The successful and speedy Comanche Peak steam generator replacement was marked by real teamwork from the get-go.
Shattering a world record…
Careful Planning, Hard Work And Luck
On April 20, 2007, TXU closed Comanche Peak Unit 1’s breaker and ramped up in power, completing a combined steam generator and reactor head replacement outage that began on February 24, just 55 days earlier. That’s right, just 55 days. While steam generator replacement projects have been performed in the U.S. and around the world many times over the past two decades, short outage durations have come to be known as the measurement of success—and competition for “who” is the best is measured in days and hours of outage durations. Interestingly, until Comanche Peak was completed, the world record stood at 63 days, 13 hours. That was until a TXU-Bechtel team of highly motivated, experienced personnel shattered that record by more than a week at the Comanche project. (Bechtel was TXU’s prime contractor for the replacement of the steam generators.) Here’s how the team did it.
Comanche Peak is a two-unit nuclear generating station located about 60 miles southwest of Dallas. Each unit generates 1150 MWe, and the pressurized water reactor design includes four steam generators per unit, with each steam generator weighing about 400 tons, nominally 15 feet in diameter and about 70 feet long.
During original construction, the containment structure around the nuclear steam supply system was completed after the steam generators were installed. Consequently, there was no existing access to remove and replace the steam generators. The replacement methodology would dictate an opening through the containment wall. This containment access opening had to be located approximately 100 feet above the ground elevation outside, directly above the containment building’s only equipment hatch. The logistics associated with creating and closing an opening above the equipment hatch without impacting the normal flow of tools and equipment to support the outage through the hatch took years of planning and coordination. Furthermore, the existing crane in the containment building did not have adequate capacity to lift the old or new steam generators, so an alternate lifting system would be installed.
After the project was awarded in 2004, TXU challenged Bechtel to work with it to match or beat previous steam generator replacement (SGR) outage durations, focusing on 65 days as an aggressive but attainable schedule. In addition to the replacement of four steam generators, the scope of the project included the replacement of the reactor head, along with installation of new cabling, new cable trays and a new air-handling unit with all new ductwork. The upgraded design of the new steam generators included the installation of new, rerouted main feed water piping, as well as new hangers, snubbers and whip restraints.
The rerouted feed water piping interfered with existing containment building ventilation ductwork, so the ductwork required rerouting, along with new hangers. The new steam generators’ instrument tap locations required removing the old and installing new instrument tubing, again with all new seismicallydesigned hangers. For planning purposes, the scope of work to be performed during the Comanche Peak outage was considerably greater than that of comparable projects. In fact, by the time planning was completed, there were over 5000 measurable activities in the replacement’s schedule.
The four new steam generators were built in Spain and arrived on site December 9, 2006. Their journey began on an oceangoing vessel from Spain to Houston, and continued by train to the site, using an upgraded train track that hadn’t been used since the power plant’s original construction in the 1980s. The new reactor head, also fabricated in Spain, was outfitted with new control rod drives in Pennsylvania and transported by barge to Houston, then trucked to the site.
Application of technologies
During the replacement planning process at Comanche Peak, significant consideration was given to the use of both proven technologies and new technologies. Over the years, TXU and Bechtel had developed strong working relationships with specialty contractors who had been able to demonstrate high-performance capabilities. Integrating these specialty contractors and their technologies with the project team proved crucial to the project’s success.
Every aspect of the Comanche Peak project was reviewed and scrutinized to determine if the current technology was the best available, or if something new might be worth pursuing. This approach to technology use was applied to rigging, concrete removal, templating, decontamination, welding, non-destructive examinations and other activities within the project.
To handle the large components through an opening in the building, a lifting system would be needed. Prior to the outage start, erection of the Outside Lift System, or OLS, was completed. The Comanche Peak OLS was by far the tallest that had ever been used for replacing steam generators. Because of the OLS height and the presence of certain underground safety-related commodities in the vicinity, the OLS header beam was tied back to the containment building.
Large Kevlar slings, six on each side, running from the header beam to the baseplates on the containment wall, were installed in the unlikely event that a tornado would strike. If the OLS were to take a direct “hit” from a tornado, the slings would keep the header from falling away from containment and potentially damage important underground commodities.
