It is common practice to operate substation transformers in parallel. The operational benefits of paralleling transformers include improved maintenance, reliability and power quality. Also, distributing the load evenly usually extends the life of transformers.

Typical Methods

Operating the parallel paths often involves three-phase power transformers equipped with load tap changers (LTCs) or, by user preference, non-LTC transformers in conjunction with step-voltage regulators. When the latter is selected, the regulators are most often of single-phase construction.

When operating the LTCs in parallel, the transformers (or regulators) must be kept on as nearly the same tap position as is realistically possible. When the taps are not on the same position, the paths in parallel do not exhibit the same voltage transformation ratio, and a circulating current is driven through the transformers and the associated bus work. This is undesirable because it represents a real power loss.

The two procedures most commonly used to accomplish the parallel operation are known as the “circulating current method” and “master follower method.” The circulating current method dominates in the United States, but the master follower method is mainly used worldwide. Each of these methods has advantages and disadvantages. The key common disadvantage is the necessity for communications between the controls of the tap changers of the paralleled devices. This usually is accomplished with circuits interconnecting the controls, where the objective is either to signal the respective controls of the circulating current so that the next transformer to accomplish a step change is the one that will bring the tap positions nearer to the same position, or to signal from the “master” transformer to the “follower” that the master has performed a tap change operation and the follower needs to accomplish the same action. With the circulating current method, it is accepted that the tap positions may be one or more steps apart for extended periods.

Another little-known procedure, the negative (or reverse) reactance method, can be used to control the apparatus in parallel. This procedure was used in the past but fell from favor due to an inherent error in the operation or from the difficulty in selecting the setpoints for its use. The technique is seldom used today despite several positive attributes, the most notable being that it doesn't require any intercontrol wiring. Each control operates wholly independently of the other, yet the controls accomplish the objective of biasing the sensed voltage so that, as with the circulating current method, the controls command the tap changers to maintain equal tap positions. Perhaps the most interesting fact, in this regard, is that every LTC control in service that conforms to the applicable standard, ANSI/IEEE Std. C57.15, already includes all the provisions necessary to accomplish the negative reactance paralleling.

Understanding Negative Reactance Control

As is explicit in the name, negative reactance paralleling uses the negative polarity capability of the reactance setpoint of line drop compensation. To understand the principle, consider that line drop compensation (LDC) is normally set on an LTC control to model the system to control the voltage at the load, which may be some distance from the transformer or regulator location. Normally that load will be at a high lagging power factor and the resistive component of the LDC (LDCR) will dominate the correction. When transformers operate in parallel, identical except for their respective tap positions, the circulating current is almost wholly reactive; that is, low power factor. The LDC component dominating the correction in that case is the reactive component of the LDC (LDC X).

In the usual case, a lagging power factor load and setpoint values of LDC R and LDC X are positive so both elements will act to boost the voltage at the substation bus to hold the voltage desired at the location of the load. The setpoint voltage is the voltage desired at the load (see the phasor diagram in Fig. 2). The reference phasor is typically about 118 V to 124 V. If there were no load current, or if the LDC R and LDC X values were zero, then the IR and IX phasors would be of zero magnitude. With IR = 0 (V_R= 0) and IX = 0 (V_X = 0), the voltage at the load would be calculated by the control model to be identical to that at the substation bus. As line drop compensation comes into play, the V_R and V_X phasors result in a boost voltage required at the bus to compensate for the drop in the line.

Now consider what would happen if the LDC X value is “reversed” or made negative. The lagging reactive current will multiply by the LDC X (negative) value to buck the voltage or lower the tap position. Similarly, a leading reactive current will multiply by the LDCX (negative) value to boost the voltage, or raise the tap position. It must be anticipated that identical transformers operating in parallel will occasionally be on different tap positions. In this case, the transformer on the higher tap position will drive a current that lags the voltage. By the nature of the system, that current will lead the voltage in the transformer that is operating on the lower tap position. Figure 3 illustrates the situation with the transformers on different tap positions. The circulating current is “outbound” in the higher tap transformer (seen as a lagging power factor) and “inbound” in the lower tap transformer (seen as a leading power factor). The net effect of this circulating current acting on a negative value of LDC X is to bias an LTC driving a circulating current (on a higher tap position) to “lower” and to bias an LTC receiving a circulating current (on the lower tap position) to “raise.” Thus, in a manner similar to the circulating current method, this satisfies the overall objective of having the tap changers stay on the same or nearly the same tap position while properly regulating the system voltage.

