Storming ahead2 March 2010
Rainpower has developed its own Francis turbines using a parametric approach and is further expanding the range while also preparing to launch new Pelton designs. Report by Patrick Reynolds
Rainpower has been making rapid, strident advances in research and development for its new range of turbines, initially low-to-medium head Francis, branded ‘Rainpower Storm’, and it is preparing to extend the parametrically-based analytical approach to high-head designs to establish a further range under, it is expected, another brand name.
The surge in R&D should also see the young, privately-held Norwegian company get ready to launch a new Pelton range of turbines in the coming months.
The sweeping approach to R&D arises from its seasoned engineers employing ever-increasing computing power to establish an independent portfolio by pursuing, usually from first principles, a parametric-based approach to develop suites of turbine designs and models. They have been determined to renew their own basic approach to analysis and design, and so create a fresh offering in Francis and Pelton turbines.
A key aim of the fundamental approach is to use a parametric concept where all the main dimensions for all components are based on continuous function of Unit Speed (N11) and Unit Flow (Q11), and so interpolate between models and more efficiently determine further, intermediate designs.
Research into the parametric approach began immediately after Rainpower was formed in midst of merger and acquisition activity in the industry (See box: ‘Route to Rainpower’). The first test of a parametrically-derived model took place within less than a year, in September 2008. The first customer witness test was performed before the end of the year. In total, four new Francis ‘Rainpower Storm’ models have been developed and tested, for the plants at Tjørhom in Norway, Pankou and Maoergai in China, and Keban in Turkey.
While its experienced designers had the vision for the parametric concept, setting out with the new business in a competitive market required that they had focus, too, on continuing to utilise a local, 10-year technology licence for existing high head technology, arising from a legacy relationship. They would be working on two distinct technological fronts, one for high head and the other for the low-to-medium range.
Legacy & licence
While preparing to take their research on parametric concepts for turbine design into high head turbines in the near future, Rainpower’s engineers have been working more traditionally with Kværner-originated technology where the licence applies – in Norway.
A number of renovation, modernisation and upgrade (RMU) projects for high head plants are underway, such as at Svartisen, Hol 1, Hodnaberg and Rendalen. The work involves both refurbishing existing units and supplying new equipment.
At Svartisen in the north of the country, the plant houses one of the few large high head Francis units in the world, and considered to be on the upper boundary for design technology, says Rainpower. The single, vertical axis turbine was designed and fabricated by Kværner and installed in the 1990s. It has a rated head of 543m, maximum net head is 585m and rated speed of 333rpm to give a nameplate capacity of 350MW. The runner is 4.4m diameter and weighs 36.5 tonnes.
Svartisen – turbine erosion
However, the turbine has suffered erosion from highly abrasive sediments, the flow to the plant being meltwater brought by a tunnel network from a nearby glacier. The problem for the plant – its name meaning ‘black ice’ – comes despite settlement caverns built within the tunnels. Significant wear was found after eight years of operation.
The load was considered moderate compared to those experienced in other regions, such as the Alps, Himalayas and Caucasus region of Central Asia, but apart from some traditional Kværner features that help in silt-laden flow – such as splitter vane runners, guide vane stem seal construction and a free draft tube cone – there were few anti-erosion measures. The main owner of the scheme, Statkraft, then learned that after 10 years of operations the problem had reduced efficiency of approx. 2.5%, as measured by the thermodynamic method. Damage was found to guide vanes, turbine cover, runner and labyrinth seals (wear rings). Plans for a major rehabilitation programme were restricted to 14 weeks to limit the economic impact of the shutdown. The plant, the last of the succession of relatively large hydro schemes in the country, is critical to the grid and supplies in the region.
GE, having bought Kværner, signed a refurbishment contract and the work was completed in 2003, including installation of new guide vanes and sealing rings. It also coated covers and vanes with an onsite-applied, tungsten carbide thermal spray (high velocity oxy fuel – HVOF). The programme, however, only provided for minor welding and grinding repairs to the runner. Efficiency and visual tests in February 2005 some two years later confirmed the integrity of the refurbished unit, although it wasn’t a long-term solution.