The height of the OLS and the amount of work to be performed from the work platform required installation of an elevator as well as a stair tower. During hydrodemolition, rebar removal, liner plate work and reinstallation of these commodities, a break area was installed on the runway. This area offered workers a place to eat without having to make the trek from the OLS to the ground and back.
Another interesting feature of the OLS was the use of a strand jack system in lieu of a chain jack system. Chain jack systems are very reliable and have been used on all Bechtel steam generator replacements to date. However, given the height of the Comanche Peak Unit 1 alternate access, the travel time from the opening height to the ground or vice versa was going to be almost six hours each way if the traditional chain jack system was used. The strand jack system, in comparison, was not only reliable, it traversed the same vertical distance in less than two hours, reducing steam generator lifting/ lowering time by two-thirds. This improved the project’s overall schedule while also limiting the time that the loads were on the OLS and potentially exposed to changes in the weather, such as high winds.
After OLS erection, the lifting system was load tested. The weights for the load test were large concrete blocks placed on a large frame. The load test weight requirements were 110% of the heaviest load to be lifted—which was a new steam generator. For Comanche Peak, the load test weight was 500 tons.
Hydrodemolition (the use of high-pressure water to demolish or remove concrete) was used on the containment building to create the Containment Alternate Access (CAA). Since the size and location of the containment’s equipment hatch was not sufficient to remove steam generators or a reactor head, an alternate access was needed to remove and install the old and new components. The access was located more than 100 feet above the ground, directly above the existing equipment hatch. The hydrodemolition equipment was mounted on a work platform at this elevation.
Tanker trucks brought about one and a half million gallons of water needed for hydrodemolition, preserving Comanche Peak’s supply of lake water. Twelve dieselpowered 475 hp pumps delivered water via high-pressure hoses to the two robots on the work platform. The water, at a pressure of 20,000 psi and a flow rate of 300 gpm, was directed at the containment through four rotating nozzles, each with a delivery opening of 3/8”. In simple terms, the process pushed 5500 horsepower through four small holes! This process also was extremely noisy, and hearing protection was an operational requirement.
During the concrete removal, as several layers of rebar were exposed, the hydrodemolition process was temporarily interrupted so that the rebar could be match-marked, cut and removed. A single bar weighed over 400 pounds, so each one was handled with ropes and pulleys until it was safely placed on the work platform. While the components were being moved in and out of the containment, each removed rebar was inspected, replaced if necessary and reinstalled later in the project using the cadwelding process.
The project selected laser templating to measure and determine where to machine the new steam generators, cut/machine the existing reactor coolant piping, and to ensure fit-up for welding when the new steam generators were installed. Using laser technology (i.e., laser templating), technicians measured the size of the reactor coolant piping and recorded the piping’s location with respect to the steam generator support system. A threedimensional database was created with this information. This process was done for all four steam generators.
When the new steam generators arrived on site in December, they were stored in the new steam generator storage facility (NSGSF). Technicians used laser templating to record the locations of the new primary nozzles in relationship to the steam generator tube sheet extension, which is the steam generator’s lower support system. Similar to what was done for the existing steam generator/reactor coolant piping geometry, a three-dimensional database was created for each new steam generator.
The two databases were then merged for the existing RCS piping and its corresponding new steam generator. Through model comparison and data reduction, where to cut and machine the new primary nozzles was determined. Laser templating was then used to position the cutting and weld preparation machines on the new steam generators. All of this work was completed before the outage started in February. During the outage and after the old steam generators were removed from containment, laser templating was used inside containment to position the weld preparation machines on the existing RCS piping. The weld preparations and non-destructive examinations (NDE) were then completed.
While it’s called pipe-end decontamination, this activity is really the decontamination efforts applied to the remaining stainless steel hot leg elbow and the remaining stainless steel cross-over leg elbow for each steam generator. During the outage, this process was applied a total of eight times, to four hot leg elbows and four crossover leg elbows.
