Parallel tap-changer controllers under varying load conditions (Part 1)

Size: px

Start display at page:

Download "Parallel tap-changer controllers under varying load conditions (Part 1)"

Reynard Harris
5 years ago
Views:

1 Parallel tap-changer controllers under varying load conditions (Part 1) by Prof. B S Rigby, T Modisane, University of KwaZulu-Natal This paper investigates the performance of voltage regulating relays used to control on-load tap changing transformers in a meshed distribution system. Real-time digital simulation, together with hardware-in loop connection of relays, is used to investigate performance under practical conditions, in particular, the responses of the relays to changes in network operating conditions during the daily load cycle. This article covers an investigation into the use of a numerical voltage regulating relay [1, 2] in distribution networks in South Africa. The relay is used to regulate the output voltage of on-load tap changing transformers, and uses control algorithms that prevent circulating reactive currents from flowing between interconnected transformers at different points in a network. The advantage of this technology is that it allows operation in a meshed configuration, providing better back up and security of supply. The algorithm needs no dedicated communication channels, making it cost-effective where substations are tens of km or more apart. It is not possible to test the algorithm or its settings using standard relay injection techniques as it is effectively a distributed control technique applied simultaneously at several locations, with the controllers interacting via the network. In the study a real-time digital simulator [3] was used to predict the performance of relays with the algorithm enabled prior to commissioning. The study also focused on determination of algorithm settings suitable for conditions encountered in practical distribution networks. Fig. 1: Distribution network under study. Early work [4] focused on developing a detailed real-time simulator model of the distribution network while subsequent work [5] developed a detailed real-time simulation model of the main control algorithms, to better understand operation of the relay and allow engineers to study its performance without the need for a real-time simulator. More recent real-time simulation studies, that considered how the regulators respond to typical daily fluctuations in load throughout a distribution system [6,7], show that careful consideration needs to be given to settings for the relays under the varying load conditions encountered. Study system Distribution network Fig. 1 shows the system, part of a distribution network in the upper South Coast region of KwaZulu-Natal, comprising an 88 kv substation at N which feeds three separate Fig. 2: Phasor diagram showing the effects of circulating current. energize - April Page 16

2 loads (Fig. 2). This forms the basis of the socalled Δcosϕ algorithm used by the relays to regulate circulating current [1, 2]. The algorithm monitors voltage and current at the secondary terminals of the transformer and calculates the power factor cosϕ act of the network as seen by that transformer The difference between this measured power factor and the actual power factor of the network at that point (entered as a setting cosϕ set in the relay) is used as an indirect measurement of the presence of circulating reactive current in the transformer as follows: I CIRC = I T sinϕ act I T sinϕ set (1) Fig. 3: Simplified block diagram of the relay s voltage and circulating current control algorithms. Fig. 4: System response following reconnection of the transformer at S (Δcosϕ control disabled). substations U, S and W. Each of these substations has a 20 MVA, 88/22 kv on-load tap-changing transformer supplying local loads (L 1, L 2 and L 3 ). Substations U, S and W also supply switching station E and substation F via 22 kv feeders which supply local loads L 4 and L 5 respectively. The tap position on each transformer, T 1, T 2 and T 3, is controlled by its own relay. The circulating current problem If control of the secondary voltages at each transformer is carried out independently(by changing the tap position and hence turns ratio), when T 1 is at a higher tap position than T 2, it will have a higher internal voltage in its secondary winding than T 2, driving circulating current from T 1 to T 2 via the two closed paths between substations U and S. This will be almost entirely reactive in nature, since the impedance in the closed paths is predominantly reactive. The phasor diagram in Fig. 2 shows that circulating reactive current adds to load current supplied by transformer T 1, but is subtracted from load current supplied by transformer T 2, causing the internal volt drop in transformer T 1 to increase, and that in transformer T 2 to decrease. This distortion in the voltage regulation and loading of the two transformers can cause transformer T 1 to tap up even further (to overcome its increased internal volt drop) thereby increasing the circulating current and worsening the problem. The regulating relay and its settings A symptom of circulating current flow is a variation in the power factor at the transformer secondaries from the true power factor of the distribution system and The Δcosϕ algorithm then generates two error signals, Y U and Y P [1] : Y U = (U act U set ) ΔU perm (2) Y P = I CIRC /I CIRC perm (3) where U act and U set are actual and desired (setpoint) values of transformer secondary voltage, and I CIRC perm and ΔU perm are the permissible circulating current and transformer secondary voltage regulation respectively. Error signals for voltage and circulating current are combined into a single error signa, Y, as follows Y = Y U + Y P (4) U s e o f t h i s s i n g l e e r r o r s i g n a l, Y, t o t a p u p t h e t r a n s f o r m e r (when Y < 1,0) or tap down the transformer (when Y > - 1,0) allows both voltage regulation and circulating reactive current to be controlled independently within their permissible deviations determined by the user settings in the Δcosϕ control algorithm [1]. Fig. 3 shows a simplified block diagram of the relay s voltage and circulating current control algorithms illustrating the user-specified settings required. The user designed settings are those inputs shown enclosed in hexagonal polygons in the block diagram, whereas all other inputs represent measurements made (directly or indirectly) by the relay. When designing the settings for these regulators, the U set and ΔU perm values are chosen as for a conventional voltage regulator. In the distribution network being considered, the utility practice is to choose U set to be equal to 103% of the nominal transformer secondary voltage (to allow for line volt drop downstream in the medium voltage network) and to set ΔU perm at 1,1% of U set. When choosing settings for the circulating current control algorithms, there is no established local utility practice to rely on. The manufacturer s recommendation [2] is that the value for I CIRC perm at each transformer be set to a minimum of 60% of the circulating current measured when that transformer s tap setting differs from those of the other transformers by two positions. energize - April Page 17