Replacement Runner, New Unit
Following the hydraulic machine difficulties, however, the plant soon after suffered problems with electrical equipment. There were two major shutdowns in late 2006, the first a knock-on effect of the local grid short-circuiting in a storm which overloaded the generator, and overheated the stator. A temporary repair was done but after some months a smoke detector tripped the unit and new problems arose.
Eventually, there would be approx. 6.5 months of outage due to the problem.
The experiences with the turbine and then electrical equipment emphasised to the utility the vulnerability of having a large single unit facility. Statkraft planned, therefore, for repairs to the stator as well as other improvements at the plant – installation of the previously envisaged, but delayed, second generating unit and also purchase of a reserve Francis runner for the existing unit.
As it was established, the company received, first, a contract to design, manufacture and supply the new turbine, and also provide the main inlet valve, governor and perform installation and commissioning. A 250MW unit was required, operating to the same nominal head though with a synchronous speed of 375rpm. The model witness test for the new runner took place in April 2009. It is to be operational in 2011, and will take the plant’s installed capacity to 600MW. The generator for the new unit is being supplied by VG Power, a subsidiary of Voith Hydro.
Then, in mid-2008, the company won a further contract to supply the replacement Francis runner plus a new stationary labyrinth ring and undertake model testing. The main operational data are unchanged, apart from more emphasis on overload operation due to the nature of the local grid. The runner is designed with a higher efficiency than the original and weighs 34 tonnes, not as heavy as the first unit. Factory acceptance was in September 2009 and the unit is to be online this month. All hydraulic equipment for the plant is being manufactured with strict supervision from Rainpower’s quality engineers at the Hangzhou office in China.
Both designs, for the new turbine as well as the replacement runner, are improvements on, though still under, the licensed Kværner/GE design technology, notes Rainpower technical manager Bjarne Boerresen, who was previously with Kværner and GE Hydro, in Norway.
A further challenge at Svartisen is the control system for a combination of reasons: high head and long tunnels presenting issues of water hammer and governing stability; topographical restrictions below the glacier preventing use of a normal-sized surge shaft; a requirement for isolated grid operation resistive load; and, occasional overload conditions.
In addition, and unusually, an added complication is the turbine being the source of the waterhammer between the main intake on the tunnel and a large surge shaft on a long branch tunnel, closer to the plant. Even slight guide vane movements can cause pressure transients to travel and excite the hydraulic system, which can then lead to disturbing oscillations in turbine speed, especially at the system’s critical frequency range, unless the governor response is slowed. Also, waterhammer is more pronounced at low flow rates, creating the resonance problems that may be poorly dampened.
Last year, Rainpower researchers presented a paper to an IAHR workshop to outline two possible special governor algorithms – Pressure Feedback (PF) or Water Column Compensation (WCC) – to help address the special circumstances for control systems dealing with water hammer and/or pressure waves, such as at Svartisen. In aiming to achieve both stability and speed of response in governor design, the concept is to filter-out, either actively or passively, the data feed on transients within the critical frequency range.
The researchers conclude that stability analysis of difficult hydraulic systems, such as at Svartisen, show the importance of using both frequency and time domain studies. The use of PF or WCC could increase where plant upgrades are proposed and stability in isolated grid mode is demanded, the researchers add.
Beyond Svartisen’s particular challenges surrounding the use of the established, licensed high head turbine technology, Rainpower is preparing to developing its own high head designs for Francis turbines, based on the parametric approach. Its initial R&D focus, however, has been to use the parametric approach for low-to-medium head turbines (design heads of 80m-300m).
The resulting ‘Rainpower Storm’ turbines are to be used in domestic RMU projects, like Bjelland, Haaverstad and Bagn, and internationally at schemes such including Keban, Maoergai and Pankou.
The first ‘Rainpower Storm’ model test (N11=60) was undertaken at Rainpower’s commercial risk for the proposed project at Sira-Kvina’s Tjørhom extension scheme, in Norway, but which then was postponed, although partly built, due to the economic crisis in 2008. It is expected that the first turbine to be installed will be at Pankou (N11=70) this year.
