The refurbishment of 10 Francis turbines at the U.S. Army Corps of Engineers' 2,620-MW Chief Joseph hydro project has resulted in turbines operating at the highest level of performance achieved for existing designs.
By Benoit Papillon and Thomas Freeman
The 2,620-MW Chief Joseph power plant, with its 27 vertical Francis generating units and energy production of 12,517 GWh in 2012, is the second largest hydroelectric power producing project in the U.S. It is operated by the U.S. Army Corps of Engineers and is located on the Columbia River near Bridgeport, Wash.
In 2007, Alstom was awarded a contract to refurbish 16 of the turbines to increase their rated power output (under a net head of 165 feet) from 100,000 horsepower to 120,250 hp and to raise turbine peak efficiency by more than 6.5%. Ten of these turbines were built by Newport News Ship Building and Dry Dock Company and six by S. Morgan Smith Company. All these units were placed in service between 1955 and 1958.
This article describes how Alstom is replacing key hydraulic components in the 10 Newport News turbines to increase efficiency and output and to extend the life of the facility. Work on the S. Morgan Smith units is planned to begin by the end of 2014.
In 2001, an Alstom research and development program targeting Francis upgrades in the U.S., and more specifically the Chief Joseph project, began hydraulic design of the new Chief Joseph units, well before the request for quotation from the Corps. Partial information on the Chief Joseph water passages and targeted turbine output increase from the Corps allowed Alstom to design a runner that would be adaptable to the geometry of either Newport News or S. Morgan Smith designs. The main parameters of the existing turbines (wicket gate pitch diameter, distributor height, runner band height, synchronous speed, net head range and targeted output) was such that the new runner design appeared to be out of Alstom's usual design family. The design required a significant increase in best efficiency discharge for such unit speed. Nevertheless, the strategy was to use optimal stay ring and stay vanes, wicket gates and draft tube designs to set a new reference for this high unit discharge design.
From this R&D program, an innovative runner design was conceived, using new techniques.1 Model testing in the Alstom laboratory in Grenoble, France, showed peak efficiency corresponded to the highest model efficiency ever measured in the Alstom laboratory at that time.
During bid preparation in 2006 and after contract award in 2007, further development was needed. Even though the best efficiency point in head and discharge estimated in the 2001 R&D program was close to the one defined in the 2006 request for quotation, it was necessary to take into account geometrical parameters specific to the Newport News turbines. Several computational fluid dynamics (CFD) analyses were performed to estimate the performance of existing components compared to Alstom's reference design and to adapt the runner to the Chief Joseph water passages. Those calculations included entire spiral case, stay ring and distributor simulations, as well as runner and draft tube CFD analyses.
Several periodic CFD calculations using fine meshes were also performed and showed a flow separation on all 23 stay vanes (see Figure 1). Modifications were designed to eliminate this flow separation and to be closer to the company's reference design performances. Using CFD, two designs were tested at the laboratory. The final one consisted of modifying the hydraulic profiles of the stay vanes by adding extensions to their leading edges in order to obtain inlet angles compatible with the flow exiting the spiral case (see Figure 1). CFD analyses also showed that changing the wicket gate hydraulic profile would increase turbine efficiency.
During the contract phase, following and in parallel with CFD analyses, a model test was performed using a fully homologous model. Measurements made included efficiency, cavitation, runaway speed, pressure and shaft torque pulsations, axial thrust and wicket gate torques. Tests were performed at plant Thoma number2 and with crown draining holes opened. Pressure pulsations were measured in the scroll case, between the runner and the wicket gates, and in the draft tube cone, at four locations.
During model testing, different configurations were tested to validate CFD analyses and to better estimate improvements provided by the Alstom components. Laboratory comparisons between the reference and Chief Joseph draft tubes confirmed results of the CFD simulations, indicating similar loss levels for Q/Qopt<1 but higher losses (by up to 1%) at Q/Qopt>1 in the Chief Joseph draft tube, where Q is turbine discharge and Qopt is turbine discharge corresponding to the peak efficiency. This confirmed the impact of flow separations shown by CFD simulations in the draft tube at high discharge.
