By Keith MacLean and Eric Fournier, New Brunswick Power Generation, and Javier Gomez-Grande and Tony Scandroli, Welding Services Inc.
New Brunswick Power (NB Power) is the largest electric utility in Canada’s Maritime provinces, with a generating capacity of approximately 4,000 MW from a mix of nuclear, fossil and hydro power. In NB Power’s generation portfolio, the 490 MW coal-fired Belledune Generating Station is the newest boiler, having entered service in 1993.
In 2007, NB Power embarked on the world’s largest Inconel waterwall weld overlay project at the time at the Belledune Generation Station. Project scope included repairing existing waterwall fireside corrosion and preventing future low NOX firing waterwall wastage. The project involved investigation, engineering and repair work.
Between 2000 and 2003, NB Power made an initial investigation of the best practices for repairing and preventing waterwall low NOX wastage. In 2005, waterwall corrosion risk factors were reviewed and a detailed inspection of the Belledune waterwalls was planned for 2006.
In the 2006 outage, severe waterwall wastage was found, with some areas experiencing up to 50 percent fireside wall loss. This corrosion was caused by the fuel blend’s high sulfur petcoke component. Action had to be taken to limit further corrosion damage during the coming 12 months of operation, until the 2007 outage when the weld overlay would actually be performed.
All repair methods were investigated, including sprayed plasma-type coatings, weld overlay with either 309SS, Alloy 33 or Inconel 622. The options of field-weld overlay versus new-shop weld overlay panels were also reviewed.
A detailed engineering analysis of the waterwall shrinkage that would be experienced with this large amount of weld overlay focused on waterwall distortion and interaction with the buckstay support system. This was done to ensure that no damage would be done that would impede tilting the burner nozzles in this tangentially fired boiler. The analysis was handled by a detailed finite element analysis (FEA) of the waterwalls and boiler buckstay system.
The 11,000 square feet of Inconel 622 weld overlay was performed in the 2007 outage. An inspection after one year of full-load operation revealed that the weld overlay was standing up well, as were the surrounding non-weld overlaid areas. Not even one tube leak was experienced during this work, either prior to, during or after the weld overlay outage.
Preparing for Waterwall Tube Wastage
In the mid-1990s, NB Power studies monitored gases such as carbon moNOXide and hydrogen sulfide along the sidewalls that would be indicative of waterwall corrosion risk. With the existing coal-only fuel blend, no waterwall corrosion was detected.
In 2000, NB Power reviewed the waterwall damage experienced on an Italian 660 MW Babcock & Wilcox (B&W) opposed wall-fired boiler. Operation of this boiler, with over-fire air (OFA) and a high sulfur fuel, resulted in waterwall wastage rates of up to 0.160” per year.1 Clearly, waterwall tube wastage was something for which NB Power needed to be prepared.
In 2003, following meetings to discuss waterwall wastage repairs, NB Power concluded that weld overlay was the preferred repair method in case waterwall wastage problems were experienced in the NB Power fleet in the future. The other option, sprayed-type coating, was not considered even though it would have been less expensive because of concerns about flaking and cracking of the protective layer.
Starting in 2001, petcoke was co-fired with coal at Belledune. Starting at about 5 percent, the amount was slowly increased to approximately 25 percent by late 2005. At that time, the decision was made to perform a complete furnace waterwall inspection during the spring 2006 shutdown. The thinking was that Belledune was now operating in a risk area for experiencing waterwall wastage, mainly due to the larger amounts of high sulfur petcoke being co-fired with the coal, along with the use of the original separated over-fire air ports (SOFA).
Petcoke is slower to burn than coal and the combustion system can deposit partially burnt petcoke particles on the waterwall tube. Using high sulfur fuel and running the main burners at a stoichiometry below a value of 1 are key risk factors in the corrosion of the fireside of the waterwall tubes. The boiler also had a NHIPA of 1.8 (net heat input divided by cross sectional area) that was high enough to be another risk factor.
2006 Inspection Results
The 2006 inspection showed heavy fireside corrosion loss of up to 50 percent of tube thickness, with a distinct corrosion pattern that was keyed to the combustion system details.
