Using Rapid Prototyping Methods to Repair Runner Cavitation Damage

A new technique developed by Ontario Power Generation allows for accurate, quick repair of cavitation damage on a runner blade. The technique is based on rapid prototyping methods and shows promise for other hydro plant applications.

By Darrell Lewis

To overcome limitations involved in performing repairs of turbine runner blades suffering from cavitation damage, the author developed a new repair technique based on rapid prototyping methods. This technique consists of using a portable scanner to produce a three-dimensional (3D) image of the runner blade surface that can be downloaded to a computer. The data can be analyzed and used to build a replacement piece that is then welded to the existing turbine runner.

Ontario Power Generation (OPG) has applied this technique twice, with good results. The first application showed that the technique was feasible and provided good correlation between the replacement piece and the damaged runner blade. The second application, at a working hydro facility, allowed OPG to continue running the unit while the replacement parts were manufactured. The utility plans to use this new technique for future cavitation repair work at its hydro facilities.

Standard cavitation repair techniques

Small bubbles can form at a pressure less than the vapor pressure of the water. If the vapor bubble collapses near the runner surface, highly localized pressure forces can remove runner material. This phenomenon, known as cavitation, can result in significant damage to a turbine runner.

OPG operates 65 hydro stations with a total capacity of 6,963 MW. The smallest station has a capacity of 1 MW; the largest more than 1,400 MW. Repair due to cavitation erosion is the most frequent maintenance task OPG personnel must perform on turbine runners. Historically, the average interval between repairs has been about 12 years. However, this can be highly variable among units as newer runner designs have been greatly improved to reduce or even eliminate cavitation.

For runners that experience cavitation damage, repair typically begins with an inspection to identify the damaged area(s). Personnel then must grind the damaged area to remove the pitted surface. Frequently, this damaged area can be far larger than what appears upon first inspection. Once surface preparation is complete, the lost volume of the runner is built back up using weld overlay. Finally, the surface is ground back down to best approximate the original runner surface.

Many factors make this repair method time-consuming and problematic:

• Cavitation damage typically occurs on the downstream side of the runner blade. On vertical turbines, this involves an overhead weld deposition, which increases the difficulty of the repair;
• When adding material, the welder must make a best judgement of the original runner profile and usually has no template. Grinding the weld material to best approximate the original profile is time-consuming and can be inaccurate;
• The weld must be layered to a depth greater than the original profile to allow the worker to grind back to a smooth surface. Stainless steel, which is frequently used for weld overlay, is difficult to grind;
• Weld pool shrinkage during the overlay can cause runner profile distortion. A strongback or brace typically is required to help limit this movement, but shrinkage stress stills exists and the final shape is uncertain when the brace is released;
• Stress-relieving techniques are difficult and probably will not correct weld-induced distortion;
• Overlay welding can generate a large volume of welding fumes in a confined space; and
• Runner blade distortion is proportional to the amount of weld deposited on the runner. A large weld repair likely will result in greater runner blade distortion.

Developing a new technique

Based on new rapid prototyping and 3D modeling methods used in manufacturing and other industries, OPG personnel decided to investigate the possibility of utilizing some of these methods for repairing hydro turbine runners and other complex surfaces, such as spill rings and wicket gates.

OPG wanted to improve in-situ repair to achieve five goals:

• Increase the interval between cavitation repair outages;
• Decrease unit outage time during the repair;
• Minimize runner blade distortion as a result of the repair;
• Maintain the original runner surface contour; and
• Eliminate the need for a strongback during welding repair, as this device typically is heavy and awkward.

The technique OPG developed is based on use of a portable stereoscopic handheld laser scanner. This scanner projects a laser beam onto a surface, captures reflected photons from this beam, and creates a geometric shape or “point cloud” that represents the 3D surface being scanned. Data point position is determined by referencing to an internal coordinate system created by adding randomly positioned targets affixed to the runner surface.

This initial graphics exchange specification (IGES) file represents the surface of a runner after it was ground to remove cavitation damage. This file is used to create another IGES file that represents a 3D plug or insert needed to restore the runner to its original profile.

The completed scan has a resolution of +/- 0.004 inch. The resultant data file also depends on the area covered and data capture rate.

