By Sean Ricker and Robert Bjune, South Texas Project, and Matthias Svoboda, Alstom Power Service
The South Texas Project (STP) is a two-unit nuclear facility southwest of Bay City, Texas. Both units are Westinghouse pressurized water reactors with Westinghouse main electric generators. The electric generators are rated at 1,504 MVA and have an average electrical output of 1,400 MWe. Unit 1 began commercial operation in 1988 and Unit 2 in 1989.
The generator rotor and core are cooled by hydrogen gas. The stator bars are cooled by a Westinghouse designed stator cooling water (SCW) system. Each generator has 72 stator bars consisting of 36 top bars and 36 bottom bars stacked two to a slot. All stator bars are hydraulically in parallel. The cooling water is deionized via a mixed-bed demineralizer, at a low dissolved oxygen (<50ppb as per original equipment manufacturer specifications) and neutral pH. The source of makeup to the system is deionized water at a dissolved oxygen level of 10 to 15 ppb. The makeup is run through the demineralizer prior to service to reduce conductivity to 0.055µS/cm. Makeup to the system during operation is negligible as the system is very tight with few leaks.
During operation the chemistry of the SCW systems has been maintained extremely well. Conductivity is almost without exception <0.1µS/cm, DO is typically <5ppb. These values have been maintained this way for the life of the plants. Continuous online monitoring is utilized for conductivity and DO and backed up by periodic sampling by chemistry.
The lay-up of the SCW system during outages has primarily been maintenance driven. During outages in which the stator must be vacuum dried, it was done so (for example, for DC Hipot testing). This was typically only every six to 10 years. However, when not required to be dried the system was merely drained to the extent possible given the existing piping, and left open to air for the duration of the outage.
Stator Cooling Water System Problems
Throughout the units’ more than 20 years of service, the SCW systems have been operated and laid up nearly identically. The only exception came in 2003 when Unit 1 entered an extended outage (6 months). During that time the SCW system was operated intermittently and chemistry values were not always kept in their ideal bands. Prior to that outage both units ran with copper levels <10 ppb. However, following the extended outage Unit 1 regularly ran with copper concentrations near the alert limit of 20 ppb.
During the Fall 2009 outage, the SCW system was drained, blown down and vacuum dried to perform DC Hi Pot testing. Following startup, higher than normal copper levels were found in the system. During that 18-month operating cycle, five sets of filters became plugged and required changing and copper levels got as high as 240 ppb. This is well above the normal levels of <20 ppb.
During the resulting investigation, STP decided that their current use of 30 micron nominal string wound filters was outdated. They have since changed to an improved 30 micron absolute pleated polypropylene filter. The use of this new filter was initially thought to have been a contributing factor to the differential temperature issues but was later cleared of suspicion. They have been in service in Unit 1 since May 2011 and have been successful at maintaining low copper levels without plugging.
The rising differential temperatures were seen on both the top bars and bottom bars equally. However, the differentials themselves were actually being caused by roughly 50 percent of the top bars and 50 percent of the bottom bars rising in outlet temperatures while the other 50 percent remained stable or actually lowered in temperature. This data suggests that the 50 percent that were rising were fouling and losing flow while the others were actually gaining flow in return.
When this data was examined further it was found that those bars whose temperatures were rising had their inlet and outlet water connections lower than the bars themselves (see Figure 1 and Figure 2). This means that the bars that are drained during most outages are the ones that were fouling. Alternatively, the bars that get their water connections higher than the bars, and therefore remain filled during outages, showed little to no signs of fouling.
Cleaning the System
For chemical cleaning, acid cleaning is an efficient option. However, this poses substantial safety problems, both for handling and for the machines. It also brings complicated waste treatment. In order to minimize the risks, the machine has to be disassembled and the conductors cleaned individually. With this, only some components, but not the system, are cleaned.
Alstom’s proprietary Cuproplex method is based on the complexing agent EDTA (Ethylene diamine tetra acetic acid), which is applied through the whole system by recirculation. It dissolves only copper oxides and does not react with copper metal.
The reagent and the dissolved copper are absorbed in the mixed bed and then disposed of as solid waste.
The method can be applied either during a generator shutdown (offline) or when the generator is in operation (online). The latter method was chosen in the present case, as the last planned outage was only three months before and demand was at a peak.
The progress and result of cleaning are monitored by regular water analysis. A spectrophotometer, a conductivity meter and an oxygen meter are used. Information is supplemented by data from the power plant chemical laboratory.
Leading up from 1980 to June 2011, this method has been applied in 208 cleanings, on generators from 10 different manufacturers. Included are 51 online cleanings since its first application in 1996.