Track-mounted, automated overlay machine at Belledune. Photo, New Brunswick Power.
The 2006 inspection showed similar corrosion patterns on all four walls, with a slight difference between the long and short walls. The furnace is slightly rectangular, with the east and west walls being 48 feet wide and the north and south walls 50 feet wide. This 5 percent difference in wall length is large enough to noticeably affect the corrosion pattern.
A better way to understand this is to look at the burner angles. For the short walls, the burners are pointed into the furnace at 42 degrees away from the wall. On the longer walls the burners are pointed 39 degrees away from the wall. This affects the burners’ aerodynamics and how the flow interacts with the furnace waterwalls.
On the shorter east and west walls, while the corrosion was more aggressive, the area affected was more distinct, with less square footage involved. These burners are pointed further away from the wall, which helps to protect the wall immediately downstream of the burner corner. However, there is more distinct and aggressive corrosion taking place when the combustion products touch the wall or recirculate to the wall in a large eddy. The longer north and south walls suffered less corrosion thickness loss, but the area corroding was more diffuse, so it involved more square footage. With these burner corners being tilted closer to the wall, the combustion products impinge over a wider area but at a substantially slower corrosion rate.
For all walls, the most severe corrosion area was in a concentrated area between the top burner and the SOFA. This gap area corresponds to an area of high heat input, which accelerates the corrosion rate. Also, computational fluid dynamics (CFD) models show this is an area where the upstream burner flame hits the downstream burner and is deflected up and over, then it hits the wall. This would be a source of partially burned fuel hitting the gap area. Also, there is no concentric firing system (CFS) air nozzle on the top of the highest coal nozzle, which may also accelerate corrosion in this area. The gap area showed corrosion loss of up to half of the original tube wall thickness.
The less severe but larger wall area loss covered the area from well below the bottom burner to above the top burner. From the corrosion pattern extending below the bottom burners, it looks like this is a combination of the bottom flame rolling down under the downstream burner and hitting the wall, and there is no CFS air nozzle below the bottom burner so the lower wall lacks a protective high oxygen air blanket. This corrosion was worse about two thirds of the way along the wall, where the fireball would be closest to the wall. This was also where the flames would tend to lick the walls and deposit partially burned petcoke fuel particles. The high sulfur and high carbon content of partially burned petcoke particles is perfect for corroding the carbon steel waterwalls under low stoichiometry conditions.
Close-up of weld overlay prior to buffing and inspection. Photo, New Brunswick Powerr.
This location is also where the protective high-oxygen CFS air blanket along the waterwalls would be exhausted. The air blanket is composed of some auxiliary secondary air nozzles that are directed away from the center fireball, closer to the walls. There is one CFS nozzle above and below each auxiliary air nozzle; this is the Alstom CFS system. Key points include:
- The original design had a CFS 22 degree angle to protect the walls. The angle is from the fuel nozzle toward the wall until the CFS auxiliary secondary air nozzle angle is reached.
- The CFS angle was reduced to 0 degrees during commissioning to increase wall slagging, thereby decreasing waterwall heat transfer in the hope of increasing reheat temperatures upward, closer to the design value.
- In 1996, the CFS was re-instated at 7 degrees, to restore some waterwall protection.
- These changes were acceptable with the coal-only fuel blend at the time.
- It would have been useful to have looked at changing the CFS angle back to 22 degrees, when petcoke firing was started. However, this change would not have stopped the corrosion, only slowed the rate.
On-line Until 2007
It was important for NB Power to delay performing any waterwall repairs in 2006, which would have meant an extended spring outage or a second outage in the fall. Both options would have been extremely costly.
After an engineering review that included input from Alstom Engineering, the following preventative actions were taken in 2006:
- Cut petcoke from 25 percent to 15 percent (by weight)
- Reduce OFA usage by approximately 25 percent
- Modify CFS air flow back toward the walls
- Limited use of emergency carbon steel
- Pressure boundary weld overlay performed on two walls.
As a result of these corrective actions, Belledune was able to run at full load for a year until the next outage, with no forced outages. All tube thicknesses remained >0.120” so all tubes could be alloy overlaid without performing an intermediate step of carbon steel weld overlay.
