WIND TURBINES IN WEAK GRIDS CONSTRAINTS AND SOLUTIONS J O G Tande and K Uhlen SINTEF Energy Research, Norway SUMMARY Cost efficient utilisation of the wind energy resource requires wind farms to be located at areas with good wind resources. Such areas are often in proximity to a relatively weak grid that may represent technical constraints for exploiting the wind resources. Hence, aiming for cost efficient utilisation of wind energy resources it is a challenge to find ways to overcome the technical constraints, i.e. basically issues related to thermal capacity, voltage quality and stability. 58 57 56 55 54 1 53 Figure 1: Example network with small wind farm. We consider a small wind farm on a medium voltage feeder as shown in Figure 1 to illustrate issues of voltage quality. We demonstrate that the development of IEC 61400-21 [1] provides firm basis for assessment of the impact of wind turbines on voltage quality. This make previously used simplified rules unnecessary, and opens for increased utilisation of wind energy in weak grids. Further, we demonstrate that possible voltage quality problems due to a small wind farm may be overcome simply by selecting an appropriate wind turbine type, and/or by adjusting wind turbine control parameters. The assessment includes determination of voltage profiles (Figure 2), emission of flicker, voltage dips and harmonics. Voltage (pu) 1.06 1.04 1.02 1.00 0.98 0.96 0.94 min load/max wind max load/0 wind 0.92 0 10 20 30 40 50 60 Node number Figure 2: Result of example load-flow analysis. We consider the case of a large wind farm in a regional 132 kv grid distant from the main 300 kv transmission grid to illustrate issues related to thermal capacity and voltage stability. First we assume the regional grid to be connected to the main transmission via two separate feeders. In this case, depending on the amount of injected wind power, an outage of one of the feeders may cause a voltage collapse as shown in Figure 3. Hence, this could be a constraint for the wind power development in the regional grid. 138 136 134 132 130 128 Bus 1 126 124 Wind farm 122 120 0 20 40 60 80 100 120 140 160 Wind power injected (MW) Figure 3: Voltage dependency of injected wind power. Voltage (kv) This problem can be solved however by installation of a inexpensive system protection scheme that send a signal to automatically reduce the wind power generation when a line outage is detected. Sometimes regional networks are connected to the main transmission via one feeder only. If this was the case, we show that application of an SVC unit or a similar device for continuous voltage control allows increased wind power penetration. We have not made any detailed assessment of the costsavings associated with the suggested solutions compared to the alternative of conventional grid reinforcement by installation of new lines. This is because such numbers will be very case specific. We suspect however that for a majority of projects, the solutions considered will be more economic than installation of new lines. [1] IEC 61400-21 (2001) Measurements and assessment of power quality characteristics of grid connected wind turbines. FDIS.
WIND TURBINES IN WEAK GRIDS CONSTRAINTS AND SOLUTIONS J O G Tande and K Uhlen SINTEF Energy Research, Norway ABSTRACT The development of IEC 61400-21 provides firm basis for assessment of the impact of wind turbines on voltage quality. This makes previously used simplified rules unnecessary, and opens for increased utilisation of wind energy in weak grids. Assessment of thermal capacity and voltage stability constraints call for additional analyses and innovative solutions. These are pinpointed by examples enabling cost-effective means to allow increased penetration of wind power in weak grids. SYMBOLS ψ k N T U LV U T c(ψ k,v a ) cos(ϕ) d E Plt E Pst k f (ψ k ) k u (ψ k ) N 10 N 120 P lt P mc P n P st S k U LV U MV U n v a network impedance phase angle LV transformer deviation from nominal ratio voltage drop from LV transformer to consumer voltage drop due to transformer losses flicker coefficient for wind turbine power factor of wind turbine sudden voltage reduction long-term flicker emission limit short-term flicker emission limit flicker step factor for wind turbine voltage change factor for wind turbine number of wind turbine starts within 10 min. number of wind turbine starts within 2 hours number of wind turbines long-term flicker emission maximum permitted power from wind turbine rated power of wind turbine short-term flicker emission network short-circuit power voltage at low voltage consumer voltage at medium voltage node nominal voltage annual average wind speed at wind turbine hub 1 INTRODUCTION The total operating wind power capacity in the world has increased from approximately 2000 MW in 1990 to well over 16 000 MW by end of 2000. The development has created a booming wind power industry. Continuous rapid growth is expected indicating 50 000 MW of operating wind power capacity by year 2010. Driving forces for this development are environmental