Because of the large quantity of collected raw data, the data is converted to a smaller polygonal mesh stereolithography (STL) file for easier use. The polygonal mesh is a series of triangular faces and edges that represent the surface to be modeled. The smaller the mesh size, the higher the resolution and the better the creation of a smooth appearance of the model. A laptop with enhanced memory and some computer-aided drafting (CAD) capability can usually suffice to do the work.

The STL file is then converted to a neutral file format known as initial graphics exchange specification (IGES). This format is common in 3D CAD software packages. From the IGES file representing the runner surface, the software technician then can create an IGES file that represents a 3D plug or insert. The portion of the insert that mates with the runner pocket is obtained directly from the runner surface data.

The outer surface of the runner insert is determined by analyzing the surrounding non-cavitated runner surface. This surface information allows the developer to design a smooth contour for the new insert that best matches the original runner profile. Additionally, an edge bevel and other features can be designed to facilitate the final installation.

At this point, the IGES file of the runner insert is complete and can be sent to vendors to create the actual solid from the material of choice, either by casting or CNC machining.

This model of an insert (foreground) needed to repair runner cavitation damage was machined using a CNC router. The holes provide locations for plug welds when the insert is welded onto the damaged runner.

Applying the technique

OPG personnel performed the first test of the new technique on a discarded propeller runner at the Caribou Generating Station. Personnel removed a portion of the runner blade, transported it to the maintenance shop at the dam site, and selected an arbitrary region to repair. Personnel then ground this region to replicate a typical cavitation repair job.

OPG hired Agile Manufacturing in Uxbridge, Ontario, Canada to scan the runner area and existing surface surrounding the damaged area. From this scan data, an IGES file was created that provided OPG with a true representation of the repair surface and undamaged peripheral surface. This latter data is needed to develop the “missing” runner surface that was eroded. Using various computer software programs, such as Solidworks and Rhino3D, OPG personnel could then produce a 3D insert that both fit the runner void and closely matched the existing runner surface.

The outer surface contour of the insert was approximated by referencing and extrapolating scan data from the undamaged periphery surrounding the void using the 3D modeling software.

The file representing the insert was now complete, and OPG used the computer model to simulate a test fit. The completed file was used to produce a CNC machine file. A computer program was written for a five-axis milling machine and, once complete, the newly created machine file was uploaded. Agile Manufacturing machined the part from an aluminium billet for ease of milling. The completed test piece was fitted to the mock runner section, and the fit was dimensionally correct, with good correlation with the runner surface.

Building on the success of this experiment, OPG decided to use this technique to repair an in-service turbine. Personnel selected a 107-inch-diameter Morris turbine runner at the 18.5-MW Ear Falls Generating Station.

The turbine was shut down in the fall of 2008 for routine maintenance, and inspection revealed cavitation damage to the runner blades and throat ring.

Fitting a machined insert to the damaged turbine runner for which it was created shows the good correlation resulting from Ontario Power Generation’s technique.

OPG personnel selected two runner blades and about a third of the throat ring section for repair. Eight locations were found to have severe cavitation-induced pitting. The damaged areas were ground to remove the pitted surfaces, then a scanning technician with Agile Manufacturing scanned the area. When scanning was complete, in about four hours, the unit was returned to service.

Once the eight runner insert IGES models were developed, the completed files were e-mailed to MA Steel Foundry in Calgary, Alberta. Site staff then created a casting mold and produced a sand casting. Details such as casting shrinkage had to be compensated for to produce an exact match with the computer model. MA Steel Foundry used a software program called Magmasoft to simulate and optimize the casting pour to eliminate any possible voids and defects. For this test, the runner pieces were cast from 316L stainless steel.

In the spring of 2009, the unit was shut down to install the new runner inserts. During the six months the unit had run, minor increased cavitation damage occurred in the areas to be repaired. Despite this, the cast inserts fit well, and only minor grinding was necessary to fit all pieces. The inserts were edge bevelled, then welded into place by OPG personnel. On larger inserts, personnel drilled a few randomly placed 1-inch-diameter holes to facilitate plug welding.

During this shutdown, OPG also decided to test the new repair method under real-time conditions. Site staff selected three runner blades with cavitation damage. The damaged surfaces were ground and scanned on May 21, 2009. New inserts were designed and were e-mailed to the foundry for casting. The pieces were then created by the investment cast process from 316L stainless steel. The completed parts were shipped from the foundry on May 28, 2009. Thus, total duration from start of scanning to completed part was just seven days. The new inserts were welded in place the following day, and the unit was returned to service

After a year in service, OPG shut down this unit to inspect the repaired runners. No visible damage was noted using a diving team equipped with video cameras. The unit was returned to service.