Reducing petcoke usage between 2006 and 2007 cost NB Power more than $4 million, but this was much less expensive than taking a second outage later in 2006 to do the weld overlay.
Weld Overlay Investigation and Engineering
The first decision that had to be made was whether to buy new shop alloy overlaid waterwall panels to replace the damaged waterwall sections or perform field-weld overlay on the existing waterwalls.
The cost of the new panels, both in purchase price and installation cost, would have been high and the cost of replacement power for the extended outage necessary to install the new panels would have been prohibitive. Because NB Power caught the corrosion problem in time, the tubes were still thick enough to weld overlay, so field-weld overlay was chosen as the best solution for Belledune. (In some cases, especially for boiler waterwalls that are corroded too thin to allow field overlay to be performed, the option of new shop alloy weld overlay panels is still a good one.)
Selecting the alloy for the weld overlay was of the highest importance. Two key factors at Belledune were that the boiler was relatively new, with a long life ahead, and that the petcoke could be more aggressive than the typical coal corrosion.
The least expensive material to use was 309 stainless steel. A newer material called Alloy 33 was middle-range in cost and the most expensive material was Inconel 622.
309 Stainless Steel (ss): The concern with this material was that it might not offer enough corrosion protection from petcoke. It also has a higher co-efficient of thermal expansion than the carbon steel tubes.
Alloy 33: This was a newer material with a higher chrome content than 309SS, so it could offer very good corrosion resistance. However, field use of this material has just begun and it has a higher co-efficient of thermal expansion than the carbon steel tubes. For this very large job, on a relatively new boiler, this was not the time to test a new material.
Inconel 622: This material, while expensive, has a large experience base and has given good service to many utilities. It also has a co-efficient of thermal expansion that is very close to the carbon steel waterwall tubes. This material was selected.
NB Power also investigated previously undertaken weld overlay projects, the most applicable being Tampa Electric at its Big Bend Generating Station and Dominion Electric at Brayton Point. Welding Services Inc. (WSI) performed the weld overlay on these units and was awarded the work for the Belledune weld overlay.
When large amounts of weld overlay are done, the associated waterwall shrinkage should be looked at formally. There will be both a horizontal and vertical shrinkage of the boiler due to the normal cooling of the weld beads.
Areas that are not weld overlaid will not shrink and will restrain the shrinkage of the welded areas, resulting in buckling of the waterwalls to balance the two different length sections of the waterwalls.
In Figure 1, the distortion (in and out buckling) of the south wall perpendicular to the wall is shown. On the east wall section the horizontal shrinkage is parallel to the wall, which shows wrinkling at the buckstay restraints. This is all greatly magnified to emphasize areas that may be a problem. (Units are in inches.)
The buckstay system must be investigated to ensure that the shrinkage does not cause tube-to-buckstay clip welds from being torn out and causing forced outages.
WSI performed the finite element analysis (FEA) work for the Belledune boiler. The FEA results revealed the waterwalls would show buckling in the vertical direction. This was predicted mainly near the corners, where the weld overlay ended and met the rigid non-shrinking corners. There was a concern that this would load up the burner corners where the tilting burner nozzles were located, but it was determined that the wall buckling would not hamper free burner operation. Plus, the vertical buckling of the waterwalls was not a concern for the waterwalls or the buckstays.
However, the FEA did show a concern for the horizontal shrinkage of the waterwalls.
Belledune had a highly constrained buckstay system that would not tolerate horizontal shrinkage. This meant a risk existed that when the waterwalls shrank horizontally they would break the tube-to-buckstay clips. After all areas of concern were fully investigated, some clips needed to be disconnected prior to the weld overlay and then rewelded once the weld overlay was complete. This work was completed and no weld overlay-related tube leaks were experienced during the weld overlay outage, during unit start-up or in the next two years.
Many different methods of sequencing the welding were considered. Because of concern that some areas of the boiler were thinned too much to perform weld overlay, two main patterns were initially analysed.
