four AP1000 units currently being constructed in China.
On April 17, 2012, Nuclear Engineering International published a brief story titled Sanmen RCPs to be shipped by June that said that the first two reactor coolant pumps for the Sanmen power plant — the first of a kind AP1000 — had completed 500 hours of testing and would be shipped by the end of the second quarter of 2012. The article ended with the following statement with no further details provided.
…SNPTC, Westinghouse and Curtiss-Wright had “jointly overcome the challenges with the AP1000 RCPs.”
On May 3 2013, Curtiss-Wright submitted a Part 21 report of defects and noncompliance titled Insufficient Process Control on Pump Impeller that described how a 3″ by 2.5″ piece of the pump impeller, which had been supplied by Wollaston Alloys, had separated from the leading edge of one of the pump blades during final testing.
On June 17, 2013, Wollaston Alloys submitted a document titled Report of Potential Substantial Safety Hazard in accordance with Title 10 Code of Federal Regulations, Part 21. That document described some quality control issues, the extent of the conditions discovered and actions taken to prevent them from recurring.
As a result of the defects, SNPTC shipped three of the four pumps it had already received back to the US for component replacement and factory retesting. The first two repaired pumps were shipped back to China in August 2013.
On September 19, 2014, Nuclear Engineering International published a story titled Design issues resolved for China AP1000s that led off with a statement about the resolution of first of a kind engineering issues sourced to Westinghouse CEO, Danny Roderick. It continued with a description of the installation plan for redesigned and remanufactured pumps to support a schedule starting date of December 1, 2015 for Sanmen 1.
The article explained the status of the pumps issues as follows.
Timothy Collier, Westinghouse vice president and managing director, China, said in an April 2014 press conference that the problem with the reactor coolant pumps, manufactured by Curtiss-Wright flow control in the USA, was a matter of assessing mechanical tolerances in the pumps. He said that it was not a major design issue, but more one of validation and verification: some pumps had passed factory tests, and others had not. He said that the effort to resolve the issue involves Westinghouse working with Curtiss-Wright, outside experts, Chinese owners and the Chinese regulator the NNSA.
Though I am not a credentialed design engineer, I’ve worked with a lot of them and been involved in a number of engineering troubleshooting endeavors. The problems that are often the most difficult to solve are the ones that don’t always occur. Intermittent failure modes are hard to track down to the source of the problem, which makes it hard to know when you’ve found and corrected the root cause. Without correcting the source, you are never quite sure that you won’t have a similar problem later.
Based on the statement from the February 16, 2015 article with which this piece starts, it seems that there are still unknown issues with the pumps that require the suppliers to continue “working to redesign a coolant pump.”
When you notice the time delays associated with the above chain of events, take into account the fact that each pump is so large that it requires transport by a heavy lift truck, realize that the lead units are located half way around the world from the source factory, know that every day of delay for each AP1000 reactor plant costs between one and two million dollars, and remember that there are already eight units under construction with hopes for “well over 30 AP1000 units to be in operation by 2030”, you can perhaps understand why I’ve been having so much trouble sleeping.
There is so much riding on the success of the AP1000; I certainly hope that the reactor coolant pump issues get resolved without too much additional delay. However, I have to admit that I am no longer very optimistic and don’t see much evidence of a plan B for this particular design. There are, of course, an infinite number of ways to avoid the specific lesson that should be taken away from this saga and many nuclear innovators are working on projects that do not include this weak link.
One of my favorite ways to avoid the project risk that comes with design or manufacturing challenges associated with key components is to eliminate as many of those components as possible. NuScale, for example, has no worries about finding a reliable design or a reliable supplier for reactor coolant pumps. Its natural circulation design does not include any pumps in the first place.
I’ll close with a 2009 clip of Ted Rockwell, who served as Admiral Rickover’s technical director for the USS Nautilus and the Shippingport reactor, testifying to a US Senate committee about the advantages of building reactors that do not need to use the “world’s largest” components.
Food for thought:
For Want of a Nail
For want of a nail the shoe was lost.
For want of a shoe the horse was lost.
For want of a horse the rider was lost.
For want of a rider the message was lost.
For want of a message the battle was lost.
For want of a battle the kingdom was lost.
And all for the want of a nail.