Efficiency comparison between original and new wicket gates also showed good agreement with the CFD analyses, indicating improvements of 0.2% to 0.5%, depending on load conditions. Finally, comparison between original and modified stay vanes showed losses much higher on the model than in CFD simulations, indicating the difficulty, in some cases, of precisely estimating losses due to flow separations, especially when they can have impacts downstream from the components where they occur. Figure 2 on page 10 shows improvement from the optimized stay vane and new wicket gate profiles.
The final new turbine configuration with optimized stay vane profiles, new wicket gates and a new runner reached a high level of performance, close to modern designs. Cavitation behavior of the runner was excellent, including no inlet cavitation and a minimum 12-foot margin between sigma incipient and sigma plant, indicating the prototype runner would be cavitation-free. All other measured parameters were inside allowable limits.
Even though efficiency increase with new wicket gates was slightly lower than values Alstom encountered in some past contracts, using new wicket gates was still profitable in a short period of time when compared to the rehabilitation of original wicket gates, considering increased efficiency of the hydraulic design, the better surface finish of new stainless steel gates on the prototype, added longevity and the minimization of possible impacts on the schedule (difficult and time-consuming repairs of old carbon steel wicket gates).
Mechanical design, fabrication and installation at site
New runners were fabricated using cast components in stainless steel ASTM A743 CA6NM (crown, band, blades and runner cone). These components were welded full penetration with 410 Ni-Mo electrodes using a standard process at Alstom.3 Particular attention was paid to ensure homology of performance between the prototype and model. Half of the IEC homology tolerances2 had to be respected. Much better than that was generally achieved.
New wicket gates in stainless steel ASTM A743 CA6NM were fabricated using the electroslag remelting casting process. This process ensures a good control of the solidification rate (directed solidification), practically eliminating shrinkage porosity and segregation. All wicket gate surfaces were machined, respecting easily half of the IEC homology tolerances.
Stay vane extensions were fabricated from carbon steel ASTM A516 plates fully machined to obtain an accurate hydraulic profile. They were welded to existing stay vanes using E7018 electrodes and positioned using templates on the existing stay vanes. Templates and 3D measurements using optical instruments such as a laser tracker were used to confirm the correct alignment of the extensions.
Detailed stress analysis using finite elements calculations were performed on all the new components to confirm that mechanical stresses were inside allowable limits. On the runner and wicket gates, the procedure was standard, using as criteria the lesser of one third of the yield strength and one fifth of the ultimate strength of the material during normal operation. Under runaway condition, the maximum allowable stress was less than two thirds of the yield strength of the material.
Detailed stress analyses were needed to ensure the steel extensions on the original stay vanes were not inducing unacceptable stress levels in the stay ring or stay vanes. Conservative hydraulic loads, compatibility loads, accidental loads (impact) and vibrations were considered. These analyses showed that the stay vane modification design was safe and reliable, in particular because of the following characteristics:
- There is no connection between the stay vane extensions and stay ring shroud, to avoid imposing any relative displacement from one component to the other.
- There is no welding of the stay vane extension on the existing stay vanes leading edge when closer to 4 inches from the end of the radius area of the junction between the stay vane and the stay ring shroud.
- Special attention is paid to the geometry for the weld extremities, including the incorporation of stress relief cutouts, to reduce the stress concentration to within industrial acceptance standards.
New prototype turbine
After installation of the first unit, a startup procedure was applied to collect data on the turbine and generator. Tests were performed in no load and load conditions and during load rejections, including measurements of rotating parts runout, vibration of fixed parts, hydraulic thrust, draft tube pressure pulsations, pressure at different locations in the water passages and a friction test. These measurements showed the turbine was operating satisfactorily. The refurbished turbine was much quieter than the originals, except during partial load operation of Q/Qopt of 0.79 and 0.86, where intermittent pressure pulsations generating impulse shocks and banging noises occurred (one every few minutes). Alstom eliminated this high partial discharge instability (not observed during model testing) with an injection of atmospheric air (less than 0.1% of the turbine flow rate) induced by the installation of a runner cone extension designed to have no impact on efficiency. More on this phenomenon is available.4
Performance testing of the upgraded turbine was conducted in accordance with ASME PTC 18 Code.5 IEC standard 0416 was also invoked for some aspects of the analysis. Flow rate was measured by the pressure-time method using digital data acquisition and the ultrasonic method5 using two planes with four chordal paths in each plane. Two flow measurement methods were used because of concern that a downstream bend in the penstock was too close to the acoustic flowmeter section to satisfy test code requirements.
