Study of flow characteristics in the S-shaped region of a pump-turbine at Baoquan pumped storage power station

The rapid development of local economies makes the load on the power grid steadily larger, and brings severe challenges to its regulation. Compared to other types of power station, pumped storage stations have big advantages and offer a variety of benefits – e.g., balancing load peaks and troughs, modulating phase and/or frequency, providing emergency backup, and so on. Nowadays, pumped storage is the best and cheapest bulk method of storing power in the world. Its conversion efficiency can be as high as 75%, and the capacity ratio of pumped storage is important in judging national energy usage efficiency and increasing energy use (Zhao et al. 2012). The power grid is likely to be most effective only when the capacity ratio of pumped storage is 5–10% of the power system (Mei 2000).

Baoquan power station is a regulation and pumped storage power station. It is equipped with four single-stage and reversible pump-turbine units manufactured by ALSTOM, with unit capacity 300 MW. It is strategically significant, easing electricity supply, enabling optimization of the structure of China's central power grid, promoting the national network, and ensuring safe and stable working of the grid. Despite this, many problems arose during operation of 1# unit. For example, it did not operate stably when its speed was rated prior to grid connection, the guide vane aperture changed significantly, vibration increased, and the frequency of speed variation was too great, so study of the ‘S’ characteristics became urgent. In this study, the unit's internal flow characteristics in the S-shaped region were revealed using CFD numerical simulation and the results are expected to provide theoretical support for stable operation.

To date, references about pump-turbine ‘S’ characteristics have been limited. Chen & Xie (2001) determined the real flow in the ‘S’ characteristics area and discussed the reasons for unstable flow. Using guide vanes that were misaligned, You et al. (2006) suggested that some difficulties could be solved, e.g., grid connection under low-head-generating conditions, and zero-load instability after load shedding in turbine mode. Cheng (2008) clarified damage to the S-shaped region and proposed using guide vanes controlled asynchronously to avoid the region. Nicol & Alligne (2008) studied pump-turbine pressure fluctuation under runaway conditions. Martin (2010) established a mathematical expression for the complete characteristics curve of a pump-turbine, and analyzed the ‘S’ characteristics theoretically. Using visualization research and the pressure fluctuation in a pump-turbine under working conditions for which it was not designed, Hasmatuchi & Roth (2010) analyzed the flow status in S-shaped region. Wang & Liu (2010) developed an understanding of the rules relating vortex structure and guide vane opening, and how they affect ‘S’ characteristics. Zhang et al. (2011) studied the internal flow field distribution under different pump-turbine working conditions using CFD numerical simulation. Ji & Lai (2011) worked on numerical simulation of the S-shaped region in low specific speed pump-turbines. Using numerical calculation, Xiao et al. (2012) analyzed the ‘S’ characteristics and pressure fluctuation of pump-turbines under non-synchronous guide vane conditions. Liu et al. (2013) analyzed the pump-turbine's ‘S’ characteristics in both synchronous and non-synchronous guide vane systems, using complete three-dimensional numerical simulation.

The subject of investigation was the flow channel, and the calculations took account of the spiral case, guide apparatus, runner and draft tube. A schematic diagram of the turbine system (flow channel) is shown in Figure 1 and the design parameters are given in Table 1.

The turbulent model used was SSTk-ε. Considering the influence of flow on vortex viscosity, the model had high accuracy and was applied to various pressure gradient turbulent calculations. The basic import discharge and export pressure were the boundary conditions, and the reference pressure was one standard atmosphere. The near wall surface used the standard wall function, and solid surface used no slip boundary conditions. On account of its complexity, the calculation region was divided using an unstructured, tetrahedral mesh that is highly adaptable. The numbers used in the mesh unit are shown in Table 2. The operating head range was between 494.0 and 570.4 m, the corresponding unit-speed (n₁₁) range was from 43.192 to 40.196 rpm, and the guide vane opening range was 7.183 to 3.592 °.

In the model tests, 29 operating points were selected on the ‘S’ curve. The guide vane openings of the points were 7, 20 and 23 °, and 3-D turbulent simulations of every point were carried out. The results of the tests and calculations are shown in Table 3. It was found that all relative errors were less than 6%, indicating that the calculation method used in this study was reasonable and feasible, and that the essential features of the unit's ‘S’ characteristics could be reflected using CFD simulation.

When operated as a hydro-turbine, the unit's ‘S’ curve was distributed in the first and fourth quadrants, and went through five modes – turbine, runaway, braking, zero and reverse pump, respectively. The nodes of the working area partition were the runaway and zero flow points. As a result, the flow regime changes in the S-shape region could be understood by analyzing four points, relating to the turbine, runaway, zero and reserve pump modes, respectively, on the ‘S’ curve of guide vane opening. The streamline distributions in the flow channel for the four modes, with guide vane openings of 7 and 20 °, are shown in Figures 2,³–4.

