Africa, Asia and Oceania, Dam Safety, North America, Turbines and Mechanical Components

Roundtable Discussion: Mitigating Seal Failures

Issue 10 and Volume 36.

Rehabilitation and Repair

Personnel from power producers in Thailand, the U.S., Canada and South Africa share their experiences with seal failures and how these problems were resolved.

Owners of four hydroelectric plants and dams share experiences with seal failures, including what was affected and corrections taken. Facilities are:

  • Thailand – Palapum Khunthongjan, engineer for Electricity Generating Authority of Thailand (EGAT), shared data on the 500-MW Lamtakong Jolabha Vadhana (LTK) pumped-storage plant on the Lam Takhong River in Nakhon Ratchasima province.
  • U.S. – Karen Case, P.E., senior mechanical engineer for Sacramento Municipal Utility District (SMUD), shared information on the 150-MW Camino plant in Camino, Calif. The plant is part of the Upper American River Project, which includes eight powerhouses containing 11 turbines.
  • Canada Personnel at a Canadian generating station, who asked to remain -anonymous because of ongoing concerns, contributed their experience dealing with seal failure.
  • South Africa – Leon Tromp, P.E., chief engineer with Lesotho Highlands Water Commission, shared data on sealing between two concrete slabs of Mohale Dam, a 145-m-high concrete-faced rockfill dam (CFRD) in Lesotho, South Africa.

Q: What type of seal failure did you experience?

Thailand: We experienced seal failure on turbine shafts.

U.S.: SMUD dealt with failure of a carbon segmented turbine shaft seal.

Canada: Our issue dealt with placing incorrect Kaplan blade seals on a turbine repaired offsite from the hydro project.

South Africa: The face slab in the middle of Mohale Dam on the joint between slabs No. 17 and 18 cracked and necessitated replacing joint seals between the slabs.

Q: What were the initial circumstances that led you to suspect a seal failure?

Thailand: The LTK Unit No. 1 seal started to break on the outside of the carbon ring leading to water leakage.

U.S.: On Unit 2 at Camino, personnel noted excessive water in the turbine pit, combined with sediment.

Canada: Our operator noted oil leaking into the river. The leak was minimal, but on the river, it was visually severe and we estimated the amount leaked was about 2.5 liters.

South Africa: In February 2006, Mohale Dam developed a compressive vertical crack on the downstream face between slabs No. 17 and 18. The crack started from the top and propagated down to the bottom of the face slab, following heavy rains that started in January. Although the reservoir water level was almost constant at 2,075 masl for the next month, the seepage rate almost doubled from 350 L/s to 600 L/s in one day.

Q: What did you determine was the reason for the seal failure?

Thailand: The shaft seal failure was due to either low-quality original equipment manufacturer (OEM) construction or adhesive glue degeneration around the carbon ring.

U.S.: The Unit 2 seal rings failed at Camino after a major outage. The seal rings were replaced and inadvertently fabricated to an incorrect dimension. This resulted in hydraulic instability and high vibration such that the stationary and rotating rings made contact. The reason for the seal failure was a combination of seal wear and insufficient cooling water supply to the packing box, which would have prevented tailrace water pressure inundating the seal with sediment.

Canada: Several months earlier, the turbine was shipped to the contractor’s shop to undergo a repair. It appears the root cause for the incident was that information on hand for replacement seals at the project was correct, but the contractor had information that differed from that. Based on this incorrect data, the contractor acquired seals from its supplier.

Additionally, we do not think the contractor did a thorough enough pressure test on the turbine after the seals were installed to ensure the turbine worked correctly without leakage or wear on the seals. It appears the job was rushed and not enough diligence was given to make sure the seals could not fracture.

South Africa: Movement occurred on the dam embankment, mainly on the upstream face, causing concrete spalling, and shearing damaged steel rebar in the concrete slab on the vertical joint. The centroid (center of mass of a geometric object of uniform density) of the curve moved almost 500 mm towards downstream. A 42-mm cross-valley movement from each side was measured on the crest in the center of the dam.

Q: How did you respond, and what plan was implemented to mitigate the failure?

Thailand: EGAT contracted a product/service provider to dismantle the unit, and EGAT personnel along with the contractor inspected the broken seal. We conducted research on material types used for seals that would replace the damaged seal. EGAT compared advantages and disadvantages of seal properties that used polymer materials because of the following considerations needed for the unit’s shaft: high running speed, abrasions, high compression and immersion.

U.S.: SMUD replaced the carbon segments and springs and increased the pipe size for the cooling water to allow higher flows.

Canada: The seal vendor and the contractor worked together to develop a seal that would rectify the issue.

South Africa: In March 2006, Lesotho Highlands Development Authority (LHDA) appointed a panel of experts to investigate the cause and extent of the crack and assess the safety of the dam structure. We decided to design the compression joint between slabs 17 and 18 above the water line. But for underwater rehabilitation, we used other solutions that included using divers and dumping sand and gravel to fill the orifices in the shear failure zone.

Work began on March 17, 2008, and was completed six months later at a cost of US$1.04 million.

Q: How did you choose replacement seals, including material and technical considerations?

Thailand: EGAT concluded that the polymer material group including ultra-high-molecular-weight polyethylene (UHMW-PE), elastic polymer or fiber-reinforced phenol resin are preferable for use with the shaft seal, instead of antimony-impregnated carbon.

U.S.: We used OEM specifications for material and fabrication. However, we are very interested in a future modification of this seal due to the relatively short life of the carbon segments.

Canada: We allowed the contractor to rectify the problem with the seal supplier/manufacturer.

South Africa: The replacement seal (see Figure 1) was similar to the original joint seal, which was made up of a U-shaped copper plate that has an inverted U in the middle to allow for horizontal displacements. In addition to the original design, the seal repair included installing a rubber water stop.

The dynamic sealer used for the top part can accommodate 30 m of water pressure. A special seal was also developed for deeper joints.

Figure 1 — Joint Seal Replacement at Mohale Dam in South Africa

This figure details the components involved in repairing the crack and replacing the seal between concrete slabs No. 17 and 18 at Mohale Dam.

Q: What was the outcome of replacing the seal(s) in terms of time to return the equipment to service?

Thailand: The turbine shaft seals of Unit 1 and 2 at LTK are constructed with antimony-impregnated carbon. During the period of operation from 2004 to 2013, there were 13,200 hours of unplanned outage. We learned there were 11 outages due to failure of these seals, which amounts to about 12.76%, or 1,685 total hours of the unplanned outages. Correcting the seals issue allowed LTK to reduce 168 hours of loss each year.

U.S.: After replacing the seal on this unit, leakage into the turbine pit dramatically reduced. It took about three weeks to disassemble, replace and reassemble Unit 2.

Canada: This event happened some years ago, and the production loss was significant for us. It took about four months to solve the issue and place the turbine back in service.

South Africa: The repairs covered a 17-m-long section of the crack above the top of the catchment’s water level. The seepage decreased after rehabilitation from 600 L/s to 500 L/s when the water level reached full supply level. At lower water levels. seepage decreased to between 300 L/s and 400 L/s.

Gregory B. Poindexter is associate editor for Hydro Review.