To ensure the downstream bend was not inducing a large bias in the flow measurement made by the acoustic method, CFD analyses of the penstock (see Figure 3 on page 10) were performed to estimate the flow measurement error due to the non-uniformity of the flow in the acoustic flowmeter section. This analysis was used to determine the best acoustic flowmeter parameters for this case. A CFD investigation was also performed to evaluate the potential usefulness of an 18-path acoustic flowmeter (two planes with nine paths per plane). Details on this work are available.7
Figure 4 shows that prototype efficiency corresponds to the transposed model test efficiency within the uncertainty of the test (about 1.37% at peak efficiency). The difference between peak efficiency from transposed model and prototype performance testing is about 0.1%. Pressure time and acoustic measurements are also consistent. For instance, at peak efficiency, the two methods indicate turbine efficiency levels differ by less than 0.1%. When compared to the original turbine, the measured peak efficiency of 95.62% on the prototype represents an increase of more than 6.5%.
After two years of operation, the first rehabilitated turbine was inspected in February 2012 to check the presence of cavitation erosion on the runner or other components. This inspection did not detect any cavitation or even any frosting over runner areas prone to cavitation (for instance, the trailing edge of the blade close to the band). This confirms the good geometrical homology of the prototype runner when compared to the model runner and the fact that this runner is cavitation-free. No (or very few) cavitation repairs will be required during its lifetime.
Chief Joseph is one of the biggest turbine rehab projects undertaken by Alstom in the past few years in North America. This project spans from 2001 to 2014 (installation of the 10th Newport News turbine) and even into 2017 (including S. Morgan Smith turbines). Alstom conceived the most efficient runner designed by the company at that time and brought the Newport News turbines to a high level of performance, increasing by more than 6.5% the turbine peak efficiency when compared to the original turbines.
Mechanical design, fabrication and installation were performed to achieve optimal value for the Corps. After putting the first unit in service, a high partial discharge instability was observed. Alstom eliminated this problem by installing a runner cone extension that enhanced aeration through the runner cone center. Performance testing on-site showed results similar to the transposed model results, confirming the good geometrical homology of the new turbine components and the high performance level measured at the laboratory. The Chief Joseph turbine rehabilitation is a success.
For their contributions to this article, the authors thank: Francois Paquet, head of hydraulic engineering, and Frederick Mathieu, hydraulic design junior engineer with Alstom in Canada; as well as Agnew Rochas, hydraulic project engineer, and Rachel Chiappa, technical tendering engineer with Alstom in France.
1Mazzouji, Farid, et al, "Refinement in Francis Turbine Design, " Proceedings of Hydro 2003.
2Hydraulic Turbines, Storage Pumps and Pump-Turbines - Model Acceptance Tests, IEC Standard 60193, 2nd edition, 1999.
3Sabourin, Michel, Louis Mathieu, and Benoit Papillon, "Francis Runner Fabrication Process: The Cornerstone of Power Plant Reliability," Proceedings of HydroVision International 2004, PennWell Corp., Tulsa, Okla., 2004.
4Papillon, Benoit, et al, "High Partial Discharge Hydraulic Instability of Francis Turbines: Understanding and Minimizing the Impacts," Proceedings of Hydrovision International 2012, PennWell Corp., Tulsa, Okla., 2012.
5Hydraulic Turbines and Pump-Turbines - Performance Test Codes, ASME PTC 18-2002, 2002.
6Field Acceptance Tests to Determine the Hydraulic Performance of Hydraulic Turbines, Storage Pumps and Pump-turbines, IEC Standard 041, 3rd edition, 1991.
7Debeissat, Frederic, et al, "CFD to Improve Turbine Discharge Measurement at Site," Proceedings of HydroVision International 2012, PennWell Corp., Tulsa, Okla., 2012.
Benoit Papillon is principal engineer with Alstom in Canada. Thomas Freeman is senior mechanical engineer with the U.S. Army Corps of Engineers.