How standards constrain an innovative pole design which can reduce CO2-emissions in large hydro power generators

Thomas Hildinger Voith Hydro Holding GmbH & Co. KG Alexanderstr. 11 89522 Heidenheim Germany Achim Frank Voith Hydro Holding GmbH & Co. KG Alexanderstr. 11 89522 Heidenheim Germany Klaus Reppe Retired Burscheider Weg 18g 13599 Berlin Germany

Introduction
The efficiency of large hydro power generators under load can be improved by changing the traditional symmetrical pole design to a novel unsymmetrical design. A higher efficiency leads to more production of electrical energy and consequently to higher profit. Better generator efficiency benefits the owner and the environment through the reduction of tons of carbon dioxide emissions due to the higher efficiency in the energy convertion process. Even hydropower plants may produce CO2 and have also to contribute together with other renewable energy sources for the highest possible sustainability. An increase of the generators efficiency of one-hundredth of one percent will save thousands of tons of CO2 during its lifetime. There is a simple way to improve the efficiency of a typical large hydro power generator which can improve customers benefits as well as our environment. The idea is to use unsymmetrical poles instead of symmetrical poles. According to simulation results, an unsymmetrical design leads to a more homogeneous field and lower saturation effects under load conditions which help to reduce additional losses in the stator winding. These specific additional losses are called longitudinal field losses because they are caused by a longitudinal field inside the stator slots because of saturation effects in the stator teeth. Unfortunately the advantages of the proposed pole design are only measurable under load. Since, according to the most common standards, the ordinary tests to determine the generator losses and efficiency are performed during no-load and short circuit tests and not under load the standard measurements do not reflect the electromagnetic reality. It might sound absurd, but large hydro generators are designed and optimized for no-load and short circuit situations while the real machines most likely will run under load and very seldon at no-load or short circuit. The concept of unsymmetrical poles for electric machinery is not totally new, but it was never done for large synchronous hydro-generators, because of inconsistency between usual standards and physical reality. Although standards like IEC and IEEE are essential for the hydro business, they do not always serve non-standard innovations like the proposed unsymmetrical pole shoe design.

1. Simulation results
To make the discrepancy between standards and physical reality clear, simulations have been performed for rated load and no-load operations. The results show that the appearance of the magnetic field for symmetrical and unsymmetrical poles for load and no-load is interchanged. 1.1 Load conditions The idea to build unsymmetrical poles is based on 2D-simulation results which are shown in figure 1 and 2 for load conditions. Figure 1 shows the magnetic field for the classic design. The density of the field lines on the surface of the symmetrical pole shoe is higher on one side of the shoe than on the other. The high density is on the subsiding part of the pole, where the torque is mainly generated. The teeth and the pole shoe are strongly saturated on that side of the pole. The field in the saturated teeth is displaced inside the slot and causes eddy current losses (so called longitudinal field losses), which can reach up to 20-30% of the I2R-losses

inside the stator core length. These high longitudinal field losses can be expected because the grade of saturation of a stator tooth changes within the time a pole needs to move over the distance of one slot pitch, so the flux change d/dt is very high. These losses additionally heat up the stator winding, which leads to higher ohmic losses. These strong saturations, which cause the longitudinal field losses, only appear under load; they cannot be measured in the no-load and short circuit runs. One can easily note that they are invisible for standard efficiency measurements.

Figure 1 Magnetic field of a machine with symmetrical poles at load

Figure 2 Magnetic field of a machine with unsymmetrical poles at load

One solution to avoid the unsymmetrical field is unsymmetrical poles. If the pole shoe is inclined, the maximum air gap on one side will be larger than the air gap on the other side, which will lead to an unsymmetrical air gap reluctance. Since the flux density under load is displaced to the subsiding end of the pole shoe, the combination of both unsymmetrical effects will lead to a more homogeneous field. But inclination is not the only way to change the shape of the air gap. It is also possible to add more iron to one side of the pole shoe so that the air gap gets smaller than on the other end of the shoe. Figure 2 shows an unsymmetrical pole shoe design which could be used instead of the pole shown in figure 1 for the same machine. The pole shoe was inclined in this example. As a result of that design the magnetic field is quite smooth on both sides of the pole shoe. But it is not the smooth field that makes the unsymmetrical design desirable. The big advantage of the unsymmetrical design is the reduction of the losses at load operation. The amount of saved energy, depends on the machine design and varies depending on parameters like stator winding slot width, strand width or maximum air gap induction. In general it was observed that the longitudinal field losses can be reduced by 30-50%. A brief numerical example will be given below for a characteristic large hydro power generator. 1.2 No-load conditions Simulations at no-load are totally inverse to the results shown and described above. The field is smooth for the symmetrical poles and less homogeneous for the unsymmetrical alternative. As a consequence the losses at no-load will be higher for a generator with unsymmetrical poles, because Lion-Heart losses will appear for the same reasons as already described above.