As a result of these tests, OPG personnel made several important observations:

• Maintenance time required for this runner repair using this technique was about 30 percent shorter than with the former method. This figure will vary based on the damage extent and number of repairs required;
• Creating inserts via machining works well for one-off parts, while casting is the preferred choice to produce volume quantities of parts;
• Runner blade distortion was minimized due to decrease in volume of weld deposition. The runner insert itself acts as a brace to help prevent distortion;
• It was easier to maintain the original runner surface. By developing the 3D model and scan data, OPG was able to create an accurate representation of the original runner profile. Using the old technique, the welder would have to build up sufficient weld to be able to grind back to create the runner contour. This task was based on judgment, and there was no practical method to determine the qualitative success of the finished shape;
• Working conditions were better because less welding was required and there was no need for additional runner blade bracing;
• Minimum insert thickness should be 0.25 inch. This makes it easier to create the part by casting or machining. Thin sections also are more difficult to weld securely to the runner surface. Thus, it is worthwhile to grind the damaged areas to at least the minimum thickness when preparing for scanning;
• Slightly damaged areas can be repaired using weld overlay; and
• Because the runner inserts are not fully fused to the runner, qualified judgment is mandatory as this method will not be suitable for high-stress areas.

OPG personnel also made the following observations, which can be used for further development of this technique:

• If metallurgical properties are compatible and proper welding techniques observed, there are opportunities to use insert materials that resist pitting by cavitation (such as inconel, stellite, and various grades of stainless steel);
• Cavitation tends to occur in the same locations. Thus, OPG may be able to replace inserts without additional scanning and model development by using previously collected historical data;
• By scanning and evaluating the runner dynamics, OPG personnel may be able to determine if the shape of an existing turbine runner blade can be modified to improve unit performance and/or cavitation resistance. If this is the case, replacement inserts can be shaped to reduce or eliminate cavitation;
• The turbine can be run longer between maintenance shutdowns. Runners subjected to cavitation need to be inspected frequently, and it is desirable to repair before the cavitation penetrates too deeply. If repair intervals are extended, the weld repair will be extensive, and the potential for runner blade distortion increases proportionally with the weld pool volume. Using the runner insert technique allows extension of the maintenance interval as long as runner structural integrity is maintained. Extending maintenance intervals by a factor of two may be possible using the new technique, as runner blade distortion is unaffected by the size of the repair;
• Scan data can be used to produce exact copies for other work. For example, a duplicate runner section can be fabricated from any material, and this copy can be set up in the maintenance shop. New parts — such as anti-cavitation lips on the runner tips to prevent unwanted vortexing — can be designed, prefabricated, and test fitted on the runner copy so that they are available for installation during a future shutdown; and
• Scanning the runner before repair establishes a base reference for comparison upon completion of work.

Conclusion

OPG’s experience with these two trials shows that this new technique meets the utility’s original criteria and is quite versatile. OPG has greatly reduced time spent at the runner, as most of the work is done using a computer. The utility now has a relationship with vendors to provide the services required to perform this technique quickly and efficiently.

For OPG, the maintenance staff have a much simpler task in just preparing the damaged runner surfaces for scanning. A technician can complete the scan within about four hours. Then an outside vendor completes the modeling and production of new parts. And company maintenance staff formerly dedicated to the runner repair can be assigned to other work. Once the new runner inserts arrive on site (in about ten days), OPG was able to install the pieces over a period of about two days.

There may be other opportunities in the future to use 3D scanning at hydro facilities. For example, this technique can be employed in areas affected by erosion (such as head gates, penstocks, and wicket gates) and for assessing any general condition by scanning the area in question. A big advantage is the ability to rapidly capture a precise 3D representation of plant condition in difficult to access areas. The unit can then be returned to service and the collected data be transferred to the computer. Upon completion of the computer analysis, repair procedures can be developed and components fabricated in the shop, thus greatly reducing plant downtime.

Darrell Lewis, P.Eng., is a senior plant engineer with Ontario Power Generation. He developed the concept for the technique described in the article and identified companies that could perform the work.

More Hydro Review Current Issue Articles
More Hydro Review Archives Issue Articles