Analyzing and Repairing a Cracked Head Cover

When a condition assessment of a turbine head cover revealed multiple cracks and anomalies, Manitoba Hydro personnel learned valuable lessons regarding the importance of records management and details of historical weld procedures, as well as challenges with cast iron repair.

By Derek Ohta

Derek Ohta, P.Eng., is senior mechanical engineer with Manitoba Hydro, a Crown Corporation in Canada. Manitoba Hydro owns and operates 15 hydroelectric generating stations, commissioned between 1911 and 2012.

Manitoba Hydro became concerned about high sump pump run times in one unit in 2008. Inspection revealed water leakage through a crack in the wetted membrane of the upper head cover. Non-destructive examination (NDE) revealed a crack greater than 12 inches long, which was repaired.

There are potential safety issues, repair challenges, and overhaul schedule risks associated with a cracked head cover. Manitoba Hydro engineering staff recommended a unit outage in 2009 to complete a thorough condition assessment of the head cover. The assessment results would be used to complete refurbishment designs prior to a planned complete overhaul to improve unit reliability.

Condition assessment

The head cover is a 21-foot-diameter, 6-foot-tall cast iron assembly comprised of upper and lower halves. The condition assessment, completed in-situ, consisted of a complete pressure wash to remove contaminants and a visual inspection of all surfaces. NDE was then completed on areas of elevated stress and areas of concern as identified by visual inspection.

This runner failed in 1927, with the runner blades sheared off the hub (at left) and the blades thrust into the head cover, punching holes in the wetted membrane (at right). The damage was repaired by welding patches using a “studded welding” technique.

Results of the condition assessment were:

  1. The crack identified in 2008 had more than doubled in length. It was previously described as a “P” shape but now had two additional crack segments propagating from the “P.”
  2. There were 11 crack indications in the wetted membrane identified using visual and magnetic particle (MT) examination. These indications appeared to be at large previous repairs, characterized by non-typical NDE indications on their perimeter weld.
  3. There was one crack visible on the dry side of the head cover wetted membrane, consisting of an extensive “spider web” of linear cracks with a total length of about 36 inches.
  4. Numerous undocumented modifications to the head cover were found, typically stiffening ribs and plates welded inside the head cover. Many welds associated with the undocumented ribs would be considered poor by modern standards, with lack of fusion being a typical defect.
  5. NDE was challenging on the cast iron due to the poor surface finish, low magnetic properties and confined working space. A variety of MT and liquid penetrant methods were tested, and wet fluorescent MT was determined to be the most accurate method of locating crack indications.

Because of the head cover condition, the unit was held on a forced outage. An engineering study was undertaken to determine suitability of the head cover for continued service and if repairs were required.

Engineering study

The head cover was supplied as original equipment in 1927, and Manitoba Hydro owned only one drawing, with no revisions. One informal document in the local files listed the maintenance history as: “1952- cracks caused by broken runner repaired.” The last major unit overhaul was in 1971, and the report described the installation of new wicket gate bushings, with no condition-based concerns.

Details of the 2008 crack repair were investigated. One important detail was that the crack extents were identified using liquid penetrant NDE, which was found to be inaccurate on this cast iron. It is possible that crack ends would be misidentified. The second relevant detail was that the crack- arresting holes were plugged with a hardwood taper dowel. A taper dowel will exert pressure on the inside of the hole, prying open the crack. The stress would increase the crack propagation rate as the wood becomes wet and swells during operation.

Manitoba Hydro’s Information Management Program Office was contacted to assist in a records search for this unit. No information was found in the files they maintain. Staff recommended contacts at an association of retired Manitoba Hydro employees. One of the association’s pursuits is a volunteer group that catalogs and maintains paraphernalia donated by retired employees. This includes historical photographs and technical notes.

Significant searching was required to reveal the 82-year history of this head cover. Photographs, taken in 1927 and 1952, were discovered that exposed the extent of damage the unit incurred during two runner failures. Both times, the runner blades were sheared off the hub, and the blades were thrust into the head cover, punching holes in the wetted membrane. The damage was repaired by welding patches using a “studded welding” technique, but the details of this technique are unknown. Photographs of the head cover repairs were analyzed and compared to the defect mapping done during the 2009 condition assessment. The result was historical, photographic evidence of the 11 repair patches and numerous undocumented stiffeners.