Figure 3 Magnetic field of a machine with symmetrical poles at no-load

Figure 4 Magnetic field of unsymmetrical poles at no-load

machine

with

2. Measurement of physical reality versus standards

All major standards for the determination of the losses suggest calorimetric measurements during three different test runs; a mechanical run to determine the windage losses, a short circuit test to determine the load-dependent losses and a no-load test run to measure the load-independent losses. The results of these three different test runs are used to calculate the losses under load condition for a certain guaranteed temperature. According to the standards the only difference between the test runs and load conditions are the temperatures inside the machine. The influence of the temperature on the windage and load-independent losses is very small and not of interest for the discussed topic, so it shall be neglected. Focusing on the load-dependent losses VSC it is common to split up these losses into ohmic losses Vohm and stray load losses Vadd, some times also called additional losses. The ohmic losses can easily be calculated if the temperature during the short circuit test SC is known:

Vohm = RSC I 2 = R20C

235 + 2 I 235 + 20

The additional losses according to IEEE are then:

Vadd = VSC Vohm

According to IEEE these additional losses which are separated at SC are not dependent on the temperature. The IEC standard uses the guaranteed load temperature instead of the measured short circuit temperature to split up ohmic and additional losses. The sum of ohmic and additional losses is supposed to be constant according to IEC. IEEE says that additional losses are constant and do not depend on the temperature. To avoid discussions about what standard is closer to physical reality, efficiency measurements must be made under load conditions. Unfortunately the standards are not consistent in terms of temperature-dependent behavior of the additional losses, it is however, even more important to point out that all standards ignore the influence of saturation effects under load operation. The procedures according to IEEE and IEC are quite good to verify contracted values, but they constrain the launch of unsymmetrical poles. In the no-load and short circuit tests, unsymmetrical poles have a less effective performance. The field under these conditions looks like the field of symmetrical poles under load. As a consequence the losses measured for unsymmetrical poles under no-load and short circuit will be higher than for symmetrical poles. In other words: Symmetrical poles are optimized for no-load and short circuit test runs and unsymmetrical poles are designed for load operation. But since the standard efficiency measurements according to IEEE or IEC are performed during no-load and short circuit, symmetrical poles seem to have a better performance. And this leads close to the absurd situation that large hydro generators are designed and optimized for no-load and short circuit while the real machines most likely will run under load and very seldom at no-load or short circuit.

To prove the simulation results and to measure the true efficiency of a generator a calorimetric measurement under full load must be performed to capture all losses that appear under load. If the customer only wants to know the efficiency and not the separated losses he can also save time, since only one instead of three test runs have to be performed during commissioning.