concerns, Kyoto targets and the improved cost efficiency of new wind farms. Cost efficient utilisation of the wind energy resource requires wind farms to be located at areas with good wind resources. Such areas are often in proximity to a relatively weak grid that may represent technical constraints for exploiting the wind resources. Hence, aiming for cost efficient utilisation of wind energy resources it is a challenge to find ways to overcome the technical constraints. Technical constraints in relation to wind power in weak grids may be associated with limited thermal capacity in parts of the grid and the effect that wind power has on voltage quality and stability. Indeed, if the weak grid is a small island grid, frequency control could also be a constraint. In this paper however, the weak grid is assumed to be part of a large interconnected power system where wind power will not have a significant effect on the frequency control. Hence, the focus of this paper is on issues related to thermal capacity, voltage quality and stability. The main points we are making are: - The development of IEC 61400-21 [1] provides basis for detailed assessment of the impact of wind turbines on voltage quality. This make previously used rule of thumbs unnecessary, and opens for increased utilisation of wind energy in weak grids. - Possible voltage quality problems due to a small wind farm may be overcome simply by selecting an appropriate wind turbine type, and/or by adjusting wind turbine control parameters. - Thermal capacity and voltage stability constraints related to a large wind farm call for innovative solutions involving e.g. dedicated system protection and dynamic compensation of reactive power. - The solutions considered may be more economic than installation of new lines. Voltage quality issues are addressed in section 2, whereas issues related to thermal capacity and voltage stability are addressed in section 3. In both sections we include examples illustrating both possible constraints and solutions. Conclusions are drawn in section 4. 2 VOLTAGE QUALITY Throughout this section we will refer to the EN 50160 [2] to characterise the voltage quality that any customer can expect under normal operating conditions. We consider slow voltage variations, i.e. voltage variations measured as 10-minute-mean RMS values, flicker, voltage dips and harmonic voltage, and describe how
wind power may influence these characteristics. Assessment is made according to [1] were applicable. For illustration, we use the example network and wind turbine specifications given in Figure 1. The wind turbines are of a conventional design, operating at fixed speed and using stall control for power limitation at high wind speeds. Each wind turbine is equipped with power electronics that limits the in-rush current to the induction generator during start, and capacitors that are switched to maintain cos(ϕ) = 1,0 during operation. Wind turbine: P n = 750 kw cos(ϕ) = 1,0 U n = 0,69 kv v a = 8,2 m/s c(55,8,2)=10,9 N 10 = 1 N 120 = 12 22 kv +/-1,5 % P mc = 1,2 P n k f (55)=1,2 k u (55)=1,5 58 57 56 55 54 1 53 Sum max load: 7,4 MW; 1,5 Mvar Sum min load: 2,2 MW; 0,5 Mvar Wind farm PCC: U n = 22 kv S k = 37,6 MVA ψ k = 55 deg E Pst = 0,7 E Plt = 0,5 Figure 1: Example network with small wind farm. 2.1 Slow Voltage Variations Load-flow analyses may be conducted to assess the slow voltage variations. Figure 2 shows results for two load situations that for the example specifications give the maximum and minimum voltage levels on the medium voltage (MV) line. Node 1 denotes the MV node at the high voltage (HV) transformer that for simplicity in this example is set to be constant = 1,0 pu. The five 750 kw wind turbines constituting a wind farm are connected at nodes 54-58, whereas all other nodes connect consumers via low voltage (LV) transformers and LV lines. Voltage (pu) 1.06 1.04 1.02 1.00 0.98 0.96 0.94 min load/max wind max load/0 wind 0.92 0 10 20 30 40 50 60 Node number Figure 2: Result of example load-flow analysis. The voltage at a LV consumer is given by: U LV = U N U U (1) MV T T LV For this example system at maximum load, U T = 2 % and U LV = 5 %, whereas the voltage drop is negligible at minimum load. This indicates that the tap changer for the transformer at node 53 must be set to -5 % for achieving a minimum voltage at LV > 0,90 pu. Assuming the tap-changer to be fixed to -5 %, the maximum voltage at LV would then be 1,09 pu. If the wind farm were expanded with more wind turbines of the same type, the maximum voltage would further increase. According to [2] the slow voltage variations shall be within ± 10 % of U n during 95 % of a week. In addition, for low voltage only, the slow voltage variations shall always be within -15/+10 % of U n. Hence, for the example system, slow voltage variations may be a constraint for further expansion of the wind farm. This constraint may however easily be overcome e.g. by adjusting the cos(ϕ) of the wind turbines. A modest reduction of the cos(ϕ) from unity to 0,98 (inductive) reduces the maximum voltage by 1,5 %, and makes room for more wind power. Actually, the wind farm may expand to a total of eight 750 kw wind turbines operated at cos(ϕ) = 0,98 before slow voltage variations again become a constraint for further expansion. We have made a simplification disregarding the deadband of the voltage regulation at the HV-transformer. If we had taken this into account, the obtained minimum and maximum voltages would be slightly different. Further, possible uncertainties in estimates of minimum and maximum load levels could impose safety margins. None of this does however change the general analysis result, suggesting that a possible slow voltage variation constraint may be counteracted on by adjusting the cos(ϕ) of the wind turbines. One argument against is that a reduced cos(ϕ) causes increased network losses. This implies that cos(ϕ) regulation should be used with care, and that alternative options should be assessed. Examples of alternative options are grid reinforcement by installation of new lines and voltage dependent reduction of wind power production, see e.g. Tande [3]. 2.2 Flicker According to [2] the long-term flicker severity shall be 1 during 95 % of a week. To ensure this, each source of flicker connected to the network can only be allowed a limited contribution, e.g. as for the example network E Pst = 0,7 and E Plt = 0,5 at the point of common coupling (PCC) of the wind farm. At other networks, different values may be found using IEC 61000-3-7 [4] as a guide. Following the recommendations given in [1], the flicker emission from a single wind turbine or wind farm may be assessed. Procedures are given both for assessing flicker emission due to starts and due to continuous operation. The procedure for assessing flicker emission due to starts
assumes that each wind turbine is characterised by a flicker step factor, k f (ψ k ), being a normalised measure of the flicker emission due to a single worst-case start. Further, the procedure assumes that for each wind turbine information is also given on the maximum number of starts, N 10 and N 120, that can be expected within a 10 minute and 2 hour period respectively. Based on these characteristics, the maximum expected flicker emission due to starts from a single wind turbine or wind farm can be calculated: 18 Pst = N S k i= 1 8 Plt = N S k i= 1 3, ( k ( ψ S ) 10,i f,i k ) n,i 3, ( k ( ψ S ) 120,i f,i k ) n,i 0, 31 2 0, 31 2 (2) (3) For deduction of the above equations, reference is given to [1]. For the example specifications we get P st = 0,71 and P lt = 0,68, and exceeding the assumed example limits. Hence, the flicker emission due to starts may be a constraint for operation of the example wind farm. This constraint may however be overcome quite easily by using another type of wind turbine with a smaller k f, basically a pitch regulated or a (semi-)variable speed type. Another alternative is to ensure that only a reduced number of wind turbines are allowed to start within the same 10- minute and 2-hour period. The latter involves altering the control system settings of the wind turbine to a smaller value for N 120, and introducing a wind farm control system that allows only a reduced number of wind turbines to start within the same 10-minute period, in effect altering in (2). The procedure for assessing flicker emission due to continuous operation assumes that each wind turbine is characterised by a flicker coefficient, c(ψ k, v a ) being a normalised measure of the maximum expected flicker emission during continuous operation of the wind turbine. To find the flicker emission from a single wind turbine, the flicker coefficient with the relevant ψ k and v a is simply multiplied by S n /S k, whereas the emission from a wind farm can be found by: P st 1 = Plt = S k i= 1 ( c ( ψ,v ) S ) i k a 0, 5 2 n,i P st = P lt in (4) because it is probable that conditions during the short-term period persist over the long-term period. Further (4) assumes that the maximum power levels between wind turbines are uncorrelated. At special condition however, wind turbines in a wind farm may synchronise causing power fluctuations to coincide. (4) would then underestimate the flicker emission. This is assessed in Tande et al [5], concluding that for common (4) conditions however, (4) will provide a good estimate. For the example specifications we get P st = P lt = 0,49, which is just within the assumed example limit E Plt = 0,5. If the wind farm were expanded with more wind turbines of the same type, the flicker emission would further increase above the acceptable limit. Hence, for the example system, flicker emission due to continuous operation may be a constraint for further expansion of the wind farm. To overcome this constraint, the straightforward approach is to select a different wind turbine type that has a smaller c value. First of all, c = 10,9 is rather high for a fixed speed, stall regulated wind turbine, so this value should not be taken as typical for this types of wind turbines. Certainly, a (semi-)variable