One pattern started the welding in the thinnest areas and worked out, which would identify any unweldable areas first and allow for tube cut-outs to be planned. If this was found to be the case, we would move to welding on the thicker tubes and weld in toward the thinner tubes until the cut-out sections were reached.
The FEA showed both welding sequences to give similar and acceptable distortion results.
Before starting the job, WSI proposed a straight band welding approach to give the shortest outage possible. This third approach was checked by WSI using the FEA tool. This more productive welding sequence proved to be acceptable, based on the FEA work. This was the welding sequence that was used for the actual weld overlay.
Weld Overlay Outage
The tube thickness repair criteria was an extremely important part of this project. Large areas of the boiler were below the original ASME code-required tube thickness of 0.165 inches. However, performing an initial weld overlay with carbon steel filler metal to reach the thickness of 0.165 inches would have increased welding-induced distortion and also increased schedule and cost, with no benefit.
The final repair criteria selected was that all tubes that had a thickness of 0.120 inches and thicker would be directly weld- overlaid with Alloy 622, without having carbon steel weld overlay performed first.
Other key criteria:
- Any tubes that were thinner than the 0.120 inches would have an initial carbon steel weld overlay performed to increase the pressure- retaining boundary thickness.
- Any tubes that were thinner than 0.100 inches would be replaced with new carbon steel tubes, prior to the application of the Alloy 622 weld overlay.
After months of detailed planning with WSI, Belledune staff and NB Power Generation Engineering, the installation of 11,000 square feet of Alloy 622 weld overlay progressed without incident. WSI also had the work scope to blast all overlay areas to a white metal finish, which is key for preventing weld overlay defects caused by any slag or scale left on the tube. WSI also had the scope to perform a detailed wall tube thickness grid prior to the start of welding to confirm that all areas still had sufficient tube thickness to allow the Alloy 622 weld overlay to be performed.
All the outage work was performed, with none of the areas requiring carbon steel weld buildup or tube replacements to be performed. WSI worked with a joint crew of its own welders and local New Brunswick welders, provided by NB Power, at a ratio of 1:1. Work was done with 16 fast-deposition rate WSI automated MIG welding machines. This work was the critical path for the outage and the work was performed around the clock, seven days a week. Actual duration of the weld overlay phase was 19 days, with approximately 580 square feet of weld overlay performed each day.
WSI completed all work five days ahead of the allowed contract outage period. This included final ultrasonic thickness measurement of the alloy weld overlay thickness. Follow-up work involved a successful hydro of the boiler by NB Power and re-welding of the buckstay clips that had been cut free from the tubes, as called for by the FEA.
In the spring 2008 outage, the boiler waterwalls were inspected, after one year of full-load operation. All waterwall inspections proved to have good results, both for the weld overlay areas and the bare tubes that were left in the less aggressive corrosion areas. Therefore, there were no further requirements to extend the weld overlay onto a wider area.
This project proved to be successful for all parties involved. Since the weld overlay project has been completed, the amount of petcoke being co-fired has been increased to 29 percent.
Authors: Keith MacLean, P.E., senior boiler engineer, joined NB Power in 1991 and worked in various coal- and oil-fired generating stations as a plant engineer. Eric Fournier, P.E., is a mechanical engineer at Belledune Generating Station, with 16 years in the project management/design field. Javier Gomez-Grande is a senior project manager at Aquilex WSI. Tony Scandroli is the senior computer analyst for finite element analysis work at Aquilex WSI.
1. Craige, S. (2000). “The Combustion and Co-firing of Orimulsion at the Brindisi SUD Power Station,” Final Report, November, 2000
Belledune Boiler Details
- Unit entered operation in 1993
- 490 MW Alstom tangentially fired boiler; Combustion EngineeringAlstom
- Controlled circulation (sub-critical boiler)
- Low NOX firing with separated over-fire air ports (SOFA); CFS angle 7, instead of the original 22 degrees
- Fuel blend: coal co-fired with petcoke (5.5 percent to 6 percent sulfur)
- Total average sulfur in fuel blend 2.25 percent
- Tube material SA210 Grade C; original tube MWT was 0.180”
- Uncorroded tube thickness is typically 0.203”