One of the biggest challenges during this process was the short time between identifying an adverse trend and the need to take action. The unit was started up from a planned outage on May 5, 2011. The adverse trend of differential temperature was first identified on May 23. At that time differential temperatures were at about 14 F and 10 F (top and bottom bars respectively) and were rising at a rate of about 1 F/4 days. By June 25, when the chemical cleaning began, temperatures had risen to 18 F and 16 F. A down-power would have been procedurally required at 20 F and a shutdown at 21.5 F.
|Figure 3 VIEW OF THE UPPER PART OF THE STATOR END WINDING WITH THE BLACK TEFLON HOSES|
An on-line cleaning has to be prepared thoroughly and adjusted to the case at hand. Thanks to efficient communication and processes this was achieved in a short time span.
The second challenge was the timing of this event. The summer of 2011 was an abnormally hot and dry one, even by Texas standards. Power generation was at a premium and the state’s power stations were regularly in a state of alert. Work that could potentially cause plant trips was minimized throughout Texas. However, STP had to decide between what was viewed as a “risky” online evolution or risk reducing power or worse. In the end, the decision was made to perform the chemical cleaning.
For STP, this process was a first-time evolution for the site, which presented another challenge. This being a first-time evolution under time pressure meant that working procedures needed to be created in a short period of time. Additionally, chemicals that had never been used on site needed to be authorized, ordered and received and unique environmental evaluations had to be performed.
The operation started on Saturday, June 25, 2011 and ended Monday, July 18 at 17:50. Throughout the cleaning the conductivity was kept within 7 ± 1µS/cm, with the exception of three deionizer resin change periods, where the reagent injection was interrupted. During the whole time the plant was operating continuously and with regard to the effects of chemical cleaning also at full availability.
Soon after the start of the cleaning the individual bar temperatures started to improve. After 48 hours the temperatures were back to normal. The cleaning was continued, as all the oxides have to be removed in order to prevent fast renewed plugging. It became apparent that the amount of oxides present was much higher than in other machines of a similar size that had problems with plugging.
This caused several challenges, as with continuing progress of the cleaning several key resources started to get short on supply. More EDTA had to be ordered, as well as personnel resources organized to guarantee continuation of the cleaning.
The total quantity of copper removed was 10.7 kg. This amounts to more than 90 percent of the oxides present in the system. More than 95 percent of this originated from copper oxides already present in the system, the rest either from metallic copper surfaces or particles. In a worst-ase scenario, assuming all of the 5 percent was from copper surfaces, this would result in a wall thickness loss of less than 0.1 µm. With oxidation, the metal removal is uniformly distributed.
This quantity of oxide is considerably more than what we normally would expect from such a machine, even considering its size.
It was decided to not continue, as further cleaning would have been inefficient and the system was already very clean considering its size.
Pressure drop across the stator, flow, as well as temperatures were already stable at design values.
The deionizer resin was changed in both vessels three times during the cleaning and once more after terminating chemicals injection.
Stator inlet pressure dropped from 36.0 psi to 29.5 psi, while flow increased from 630 Gpm to 690 Gpm. When normalized to a pressure of 32 psi, the improvement in flow was from 595 Gpm to 725 Gpm.
Temperature differentials improved from 15.7 F to 8.0 F and from 12.6 F to 3.6 F for the top and bottom bars, respectively. Average stator bar temperatures above inlet improved from 47.9 F to 45.5 F for the top bars, and from 43.1 F to 39.5 F for the bottom bars.
Strong coordination between Alstom and STP, along with clearly defined roles made a fast response possible. Staff dedicated to the cleaning on both sides enabled fast and open communication during the cleaning.
The short response time and following long intervention time tested both organizations. A regional service point for Cuproplex in North America allowed equipment to be set up by Alstom Richmond’s expert before the cleaning team arrived from overseas.
The large amount of oxides removed suggests other sources than air leaks. Such leaks would have to have introduced half a liter of air per day in average since commissioning. It should be noted, however, that such a leak is not necessarily visible by elevated dissolved oxygen levels, as the stator winding can be a very efficient oxygen scavenger.
Leaving the system open during outages, however, opens up a near infinite pool of oxygen for the duration of the exposure. The winding’s temperature distribution also points in this direction. Copper surfaces, either blank or with oxide, do not change into a deteriorating condition when dry. Reactions however take place, when wet, within a few hours. The presence of moisture, can only be eliminated by proper layup.
As is known, 10.7 kg copper (as Cu2O) correspond to 1347 grams of oxygen. This is the weight of 943 liters of oxygen (at npt conditions), or 4,500 liters of air. It also corresponds to the oxygen content of 158 m3 of air-saturated water.