The Pankou plant will be equipped with three units, each 255.1MW and operating under a rated head of 83m. For Pankou and Maoergai (N11=50), the company was hired to work in collaboration with local manufacturer Zhejiang Fuchunjiang Hydropower Equipment.
The Parametric approach
In principle, it may seem easy, but the parametric approach has taken quite a lot of man-hours to develop, says Boerresen. As such, to invest in and pursue such research took a major strategic business decision. “The biggest lesson has been that by freeing oneself from previous design experience and going back to the ‘roots’ – basic physical relations – permits new development,” he adds.
The parametric research began from fundamentals, looking at equations for fluid mechanics and turbo machinery, as well as public information on loss mechanisms and the influence of basic design parameters. The information and data on main dimensions, relationships to physical parameters and proposed handling of free variables all enabled a framework of design parameters to be formulated and optimised through iteration, including analyses and computer simulations.
The challenge was to find the best parameterisation, such that a sufficiently large design space can be defined by as few parameters as possible. Only after the initial framework was established, providing a continuous distribution of main dimensions as function of Unit Speed and Unit Flow within the applicable design range, did the detailed design and optimisation of individual models start.
For example, says Boerresen, starting from Euler’s turbine equation (a particular adaption of law for conservation of angular momentum) one can derive a relation between the turbine blade angle and the inlet diameter of the turbine. Using this combined with loss models and a large number of computer simulations one can define a relation for the inlet diameter of the turbine as function of the specific speed.
The main advantage of the parametric approach is to ensure a consistent framework for comparing individual models and, therefore, to permit interpolation between models to derive intermediate instances, or new models. Computing power is vital to the approach and helps to deliver continuously improving precision, which in turn reduces time needed in the laboratory. The lab is only needed for final verification and documentation of particular flow phenomena not yet covered in computational models. Each defined model in the data field is added with the work on each client contract.
While the approach may not be original or unique, it is not only rarely used today for hydro-machinery but is being done so on an extensive, comprehensive basis to establish a consistent, business-wide proven design platform for Rainpower. The utility of computing power, though, may well see some others undertake optimisation work using a parametric approach, Boerresen anticipates.
Parametric high head
High head research using the parametric approach has already been underway in Rainpower – for new Pelton turbines. Boerresen says that R&D has been underway for almost two years, having started soon after the commencement of the Francis turbine research and is approaching completion. While the brand name has yet to be revealed it is intended to launch the first product this year.
In principle, the approach to the Pelton research was quite similar to the Francis R&D, he says, and continues, ‘but at the same time both the product and the design team is different so, in the end, the process was quite different.’ As an example, he cites the complexity of the Pelton bucket as requiring much more work on the detailed parameterisation of the surface in comparison to the Francis blade.
He adds that starting from a clean sheet has been motivating for the engineers as it allows them free rein to discover new, interesting knowledge, but that this is more of a spin-off benefit for the goal is to develop independent, competitive technology.
Looking ahead to the next high head research – for Francis turbines – Boerresen notes that while the design philosophy will be similar to ‘Rainpower Storm’, dominant loss mechanisms are different, and so the parametric approach will not necessarily provide for continuity in the models. ‘Thus, in an intermediate range there may be a range of specific speeds where we can employ two different basic designs.’
For the moment, there is no strategic aim to research other types of turbines, such as very low head Kaplan or Bulb units, with either a parametric or other, more conventional, approaches. The low head research continues for the ‘Storm’ range just above Kaplan for N11 values of more than 70, says Boerresen.
Rainpower has chosen where it wants to focus its new portfolio – Francis, for the full head range, and Pelton.
|Route to Rainpower|
Rainpower was established in November 2007 with Norwegian parent group NLI combining the experience and skills of a few business areas - its own small hydro and hydro-related services activities; the SÃ¸rumsand workshop; and, the local subsidiary of GE Hydro, which was mainly responsible for medium and high head Francis, Pelton and reversible pump turbines.