Hill Street Substation

The Hill Street Substation of NYSEG serves the downtown area of Hornell, New York, U.S. Contrary to the more common industry practice, NYSEG operates much of the system, including Hill Street, ungrounded at 4800-V open delta. Planning indicated the need for a parallel transformer/regulator installation.

A cost study was made for the paralleling control to operate the regulators' tap changers. To do that by the usual circulating current method took an additional US$30,000. It was decided this was a good application in which to apply the negative reactance paralleling method, as it would involve no additional hardware or installation cost. All needed control functions are provided as standard in the control routinely supplied with the regulators. The regulators operating in open delta did introduce an additional complexity but that was peculiar to the connection and not to the negative reactance application. The Hill Street regulators are believed to be the first open-delta application operating in parallel by the negative reactance method.

Hill Street Application

Figure 1 shows the regulators at Hill Street Substation. There are four single-phase regulators produced by Cooper Power Systems, each 933 kVA, 7620 V (1225 A) operating ungrounded at 4800 V. In the open-delta application, two of the regulators are said to be “leading” and two are “lagging,” referring to the fact that the current signals from the regulator CT lead or lag the voltage signals by 30 degrees at unity power factor load. This fact is a non-issue as the phasing corrections are made automatically in the associated Cooper CL-5E controls. The two leading and the two lagging single-phase regulators are each in series with a 7500 kVA, 34.5- to 5.04-kV three-phase transformer wound delta-delta. The transformers each have 6.7% impedance to limit the current that will circulate when the regulators are not matched in tap position.

This system is modeled in an Excel spreadsheet. The calculation accepts as inputs the presumed substation loading in terms of magnitude and power factor. This was based on about 6000 kVA of load at 0.99 lagging power factor. In sequence it is determined that:

• The current that will circulate in the substation when there is a one tap position step discrepancy between two regulators is 69.6 A. This is the reactive current in the 4.8-kV bus (the ICIRC current of Fig. 3), which is scaled to 13.9 mA on the 200 mA base of the control and is sensed by the LDC algorithm of the controls.

• The value of the LDC X setpoint is iteratively selected. The circulating current found above will act on the setting chosen. The value selected is effectively a sensitivity adjustment for the response of the control to the one-step tap position discrepancy. The value may later be adjusted to produce greater or lesser sensitivity to the extreme ranges of load magnitude and power factor anticipated for the substation. The value finally selected for use at Hill Street is LDC X = -10 V, which produces a response in the control of 0.60 V when one-step discrepancy circulating current flows.

• Any control error is absolved for the selected load condition by selection of a positive LDC R value used in conjunction with the negative LDC X value. There is a value for LDC R that will make the control error exactly zero for the presumed load. This is found by multiple iterations to be LDC R = +1.6 V.

Different load conditions are tested to observe the error introduced as the power factor varies.

The actual load power factor at Hill Street stays uncannily close to 1.0. The table on page 37 highlights the study results.

A value of LDC R = +0.5 V is used at the Hill Street Substation. The band center voltage is 123 V with a bandwidth of 2.5 V. It is anticipated that the control error will stay below 1 V.

As a backup, the voltage limit control on the CL-5E is set for 120 V and 126 V. Should the system load change to an extent that the error is > ± 3 V, the VLC will restrict the output voltage to those limits.

The CL-5E control offers the opportunity to record selected operating parameters. “Snapshots” are recorded every 15 minutes for 30 hours. Per the associated log, the tap position of paralleled regulators always stayed within one tap position of each other. Of the 120 recordings for each pair of regulators, the regulators operating in parallel were recorded to be operating on the same tap position 87% of the time. This is excellent operational performance.