After an old steam generator was lifted out of its respective cubicle, the open-ended elbows created a very high dose rate in the general area where templating, machining and welding subsequently took place. This required the highly contaminated oxide layer inside the elbows to be removed. Pipe-end decontamination equipment was placed in the cubicle. The equipment consisted of a seal dam that was inserted down into the elbow. Another piece of equipment, called the blast head, was attached to the end of the elbow and an air-tight seal was created. Grit-impregnated sponge media was delivered through a rotating arm with a nozzle on its end, all of which was mounted inside the blast head. High-pressure air delivered the media and the overall effect was to blast the oxide layer off of the inside of the elbow. The computer-controlled rotating arm moved into and out of the elbow, effectively covering 100% of the inside surface of the elbow. The spent media was recovered through a vacuum system and emptied into shielded spent media drums. The drums were located in shielded carts for subsequent handling and disposal.
Narrow groove welding…
After installation, the new steam generators were re-attached to the reactor coolant piping by way of “narrow groove” welding. While the new steam generators were stored in the NSGSF, a narrow groove weld preparation was machined on the new primary nozzles. After machining, the weld preparations were buffed and NDE was performed to assure that no machining marks were present to provide false indications of defects in subsequent weld examinations.
Once the steam generators were removed from containment, the existing pipe ends (actually the RCS stainless steel elbows) were machined to a narrow groove weld preparation. After the new steam generator was rigged into containment, upended, lowered into place, and final fit up was achieved, the narrow groove welding process began. The track and weld head were mounted onto the nozzle-elbow configuration. Weld wire was fed into the groove, and the actual welding was monitored and/or adjusted by welding operators observing the welding through the use of cameras and monitors. As the weld progressed and suffi- cient weld material was deposited, another welding machine was mounted inside the nozzle-elbow configuration so welding could take place both inside and outside of the piping simultaneously.
Narrow groove welding provided the project with a number of benefits compared to conventional grooves. The amount of weld volume to be deposited was 70% less, reducing time and radiation exposure, as well as saving money. Production rates were improved due to automation and weld shrinkage was greatly reduced, which helped to prevent movement of the existing piping systems. This, in turn, reduced work for resetting clearances on critical components as part of restoration.
All of the narrow groove welding equipment was designed for remote operation. Therefore, during the outage, all of the RCS welding operators were stationed in a central location in containment, in a lowdose area. From this location, three, four or even five RCS welds were occurring simultaneously. Upon completion, the welds were thoroughly inspected, including with xray. Weld quality on these eight, large-bore stainless steel welds was first-time, with no defects recorded and no repairs needed.
When welding of the piping for the reactor coolant, feed water, auxiliary feed water and main steam systems was completed, the code of record required that each finished weld product be radiographically examined. The project elected to use Computed Radiography, or digital radiography, for this NDE, primarily because this technology would minimize the effects on adjacent/near work activities during the actual radiography process. Low-curie sources were used throughout the project for shooting the welds. Boundaries were established in a similar fashion, as with all types of radiography, but the low-curie sources allowed the boundaries around the weld to be radiographed to be much closer to the source, allowing adjacent work activities to continue.
On-site concrete batch plant…
Restoration of the CAA required the concrete to be replaced with material equal to or better than what was removed. Upon restoration, the concrete was tested and after the new concrete reached its designed strength, the containment building was pressurized to demonstrate its acceptability for use. The new concrete was “batched” on site using a portable batch plant. (The decision to use a portable batch plant in lieu of a local concrete company had been made by the project many months prior. Although several local companies had been given an opportunity to provide the concrete for the project, for various reasons, including logistical ones, they had chosen not to participate.)
There were a number of important considerations associated with the use of the portable batch plant. For example, the potential for dust during batching operations required submitting an application for a state air-quality permit.
Another concern involved the moving of freshly-batched concrete from the portable plant to the containment opening. Concrete trucks and drivers were rented and used for the opening, ensuring the project that the resources would be there when needed. To get the concrete from the trucks up to the opening, a pumper truck was rented and used. And, as normally planned, a second pumper truck was available on site in the event that a malfunction on the first pumper jeopardized placement activities. During placement, a seal on the delivery piping did blow out, but the seal replacement took a shorter amount of time than moving the existing pumper and replacing it. Again, luck was on the project’s side.
