Time is as important as cost when deciding whether to run, repair or replace a failed rotor. As the owner/operator of a geothermal steam turbine learned last fall, waiting until completion of the failure analysis before beginning the procurement process can result in the foreclosure of several options. Although the owner/operator did manage to get a replacement rotor online within the scheduled outage window, along with some design and material improvements to prevent future losses, if it had begun the process earlier, it could have obtained further enhancements.
The turbine in question is a 40 MW geothermal steam turbine installed at a North American facility. Steam enters the turbine through a single admission 24-inch inlet at 270 psig and 413 F. The steam path consists of four high pressure (HP) and five low pressure (LP) stages. Two uncontrolled extractions follow the second and fourth stages.
In July 2003, a turbine overspeed incident compromised the rotor’s integrity. Plant instruments indicated the rotor accelerated to 5,770 rpm, more than one-and-a-half times the rated 3,600 rpm, resulting in two-and-a-half times the normal stress levels. The plant swapped the damaged rotor with a spare and put the unit back into service. It then shipped the original rotor off to the OEM for evaluation.
The damage was extensive:
- A wet magnetic particle examination revealed an anomaly on the fourth stage in the root area of one bucket and numerous anomalies in the L-0 tie-wire braze joints.
- Finite element analysis (FEA) indicated that some of the wheel fillets experienced local yielding, increasing their risk of stress corrosion cracking.
- One L-0 stage bucket cover was missing.
- The bucket lift check discovered measurable gaps on stages 2, 7 and 8, indicating some localized tenon yielding. (Figure 1)
- Ultrasonic examination of the L-0 finger pins revealed numerous anomalies.
- Phased array testing found an indication on stage 2 under one bucket and in stage 4 dovetails.
- Further FEA suggested that significant yielding occurred in the L-1 and L-2 stage dovetails, probably resulting in large areas of yielded material. In addition, the analysis demonstrated that the L-0 wheel dovetail was stressed to the material’s yield strength.
Figure 1. Stage 7 Bucket Lift
Repair or Replace
After determining the extent of the damage, the owner/operator had to decide whether to repair or replace the rotor. The cheapest option was to keep the currently running replacement rotor as the primary rotor, perform minimal repairs on the overspeed rotor and use it as an emergency spare. However, this minimal repair approach was not financially attractive because:
- Vendors would not offer an adequate warranty since numerous highly stressed rotor sections would not be inspected and some of these areas had yielded during the overspeed incident.
- Firm pricing could only be obtained on the known scope of the job. Pricing for worst-case scenarios approached the price of a new rotor.
- The currently running spare rotor’s historical records revealed that it had numerous problems and was not a reliable long-term replacement.
After eliminating the option of only minimal repairs, the next choice was to perform complete repairs. While the anticipated repairs would cost more than half the price of a new rotor, they could be completed before the next scheduled outage. This option would also allow the metal to be upgraded in the bucket attachments in the four stages. However the bucket attachment wheel fillets that were not repaired would still be at some risk in a geothermal environment.
The owner/operator decided to replace the rotor. Normally, the original equipment manufacturer (OEM) could provide a direct replacement for the damaged rotor, but not in this case. Although the OEM had originally submitted a proposal for a new rotor, the owner/operator did not initiate the procurement process until it completed the testing and evaluation. It finally placed the order with the manufacturer in late October.
During the proposal phase, however, TurboCare Inc., Chicopee, Mass., located a forging in Germany with a suitable configuration envelope, and it also could meet the tight schedule. But in addition to meeting the schedule, the rotor metallurgy had to meet the strict requirements for operating in the harsh geothermal environment, including high toughness and ductility, low sensitivity to inter-granular stress corrosion cracking (IGSCC) and a low corrosion rate. Finding the rotor composition to be equivalent or superior to that of the original rotor material for a geothermal application, the owner/operator contracted with TurboCare to provide the rotor by the outage start date of April 2, 2004. The contract included liquidated damages for failure to achieve the contractual delivery date.
“The extremely short delivery requirement was the project’s biggest challenge,” said James Beverly, TurboCare’s senior product manager. “To meet the compressed schedule, the rotor forging was transported via a jet from Germany to New York. Although it arrived on schedule, a severe winter storm delayed its departure from New York by one week – time that had to be made up in other steps of the manufacturing process.” TurboCare reverse engineered the original rotor while the forging underwent completion of gashing (making the initial cuts in machining a gear), final heat treatment, nondestructive examination (NDE) and shipment.
“Although the delivery requirement did not allow much time for redesign, engineers had enough time to make some incremental improvements in rotor reliability,” said Beverly. A unit inspection revealed a cracked dovetail on a stage 4 bucket adjacent to the notch. TurboCare redesigned the bucket to prevent future failures, enlarged the dovetail fillet radius to reduce peak stress, and tuned the redesigned bucket to avoid low per revolution resonance. A redesigned damping system for that stage and the last stage decreased vibratory stress.
“We also upgraded the bucket materials to lower operating stress and combat the wet steam erosion and contaminant-driven erosion/corrosion common in geothermal applications,” Beverly said. Engineers used titanium for the new design stage 4 covers, the stage 7 buckets and covers, and the stage 8 and 9 covers. Stellite erosion shields were added to the leading edge of the stage 8 buckets to combat severe erosion. In addition, the gland steam shaft seal lands, which had eroded, were clad welded with Inconel 625 using a submerged arc method (Figure 2).
Figure 2. Machining Gland Steam Shaft Seal Lands in Inconel 625 Clad Material
Once machining operations were completed, the next five weeks were spent installing blades. The titanium buckets on stages 4 and 7 required “hot peening,” a specially developed method of preheating the tenon. Specific thermal gradient maintenance through the length of the airfoil was required to prevent work hardening of the tenon caused by the peening operation. Next, the bucket covers were installed and machined on an engine lathe.
“After final NDE, operating speed balance and overspeed testing, the rotor was shipped to the site for installation ahead of schedule,” Beverly said. “Further, by working closely with the customer to understand the exact requirements, we were able to provide a stronger, more erosion and corrosion resistant rotor than the original OEM version.”