3 The network power factor setting, cosϕ set, at each transformer should, theoretically, correspond to the correct power factor of the distribution system and its loads, when there is no circulating current flowing between transformers, because any measured changes in the network power factor away from cosϕ set setpoints is used by the relay to sense circulating current in its Δcosϕ algorithm [Eqn. 1]. Under practical conditions, these correct values of network power factor can be expected to change over time (even when there is no circulating current). The choice of cosϕ set therefore requires some care, as will be demonstrated in the results. The real-time simulator The real-time digital simulator (RTDS) [3] is a multiprocessor computing platform designed to allow detailed mathematical models of power systems to be analysed in real-time, as well as power system controller hardware to interact with, and control (in closed-loop) the power system plant represented in the real-time simulation model. In the first phase of the study, three separate regulating relays were used in a detailed, hardware-in-loop study of the system shown in Fig. 1. The distribution system was modelled on the RTDS simulator, while the three hardware relays performed the closed-loop control of the transformer tap positions of the simulated system in real-time. In the second phase, a software model of the regulating relay was incorporated (ie. the entire system and its relay control was modelled on the RTDS). This study was carried out to assess the feasibility of developing models of the relay for other simulation tools. Figs. 4 and 5 show the results of two hardware-in loop real-time simulator studies for a typical network event. In both, the transformer at substation S was initially disconnected (i.e. circuit breaker CBS in Fig. 1 was initially open). The regulators at the remaining two transformers were allowed to adjust their tap positions and reach a satisfactory steady-state condition. The tap changer at S was then manually set to its nominal setting (position 5), and the transformer was reconnected. This test was carried out both with and without Δcosϕ control enabled on all relays. Fig. 4 shows the response of the system following the r econnection of the transformer at S with Δcosϕ control disabled on all regulators. Fig. 5 shows the response to the same event with Δcosϕ control enabled. The figures show the response of voltage and reactive power output measured at the secondary terminals of the transformer, as well as the tap position of each transformer at substations U, S and W. The two transformers initially in service Fig. 5: System response following reconnection of the transformer at S (Δcosϕ control enabled). (U and W) start off at relatively high tap positions (12 to 13) since they are under heavy loading conditions due to the absence of the transformer at S. In both studies when the transformer at S is reconnected, its voltage initially decreases as it comes under load, and its tap position is increased over time by its regulating relay to return to within permissible range. However, comparison of Figs. 4 and 5 shows that the extent to which the tap position of transformer S is increased is quite different depending on whether or not the Δcosϕ control is enabled. With no Δcosϕ control (Fig. 4) the regulator at S only raises the tap setting to position 9, whereas with Δcosϕ control (Fig. 5) the tap setting is raised to position 12. The behaviour of the regulators at U and W is noticeably different with and without Fig. 6: Actual time varying loads: (a) at L 1 ; (b) at L 2. the Δcosϕ control enabled: with no Δcosϕ control (Fig. 4) there is no change in tap setting at W, and the tap setting at U is increased to position 14. With Δcosϕ control (Fig. 5) both U and W tap down. With Δcosϕ control employed, all three transformers adjust their tap settings to approximately the same positions following the reconnection of the transformer at S; however, without Δcosϕ control employed, the final tap positions of the three transformers are markedly different. With no Δcosϕ control, the loading on both transformers U and W increases, despite there being no change in the actual (customer) loads at L 1 to L 2. This is caused by the circulating reactive current supplied to transformer S, that enables it to reach its set point value of secondary voltage without having to move to a high energize - April Page 18