speed wind turbine would yield a much lower c value, whereas a fixed speed, pitch regulated type could actually yield a higher value of c. 2.3 Voltage Dips According to [2], a voltage dip is a sudden reduction of the voltage to a value 0,01 and 0,90 pu followed by a voltage recovery after a short period of time, conventionally 1 ms to 1 minute. The expected number of voltage dips during a year may vary from a few tens to one thousand. Start-up of a wind turbine may cause a sudden reduction of the voltage followed by a voltage recovery after a few seconds. Assuming that each wind turbine is characterised by a voltage change factor k u (ψ k ), the sudden voltage reduction may be assessed according to [1]. S n 100 ku ( ψ k (5) S k d = ) As more wind turbines in a wind farm are unlikely to start-up at the exact same time, (5) is not a function of the number of wind turbines as (2), (3) and (4). For the example wind turbines, we find that d = 3,0 %. This sudden voltage reduction is in most cases acceptable, especially considering that this would imply a voltage less than 0,90 pu, i.e. a voltage dip, only for starts coinciding with high load at the network. Further, as (5) is not a function of the number of wind turbines in the farm, we can conclude that for the example system, voltage dips are not a constraint for further expansion of the wind farm. 2.4 Harmonic Voltage [2] states limits for individual harmonic voltages up to the order of 25, and that the total harmonic distortion of the voltage, calculated up to the order of 40, shall be 8 %. To ensure this, only a limited emission of harmonic currents can be allowed. The limits depend on network specifications, and can be found using IEC 61000-3-6 [6] as a guide.
A wind turbine with an induction generator directly connected to the grid without an intervening power electronic converter is not expected to distort the voltage waveform. Power electronics applied for soft-start may give a short-duration burst of higher order harmonic currents, though the duration and magnitude of these are in general expected to be so small that they can be accepted without any further assessment. So, for the example system with fixed speed wind turbines, emission limits for harmonics is not a constraint. If however we had considered variable speed wind turbines using power electronic converters, these should be assessed. Thyristor-based converters are expected to emit harmonic currents that may influence the harmonic voltages for which limits are given in [2]. Such converters are however rarely used in new wind turbines. Rather, the converters are transistor-based, and operated at switching frequencies above 3 khz. As a consequence, the impact of such new wind turbines on the voltage waveform is commonly negligible, and not a constraint for wind power development. 3 THERMAL CAPACITY AND STABILITY This section discusses challenges and opportunities arising when a large wind farm is connected in an area distant from the main transmission grid. Consider the following case. A large wind farm is to be connected at the remote end of a wide regional network that also connects nearby consumers (distribution grids) and some other local generation. The regional grid is connected to the main transmission grid via one or two long tie-lines. The example network is illustrated in Figure 3. Long distances and limited transmission capacity in the regional network may lead to thermal overload Main transmission grid (300 kv) Bus 2 Bus 1 66 kv Distribution grid Local load (L1) ~ Local generation (G1) Bus 3 Regional distribution grid 132 kv 22 kv ~ 690 V Figure 3: Example network with large wind farm. SVC Wind farm 100 200 MW and voltage stability problems during critical operating conditions. Two examples are used to describe the relevant problems, and to illustrate possible solutions that enable secure operation of large wind farms in the remote areas. 3.1 Example 1 This case relates to the example network shown in Figure 3. Normal considerations with respect to power system security require two separate feeders to the local distribution grid. Thus, in normal operation when both lines (bus1-bus2 and bus1-bus3) are in service, the grid is strong. With modest wind power penetration, the system can handle any single contingency, i.e. outage of any line, transformer or local generation, without loss of supply to any consumer load. Consider an outage of one of the two feeding 132 kv lines as the contingency case, and an operating situation with rather low local load and low local generation. In this case high wind power penetration may cause thermal overload on the remaining feeder (bus1-bus2) and also a risk of voltage collapse because of the increased need for reactive support on the weakened transmission. Figure 4 illustrates how the problem results in a serious voltage drop when the wind power generation exceeds 100 MW. 138 136 134 132 130 128 Bus 1 126 124 Wind farm 122 120 0 20 40 60 80 100 120 140 160 Wind power injected (MW) Figure 4: Voltage dependency of injected wind power. Voltage (kv) This problem, which would result in a total voltage collapse if the wind power generation exceeded 140 MW, can be solved by installation of a dedicated system protection scheme. A robust and inexpensive scheme could be designed based on monitoring of line currents, voltage drops and breaker operations. When current limits are exceeded or a line outage is detected, a signal is sent to automatically reduce the wind power generation or to disconnect a necessary number of wind turbines. 