The results were

1. In turbine mode, the flow in the spiral case, guide apparatus and runner were smooth, and neither secondary nor transverse flow seemed to exist. A water-ring between the guide vane and runner appeared. Slight swirling flow appeared in the taper pipe section of the draft tube, but the direction of most water flow was still downward, along the pipe wall, and the back-flow appeared in the center of the taper pipe. Partial back-flow and deflected flow occurred in the outlet of the draft tube.

2. In runaway mode, the flow in the spiral case and guide apparatus remained smooth when the guide vane opening was 7 °. When it was 20 °, however, a vortex appeared in the nose of the spiral case and guide apparatus. This showed that the influence of flow deterioration spread upstream gradually as the guide vane opening increased. In the two cases, secondary back-flow and transverse flow existed, and the flow exhibited asymmetry very clearly. Violent swirling flow and central back-flow appeared in the taper pipe of the draft tube, but most flow along the pipe wall moved downstream in a spiral. Back-flow was seen very clearly in the draft tube diffuser and the internal flow in the unit worsened.

3. In zero flow mode, flow in the spiral case and guide apparatus was very turbulent, a serious vortex developed, and secondary and transverse flows were very obvious. Violent swirling flow and central back-flow appeared in the draft tube, and the flow moved downstream in a spiral and then reversed upstream, while there was almost no streamline flow out of the draft tube outlet.

4. In reverse pump mode, the flow moved from the draft tube to the spiral case, that in the draft tube being smooth. In the runner, transverse flow appeared very clearly because the rotation direction was still that of turbine mode. In the guide apparatus and spiral case, a very obvious vortex appeared in the flow channel between the guide and fixed guide vanes. Some head loss occurred because the outflow line was not parallel to the spiral case wall.

In line with references in the literature, the ‘S’ characteristics could be seen easily under low-head, turbine mode conditions, and were improved and eliminated with rising heads, in actual operations. The relationship between them was not clear, however, and in order to explore this, runaway points under 10 different guide vane openings were selected, and numerical flow simulations were carried out with respect to the spiral case, guide apparatus and runner. The streamline distributions under different guide vane opening conditions are shown in Figures 5 and 6. The velocity vector distribution on the central plane of the runner is shown in Figure 7, and the pressure distribution on the central plane of the guide vane and runner is shown in Figure 8.

As can be seen in these figures, the results were

1. As guide vane opening increased, the gap between the guide vane and runner decreased steadily, while the dynamic and static interference between them strengthened continuously. The outcomes were that both the water-ring thickness and the guide vane pressure gradient, upstream to downstream, decreased constantly. Because of this, the runner's effect on the flow spread upstream. The symmetrical distribution characteristics of internal flow and pressure worsened constantly, and a vortex began to appear in the flow channel of the guide apparatus when the guide vane opening was 13 °. As the opening increased further, a vortex began to appear in the spiral case, proving that the effect of the water-ring on the runner was actually that of containment.

2. Serious breakaway-flow existed in the runner inlet. Part of this developed gradually into a larger area of circulatory flow and then plugged the channel. Water entering the channel was forced into another one. As a result, the discharge was redistributed, the speed and torque undulated, and operation was unstable. Changes in the flow regime in the guide vane channel need further study. When the vortex first appeared in the partial flow channel of the guide vane, discharge from the flow channel was redistributed, and the hydraulic torque and guide vane opening fluctuated, while the speed fluctuation got much worse.

The complete, flow-channel numerical simulation of 1# pump-turbine, led to several conclusions:

1. The results from the CFD numerical simulation and the model tests were basically the same. In other words, the method of calculation was reasonable and feasible, reflecting the flow regime of the pump-turbine in the S-shaped region, so that the ‘S’ characteristics could be analyzed qualitatively.

2. The effect of the water-ring, between guide vane and runner, on runner's inertia force was one of containment. The serious vortex in the flow channel caused congestion, and it was fundamental to the pump-turbine's ‘S’ characteristics.

3. To inhibit or avoid ‘S’ characteristics damage on the unit, the first thing is to keep away from S-shaped region, and the second to eliminate the S-shaped region of the pump-turbine's characteristics curve. For example, opening part of the guide vane in advance (MGV apparatus) to increase the discharge rate was successful at Tianhuangping pumped storage power station. Not only did the difficulty and workload involved in controlling the equipment increase, but the faults in the MGV equipment reduced the unit's start-up success rate, while flow in the channel became more complicated. In both the hydraulic design and model tests, changing the flow regime in the channel was the fundamental measure used to solve the ‘S’ characteristic problems. Therefore, fundamental research and cooperation were needed between the scientists and engineers.

This work was financially supported by the National Natural Science Foundation of China (51179152 & 90410019), the Public Industry Scientific Special Fund of the Ministry of Water Conservation of PRC (201201085), and Foundation of the Education Department of Henan Province (12A570003).