3. Examples
To summarize the difference of the symmetrical and the unsymmetrical pole shoe design and why standards constrain the new design, two examples are given in Table 1 and 2. The first simulated machine has an apparent power of 273MVA, 52 poles and 624 slots. The second machine represents the upper limit of hydro power generators with a power of 855.6MVA, 48 poles and 576 slots. Since ohmic losses of the stator and field winding are quite the same for the symmetrical and unsymmetrical design, they are not mentioned in table 1 and 2. Machine 1 (273 MVA) Calculated losses Symmetrical poles Unsymmetrical poles Electrical osses at no-load 630.3 kW 648.6 kW Additional losses at short circuit 366.8 kW 397.8 kW Longitudinal field losses under load 376.9 kW 245.8 kW Total calculated losses for load acc. IEC 2885.3 kW 2903.6 kW at 75C Total calculated losses for load under 3461 kW 3270.8 kW real load conditions and 75C Table 1 Example 1 for a machine designed with symmetrical poles and unsymmetrical poles Table 1 points out a significant difference between the simulated losses according to IEC and the losses simulated under load condition. Since the longitudinal field losses are not included in the calculation according to IEC the results are much lower than the results for the simulation under real load conditions. In fact there is one more field dependent loss component, but since the influence of unsymmetrical poles on this effect is quite small it is not discussed here. According to the IEC interpretation of losses, the losses of the symmetrical poles would be 18.3kW lower. Even according to IEEE the diference will be in the same magnitude. But with the calculation results presumed to be correct, 190.2kW could be saved under nominal load if unsymmetrical pole shoes are used for this example machine. With a CO2-rate of 0.6 kg/kWh and 8000 operating hours a year this generator could save 1522MWh of energy and 913t of CO2 annually. Example 2 is even more distinctive than the first example. Machines of that size have very high inductions which push the electrical designers to the limit of what is technically feasible. The slots are also quite big. These are factors that lead to high longitudinal field losses. Machine 2 (855.6 MVA) Calculated losses Symmetrical poles Unsymmetrical poles Electrical losses at no-load 1930.5 kW 2006.2 kW Additional losses at short circuit 1093.1 kW 1093.8 kW Longitudinal field losses under load 2099.5 kW 1146.3 kW Total simulated losses for load acc. IEC 8495.2 kW 8568.6 kW at 95C Total simulated losses for load under 11328 kW 10220 kW real load conditions and 95C Table 2 Example 2 for a machine designed with symmetrical poles and unsymmetrical poles The longitudinal field losses reach values which are comparable with the rated power of small hydro power plants or wind power generators. In that particular case losses under load can be reduced by 1.1MW. So one unit could save 5280t of CO2 per year (based on the same numbers as were used for example 1). As in the mentioned power station 18 units will be installed the benefits (savings in losses and CO2) can be considerable.

4. Positive side effect

Another advantage of unsymmetrical poles is higher stability at rated load (referred to the same reactances). For symmetrical poles, the maximum amplitude of the field curve is located in the pole center at no-load, but moves outside the center under rated load. The proposed unsymmetrical shape shifts the maximum back to the pole center, where the magnetic main axis with the highest stability is situated.

Figure 5 Field curve of symmetrical poles

Figure 6 Field curve of unsymmetrical poles

5. Conclusion
The new proposed design for pole shoes of large synchronous hydro generators is to increase the air gap between pole and stator core from one side of the pole to the other. This leads to a more homogeneous flux density as well as less saturation and, as a consequence, lower losses in the stator winding. Since the proposed design will only have advantages under rated load, measurements must be done at rated load and not, as described before, at no-load and short circuit, in accordance with the existing standards. According to calculated results unsymmetrical poles can help to improve the efficiency of large hydro generators and consequently help to reduce CO2-emissions in the hydro business. References
Hildinger, T. , Frank, A., Reppe, K., Patent Application PT 07104 - Abgeschrgte Generatorpole , 2009. Mller, G., Vogt, K., Ponick, B., Berechnung elektrischer Maschinen, Wiley-VHC, 6. Auflage, 2008.

The Authors
Thomas Hildinger graduated in electrical engineering at the University of Sao Paulo and also graduated in business administration at the University of Sao Paulo. He started to work for Siemens in 1984 and joined Voith Siemens Hydro in 2000 when the Joint Venture was established. He is currently Head of the Generator Engineering and Development Group at Voith Hydro in Heidenheim, Germany. He is also the current Regular Member of SC1 at CIGRE for Germany. Achim Frank graduated in electrical engineering and automation at the University of Ulm and started working for Voith Hydro after finishing his diploma thesis at the Zentrum fr Sonnenenergie- und Wasserstoff-Forschung (ZSW) in 2007 in Ulm, Germany. He is currently working in product development and software development for electromechanical tools at Voith Hydro in Heidenheim, Germany. Klaus Reppe graduated in electrical engineering at the Fachschule fr Elektrotechnik in Frankfurt a.M.. He was a member of the electrical design group at Siemens Dynamowerk in Berlin for 40 years. He was and still is focused on the calculation of magnetic fields inside of large hydro generators. Since his retirement he is still working in the development of electrical calculation tools.