As planning for the project neared completion, over 1300 additional people were needed at Comanche Peak to carry the steam generator replacement. Of this number, about 400 were made up of TXU personnel, radiation protection contractor personnel, security personnel and Bechtel Field Non-Manual personnel. The remaining 900 were craftsmen from direct hire and specialty subcontractors. Of the 900 craftsmen, Bechtel typically subcontracts work scopes that require about 250 subcontract supervision and craftsmen. Typical subcontract scopes that include craftsmen were:
- Reactor Coolant Pipe Cutting, Machining and Welding
- Liner Plate Cutting and Welding
- Insulation Installation
Attracting and retaining qualified craftsmen in the numbers required for the project—about 650 in total—was a challenge. Early on, the project realized that recruiting adequate numbers of personnel would require some special things.
To effectively recruit in a tight and specialized job market, creative methods were developed to attract and retain the quantities of qualified craftsmen needed to complete the project. In parallel, contingency plans were made by the project in the likely event that shortages of craftsmen became a reality. Various project work scopes were identified and packaged. Then, subcontractors capable of performing to the project’s expectations were identified and retained. In reality, approximately one-third of the work scope originally planned to be performed by direct-hire personnel was performed by subcontractors, either through packaged work scopes or by augmenting the direct hire contingent of personnel.
A week before the outage started, the entire project moved to two 12-hour shifts, with working hours from 0530 until 1800, and 1730 until 0600. Working 12-hour shifts for several days prior to the outage start provided ample opportunities for everyone’s minds and body clocks to acclimate. Similarly, any issues with traffic backups, how long it took to enter the protected area, etc. were sorted out, prior to (and without affecting) the outage.
Even though the project worked 24/7, each and every Bechtel individual, including Field Non-Manual personnel, direct-hire craft and all Bechtel subcontractors, were assigned a day off. Once they were assigned that day, it was their day off for the duration of the outage. The project hired additional personnel to ensure that, when someone had his/ her day off, work continued.
Prior to the start of each shift, the project had its Plan of the Day (POD) meeting. Chaired by the Bechtel shift outage manager, the main focus of the POD was to present, in written form, what was planned to be performed in the next 24 hours. The POD covered project safety, project quality and the project’s schedule progress. These shift meetings lasted no more than 10 minutes. This same information was shared with the all of Bechtel’s Field Non-Manual personnel, its craft personnel and its subcontractors at the start of each shift.
The replacement outage started on Saturday, February 24, 2007 when TXU took Comanche Peak Unit 1 offline at 12:00 PM. At this time, hydrodemolition of the concrete on the outside of containment was set to commence with equipment that was positioned 100+ feet off the ground on the work platform. Unfortunately, that same day, north central Texas experienced its worst dust storm in over 20 years. Sustained winds in excess of 50 mph with gusts of over 70 mph were recorded. As a result, almost all of the project’s equipment, including cranes and man lifts, had to be secured for the day. Almost all of it. Interestingly, the only pieces of equipment qualified to operate in this tough first-day environment were, luckily enough, the work platform and the hydrodemolition robots.
TXU and Bechtel worked closely together, performing concurrent activities during plant defueling as well as during reactor core reloading, system fill and post-outage testing, such as containment pressure testing. Project engineering kept pace with the accelerated schedule, ensuring that system conditions met all of the plant’s technical requirements for the plant’s configuration.
Differing work groups pitched in and helped when requested by outage management. In lieu of reducing the work force when certain work activities were completed and turned over for testing, the work force was redirected to complete future work activities or to assist other work groups. As was the case for Comanche Peak, when all of the project’s critical path work activities and its near-critical path work activities were over 10 days ahead of schedule, opportunities for work activity completion were everywhere, furthering the project’s momentum.
World record results
The successful completion of this 55- day outage is truly one for the record books. The following results of the project speak for themselves:
- Over 1,000,000 job hours without a Lost Time Accident
- 21 of 21 (100%) large-bore piping first time quality RT welds
- 52 of 63 (83%) feedwater & auxiliary feedwater first time quality RT welds
- Completed Project with Code- Compliant 100% Computed Radiography
- Multiple ( i.e., 5 ) crane operation in containment with no mishaps
- Sustained cadweld production of over 50 per day ( plan was 32 per day)
- Never exceeded the 72-hour per week per employee rule
- ALARA Dose Goal: Plan 156 REM Actual 123.9 REM
- Breaker-to-Breaker: Plan: 65 Days Actual: 55 days (a world record by over a week).
Richard L. Miller is senior project director with Bechtel Power Corporation. Telephone: (301) 228-6215; e-mail: firstname.lastname@example.org