This linear crack, marked in red, had a “P” shape when it was first found in 2008, and extended cracking was located in 2009.

A 1930s engineering textbook was consulted to understand the 1927 and 1952 weld repair techniques. The theory behind the studded weld surface preparation appears to be that the studs provide a mechanical, structural connection between the weld and cast iron. Therefore, it is a seal weld (non-structural) at the interface to the cast iron and a structural weld at the interface to the steel stud and patch plate. This is a logical approach because cast iron welding is prone to quality issues. Eliminating structural cast iron welds significantly increases the probability of a successful repair.

Forensic analysis of cracks
An engineering consultant was employed to complete a forensic analysis of the cracks and studded weld repairs. The analysis focused on evaluation of the crack to determine the growth characteristics, supplemented by radiographic examination to understand the condition of the studded weld repairs and confirm the assumptions made based on the historical photographs.

These crack faces are from core samples of the “spider web” crack. The amount of corrosion provided a distinction between areas of older, slow growth versus areas of more recent, accelerated growth.

First, material tests were conducted in accordance with modern standards for gray iron castings. The tensile strength of the material was about 22 ksi, which is analogous to the lowest-strength, class 20 gray cast iron.

Core samples were removed from the 1.5-inch-thick head cover wetted membrane. The goal was to confirm the depth of the cracks and determine if they were stable or actively growing. Five core samples were sheared to expose the crack surface for macroscopic and microscopic examination. The result was that each crack was deemed to be actively propagating. The amount of corrosion on the crack face provided a distinction between areas of older, slow growth versus areas of more recent, accelerated growth. Most cracks were growing slowly; however, one crack (identified in 2008) was found to be propagating at an accelerated rate.

This comparison of the area where the core sample was taken shows that the extents of the porosity and steel plate are visible via radiography (at right) and are not visible in the photograph (at left).

Contractors advised that MT crack indications are often found during NDE of cast iron welds. They believe the indications show the interface of the dissimilar metals rather than being an indication of a crack. The contractors also said welding this vintage of cast iron can be difficult due to casting quality issues such as porosity, sand/oil entrapment, and inconsistent metallurgy.

Radiographic examination was required to confirm the condition of the studded weld repairs and verify the MT examination results. Wet florescent MT was deemed to be more effective than liquid penetrant, but there was still increased uncertainty in the results. It was important to accurately map out the end of cracks to drill-arresting holes, and radiographic examination provided an alternative to verify the MT results.

Radiographic examination confirmed that the studded weld repairs were made consistent with the 1930s textbook information and historic photographs. The radiography also showed no issues with the studs, tapped holes or welds used to make the repairs.

An interesting discovery during radiographic inspection was that some welds contained significant contaminants. In radiography, these contaminants appeared to be studs, nuts and scrap material laid parallel to the weld but not mechanically fastened to the base or patch material. Inserting these contaminants would significantly reduce the amount of welding required. Radiography revealed a potential quality control issue from the past, but there was no radiographic evidence of cracks in these welds.


Manitoba Hydro implemented the following repairs:

  1. Historical studded weld repairs were left as-found because they appeared to be in good condition.
  2. Two areas of linear cracking of the wetted membrane had arresting holes drilled at all crack ends, with the location verified to be at the end of the crack by radiography and MT examination.
  3. Large core sample holes were plugged with clearance fit dowels suspended from a steel plate fastened to the inside of the head cover. Smaller crack-arresting holes were plugged with a straight thread (not tapered) plug. Both plug types limited the crack propagating stresses in the cast iron.
  4. The unit was re-inspected periodically for six years to verify the measures were effective and no new cracks were forming. No significant changes were found in the historical repairs or cracks.


Special thanks to Dr. Wayne Tennesey of Test Labs International and Chad Winzinowich of Certified Inspection Services for their expertise. Gratitude goes to Hal Turner and my Manitoba Hydro colleagues for their peer reviews and guidance throughout this challenging project. My wife and daughters deserve special acknowledgement for their unwavering support.

More HR Current Issue Articles
More HR Archives Issue Articles
Previous articleHow Seattle City Light Personnel Became Fire and Rescue Experts
Next articleChina’s Three Gorges hydropower plant accused of causing landslide

No posts to display