3.2 Example 2 Regional networks are not always operated according to the (N-1) security criterion. Consider the same case except that the second feeder (bus1-bus3) is not present. Again, with modest wind power penetration an acceptable level of security is maintained because the
local generation (G1 + Wind) is able to supply the main parts of the local load in case of a critical contingency. In this case there are two main challenges. High wind power penetration will in normal operation lead to voltage stability problems and thermal overload because of the limited transmission capacity. This is basically the same problem as the contingency case in section 3.1. Since this is now a problem in normal operation, a different solution is required to be able to utilise the full wind power capacity. As illustrated in Figure 3, a possible solution is to install dynamic reactive compensators (e.g. an SVC unit) for continuous voltage control. Figure 5 shows that by applying SVC units of various ratings, it is possible to increase the wind power installation up to above 160 MW before the voltage stability limits are reached. The second problem is when a critical contingency occurs, and the regional network is isolated from the main interconnection. In such a case the isolated network may have a large but varying surplus of power due to the wind generation. Application of energy storage and automatic frequency control of the wind turbines may be necessary in order to maintain a continuous supply of power in this situation. Phase voltage (kv) 0.5 0.4 0.3 0.2 Operating point at rated Wind power production SN = 80 Mvar SN = 90 Mvar SN = 100 Mvar SN = 120 Mvar SN = 140 Mvar 0.1 0 50 100 150 200 250 Wind power generation (MW) Figure 5: Voltage stability limits. 4 CONCLUSION Until the development of [1], there was no standard procedure for characterising the power quality of a wind turbine. Basically, the manufacturer provided rated data only, giving little basis for detailed assessment of the impact of wind turbines on voltage quality. Therefore, simplified rules were often applied, e.g. requiring the voltage increment due to a wind turbine installation to be less than 1 %. Application of this 1 % rule of thumb on our example network in Figure 1 would in fact allow only one 750 kw wind turbine. As demonstrated in section 2 however, [1] gives basis for detailed assessment allowing substantially more wind power to be connected. Further, section 2 demonstrates that possible voltage quality problems due to a small wind farm may be overcome simply by selecting an appropriate wind turbine type, and/or by adjusting wind turbine control parameters. Section 3 shows that assessment of thermal capacity and voltage stability constraints related to a large wind farm calls for additional analyses and innovative solutions. These are pinpointed by examples enabling increased penetration of wind power in weak grids. We have not made any detailed assessment of the costsavings associated with the suggested solutions compared to the alternative of conventional grid reinforcement by installation of new lines. This is because such numbers will be very case specific. We suspect however that for a majority of projects, the solutions considered will be more economic than installation of new lines. ACKNOWLEDGEMENTS This paper is to a large degree based on work carried out in conjunction with the projects Wind power in distribution grids and Large scale integration of wind power financed by the Norwegian Research Council and other Norwegian entities. Figure 4 and Figure 5 are courtesy of our good colleagues Trond Toftevaag and Magni Tor Palsson. REFERENCES [1] IEC 61400-21 (2001) Measurements and assessment of power quality characteristics of grid connected wind turbines. FDIS. [2] EN 50160 (1995) Voltage characteristics of electricity supplied by public distribution systems. [3] Tande JOG (2000) Exploitation of wind-energy resources in proximity to weak electric grids. Applied Energy 65 pp 395-401. [4] IEC 61000-3-7 (1996) EMC - Part 3: Limits Section 7: Assessment of emission limits for fluctuating loads in MV and HV power systems - Basic EMC publication. (Technical report) [5] Tande JOG, Relakis G, Alejandro OAM (2000) Synchronisation of wind turbines. Proc. of Wind Power for the 21st Century, 25-27 Sept. 2000, Kassel, Germany. [6] IEC 61000-3-6 (1996) EMC - Part 3: Limits - Section 6: Assessment of emission limits for distorting loads in MV and HV power systems - Basic EMC publication. (Technical report)