Even in a machine of this size, a few kilograms of oxides can already cause considerable and visible plugging. The fact that this did not occur despite having many times over the potentially problematic oxide inventory shows that conditions in the system were quite stable during operation, allowing the oxide layer to build up without significant disturbances that would cause oxide migration.
On the other hand, had such a disturbance occurred, the resulting plugging could have been sudden and severe.
The likely source of the problems was the lay-up procedures. Often, the consequences of insufficient lay-up will only be visible years or even decades after the fact. Every original equipment manufacturer has a recommended lay-up practice for their SCW systems. In most cases, these recommendations align with EPRI recommendations, which are to leave the system in service if possible or to perform a complete vacuum dry down of the system if it will be drained for an extended period of time, typically defined as greater than three days. It is STP’s belief, as well as many in the industry, that these recommendations are difficult, time consuming, costly to follow and do not guarantee success. These systems were not originally designed with consideration to lay-up as evidence by the lengths that must be taken to do so. Additionally they do not guarantee that no problems will be encountered with regards to the internal chemistry of the machine and due to the amount of disassembly required, can also create more problems than are solved. STP has decided to institute a tailored lay-up program that will achieve near similar results at a lower cost and shorter time than the current industry standard.
In future outages, STP will be leaving clean system water in the stator winding and ring headers rather than intentionally drain them as has been done in the past. In addition, the process for filling and venting has been improved through procedure revisions to capture what was once just first-hand knowledge. These procedure enhancements also serve to enhance knowledge transfer which is a current emphasis in the industry. A new operator will not have to know from experience but will just have to follow a well written procedure.
The use of argon in the system is also being considered. Despite the makeup water being at 10 to 15 ppb dissolved oxygen, the process of filling an air saturated system with water causes the air in the water to increase to levels as high as 100 ppb. It is thought that if this air can be displaced with argon (or another inert gas) that dissolved oxygen levels can be kept within operating specifications through operation, drain down, refilling, and back into operation. This should further reduce any oxidation that takes place during refills following extended shutdowns.
Despite having a good wealth of online monitoring with regards to chemistry, the site like many others relies heavily on one differential pressure gauge as an indication of flow. However, this only tells a portion of the story. As was seen in this event, a steady differential pressure indication does not necessarily mean constant flow. In the future STP will be taking system flow readings regularly in both units in order to better monitor and trend the health of the machine. This data may be able to better identify a problem at an earlier stage before the effects are seen in differential temperatures. This will allow for better planning rather than a last-minute cleaning. It will also monitor the effectiveness of the new layup techniques. A real time flow instrument is being considered for installation in the future.
Chemical cleaning can clean plugged hollow conductors, but once a hollow conductor is completely blocked, chemicals cannot circulate inside and the channel remains blocked. It pays to act quickly once it is established that the machine has problems. Additionally, a large oxide inventory is not necessarily visible by outward symptoms, but poses an invisible threat to the system.
The stator cooling water system is often ignored due to its inconspicuous and seemingly trouble free nature compared to other systems in a power plant. It pays off to take care of this system, though, as neglect can quickly cost a lot of money when the plant has to be shut down for several days to several weeks in the worst case.
 Robert Svoboda, Christoph Liehr, and Hans-Günther Seipp, “Flow Restrictions in Water-Cooled Generator Stator Coils – Prevention, Diagnosis and Removal Part 4: Chemical Cleaning of Water-Cooled Generator Stator Coils by the Cuproplex Method” PowerPlant Chemistry 2004, 6(4), 197-202.
 Robert Svoboda and Hans-Günther Seipp, “Flow Restrictions in Water-Cooled Generator Stator Coils – Prevention, Diagnosis and Removal Part 1: Behavior of Copper in Water-Cooled Generator Coils,” PowerPlant Chemistry 2004, 6(1), 7-15.
 Robert Svoboda and Russell Chetwynd, “Flow Restrictions in Water-Cooled Generator Stator Coils – Prevention, Diagnosis and Removal Part 2: Detection of Flow Restrictions in Water-Cooled Generator Stator Coils,” PowerPlant Chemistry 2004, 6(2), 71-79.
 Robert Svoboda, Christoph Liehr, and Hans-Günther Seipp, “Flow Restrictions in Water-Cooled Generator Stator Coils – Prevention, Diagnosis and Removal Part 3: Removal of Flow Restrictions in Water-Cooled Generator Stator Coils,” PowerPlant Chemistry 2004, 6(3), 135-144.
 Turbine Generator Auxiliary System Maintenance Guides—Volume 4: Generator Stator Cooling System. EPRI, Palo Alto, CA: 2008. 1015669.
Power Engineerng Issue Archives
View Power Generation Articles on PennEnergy.com