The controls were set to hold the output voltage to 123 ± 1.25 V. Remarkably, in only one snapshot, and then on only one pair of regulators, did the record show the voltage digression to be more than 1 V from the 123 V set-point. This performance is more meaningfully stated to note that the arithmetic mean voltage for the 480 recorded values is 122.8 V with a variance of only 0.266 V.

Conclusion

Negative reactance paralleling has seldom been used for many years due to a combination of factors, including the difficulty to establish control setpoints, know when the control is operating correctly and overcome the perception that the technique is fundamentally inaccurate. With modern controls, mathematically efficient calculating procedures and improved system operation (especially as due to maintaining improved system load power factor), these factors disappear or are of much less concern.

As evidenced by the experience at the Hill Street Substation of NYSEG, there are applications where the procedure is efficiently and cost-effectively employed. Others may gain appreciable cost savings from use of this technique.

Chien C. Tschang is a senior planning engineer at New York State Electric and Gas Corp. He received a BS degree from Rochester Institute of Technology and a MBA degree from Binghamton University. Tschang has 13 years of experience in the distribution planning field and provides technical assistance to district offices in the areas of general planning studies, voltage flickers studies, construction forecast reviews and reliability studies. He also has been involved with specialized studies that assess the impact on the distribution system from line losses and distributed generation. cctschang@nyseg.com

James H. Harlow is an independent consultant specializing in matters of application and design of transformer tap changing under load and step-voltage regulators, having previous experience in these activities with Siemens Energy and Automation Inc. and Beckwith Electric Co. His professional experience includes serving the IEEE Power Engineering Society as vice president of Technical Activities and as chairman of the Transformers Committee. He is the editor of Electric Power Transformer Engineering, CRC Press, 2004. j.h.harlow@ieee.org www.harlowengineering.com

Modern Considerations of Negative Reactance Paralleling

While the procedures for negative reactance paralleling have been known for many years, it has been the advent of the modern digital LTC controls and computers that have revived the negative reactance method of paralleling.

Digital controls are precise in setting parameters and far more accurate than controls from 50 years ago. Past application guidelines stated that compensator settings should be determined by trial-and-error on the system, apparently because the alternative, manual calculations, would be much too tedious. Today, operating setpoints can be determined with a much greater degree of accuracy using computer software to model system conditions.

Digital controls can be programmed to record and display the present and past operating state of the system. This information is invaluable when using the negative reactance paralleling procedure because some voltage error is introduced due to variations in load power factor and magnitude. The error can be constrained to acceptable values by judiciously selecting the line drop compensation setpoints using system load information. Even greater accuracy can be obtained where shunt capacitors are used to maintain a high load power factor. The voltage limit control feature found on all digitial controls can be used to avoid an aggravated problem if system conditions digress from those originally anticipated.

The key benefit of this method is the elimination of control communication, which provides cost savings and improves reliability by removing potential control issues with the master/follower and the circulating current methods.

Voltage error, on 120-V base, encountered as load power factor digresses from the nominal value.

	Study presumes nominal load power factor of
	1.00			0.99 lag
	The control error in volts is			The control error in volts is
If the actual load pf is	For X=-15 V (R=0.3 V)	For X=-10 V (R=0.1 V)	For X=-8 V (R=0.1 V)	For X=-10 V (R=1.6 V)	For X=-8 V (R=1.2 V)
0.94 Lg				0.73	0.60
0.95 Lg			0.88	0.62	0.51
0.96 Lg		1.01	0.79	0.50	0.42
0.97 Lg		0.88	0.69	0.36	0.31
0.98 Lg	1.06	0.72	0.56	0.20	0.18
0.99 Lg	0.76	0.51	0.40	-0.01	0.01
1.00	0.01	0.01	0.00	-0.52	-0.39
0.99 Ld	-0.74	-0.49	-0.40	-1.02	-0.79
0.98 Ld	-1.05	-0.69	-0.57		-0.95
0.97 Ld		-0.85	-0.69		-1.08
0.96 Ld		-0.98	-0.80
0.95 Ld		-1.10	-0.89