4 tap position, thereby reducing the share of the system load it carries. When Δcosϕ control is employed, each regulator acts to minimise transformer voltage regulation and circulating current. In Fig. 5 the regulators at U and W decrease the tap settings of their transformers at some time after the transformer at S is reconnected. These changes reduce the voltages at these transformers (i.e. move them away from their voltage set points), and also reduce the circulating reactive current flowing out of U and W into S. Reduction in circulating reactive current flowing into S means that the regulator at S has to carry out a greater number of increases in tap position to restore the voltage at S to its set point value. All three transformers still return to permissible voltage levels, with no significant circulating currents flowing between them; at nearly the same tap positions; carr ying approximately the same load. Variations in network loads Figs. 4 and 5 demonstrate how hardwarein-loop real-time simulator studies are able to predict the impact of regulating relays control actions following significant network changes. Within a time frame of several minutes, it is reasonable to assume that the loads at L 1 have fixed magnitude and power factor. Under such conditions, each relay correctly interprets the change in the power factor at its transformer s secondary terminals as being due to circulating current. However, over a longer time frame, the power factor measured at each transformer could also change with genuine variations in the load conditions at L 1. In such cases, the relays Δcosϕ algorithms may misinterpret real changes in power factor as the presence of circulating currents and respond inappropriately. To study the behaviour of the Δcosϕ algorithm under realistic network conditions, data was gathered on the behaviour of the loads, with the assistance of Eskom Distribution personnel. The behaviour of the lumped loads at L 1 was determined for two typical 24-hour periods, in midsummer and mid-winter. In addition, customised models and techniques were developed for the RTDS to allow the actual time-varying loads at L 1 to be replicated during the real-time simulation to study the response of the relays. Fig. 6 shows the active and reactive power variations and power factor at two of the five loads, L 1 and L 2, over a 24-hour period. During a typical daily load cycle, the active component of load varies considerably whilst there is much less variation in the reactive component, and consequently the power factors of loads L 1 Fig. 7: Real-time simulator study to investigate the response of hardware relays to the actual 24-hour load cycle in the distribution system; fixed cosϕ set values designed for conditions at t = 10 hours. and L 2 vary between 0,87 and 0,98 (loads L 3 show similar 24-hour trends). Relay response to time-varying loads Fixed network power factor settings The real-time model of the system was then run over an extended time frame to assess the performance of the relays under varying load conditions. Customwritten, time-varying load components were used at each lumped load (L 1 ) to reproduce the known variation in these loads in a 24-hour load cycle. Fig. 7 shows the results of one such study where there has been no disturbance to the topology of the network to initiate tap changes: the regulators each adjust their tap positions in response to the normal variation in daily load. The network power factor setting cosϕ set in each relay was set at a fixed value corresponding to the power factor seen by that regulator s transformer at time t = 10 hours. Choosing this fixed cosϕ set value results in a setting close to the actual time varying network power factor for the longest possible period during the daily load cycle: Fig. 6 shows that the individual load power factors remain at, or near to, their values at time t=10 hours for an approximately 10-hour period of the day. The results in Fig. 7 confirm that with fixed settings for cosϕ set, use of the Δcosϕ algorithm with time-varying loads could cause undesirable behaviour. During the period when the relays cosϕ set settings are sufficiently close to the conditions in the network (8 18 hours) the regulators maintain the transformer secondar y voltages within the permissible deviation (indicated by the dotted lines on the plots) and there is no circulating current (transformers at the same tap positions at all times). However, under very light load conditions (0 5 hours) the network power factor is much lower than the fixed settings, and the regulators at all three transformers misinterpret this low network power factor as being due to circulating current, resulting in a large negative value of YP. This (incorrect) negative error YP in circulating current is added to the genuine error Y U in secondary voltage at each transformer to give a combined error signal Y that incorrectly lies within the permissible deviation (±1) (the permissible deviations in Y are also shown by dotted lines on the plots). Consequently, each transformer s secondary voltage remains lower than the minimum permissible value during the period 0 5 hours because of the incorrectly determined error in circulating current. Similarly, during the peak-load period of the daily load cycle (18 21 hours) the network power factor is higher than the fixed cosϕ set values in each regulator. The regulators misinterpret this high network power factor as being due to circulating current, this time resulting in a large positive value of Y P ; this incorrect positive error Y P swamps the genuine negative error Y U in secondary voltage so that, although the combined error Y lies within its permissible energize - April Page 20

5 deviation, the transformer voltages remain higher than is permissible during this threehour period. It should be noted that the error Y P in circulating current is a result of the set-point network power factor value not being updated to track varying load conditions, and not as a result of actual circulating current flow (there are no differences in tap positions between transformers at any time during the day and hence no actual circulating currents). Contact Prof. Bruce Rigby, UKZN, Tel , brigby@ukzn.ac.za v energize - April Page 21

Impact of Distributed Generation on Voltage Regulation by ULTC Transformer using Various Existing Methods

Proceedings of the th WSEAS International Conference on Power Systems, Beijing, China, September -, 200 Impact of Distributed Generation on Voltage Regulation by ULTC Transformer using Various Existing