IEEE Transactions on Energy Conversion, Vol. 6, No. 1, March 1991

Minimum Excitation Limiter Effects on Generator Response to System Disturbances

J.R.Ribeiro, Member Florida Power & Light Co.

Miami,FL

29

Abstract Minimum excitation limiters(MEL) have been in use since the first

applications of voltage regulators to synchronous machines [2] and are presently in widespread use throughout the power industry. However, a cursory literature search [l-51 indicated scant examination of the dynamic behavior of these devices. Although the minimum excitation liiiter is rarely called into action its effect can be particularly significant to the final outcome of some severe disturbances. This paper describes the digital computer modeling of two common MEL devices and addresses their effect on power system dynamics.

Key Words: Power System Dynamics, Minimum Excitation Limiter, URAL, Underexcited Reactive Ampere Limit, Steady State Stability, Excitation System Modeling, Underexcitation LimitefiUEL).

INTRODUCTION Minimum Excitation Limiters (MEL) or Underexcitation Limiters

(UEL) were introduced in the 1940's along with the continuous acting voltage regulators for two major purposes-

Y

1. Prevent generator operation below some excitation level that is associated with excessive armature core end heating caused by leakage flux [6]. 2. Prevent generator operation beyond the steady state stabdity limit [ l ,2 ]

Most interconnected systems utilize the limiters for the functions in item 1 above, but there are some instances where the limiter is set with the aim of preventing loss of synchronism in the under excited region. Some UEL models act on the voltage e m r function of the voltage regulators.However a large number of UEL devices in use today were designed to have their output fed into a high value gate in the excitation control, as shown in the block diagrams of figures 1 and 2. When the UEL set limit is reached the UEL takes over the voltage control until the generator is within the limits. This paper will focus on two UEL devices of the gate operating type, supplied by two different manufacturers. Three different types of excitation systems are examined:

-High Initial Response (HIR) static excitation -HIR rotating brushless excitation -Conventional (about 2.0 response) brushless excitation. The behavior of a plant with two units of different size and

characteristics is also examined. The scone is limited to two 'pes of UEL devices, but the results and conclusions are representative o a large number of underexcitation limiters in service.

In the next section we will discuss the UEL modeling. Following, in the Results Section, we will first discuss scenarios with machine initial conditions beyond the stability l i i i ts determined by the methods in [ l ] and then examine the UEL effect in conditions of excessive system reactive as exemplified by islanding scenarios after underfrequency load shedding.

There are three main objectives in this paper. 1. Illustrate the dynamic behavior of the UEL. 2. Demonstrate that setting the UEL limits to less than what is required by the rated thermal capability, for stability reasons, is an unwarranted constraint on the generator reactive capability.Uti1ities that are in this category can then realize measurable savings on reactor installations or at least improve operational flexibility. 3.Demonstrate that representation of the UEL dynamics is vital

5'0 Si4 426-7 EC A paper recommended and approved by t h e IEEE Energy Development and Power Generation Committee of t h e IEEE Power Engineering Socie ty f o r p r e s e n t a t i o n a t t h e IEEE/PES 1990 Summer Meeting, Minneapolis, Minnesota, J u l y 15-19, 1990. submitted January 19, 1990; made a v a i l a b l e f o r p r i n t i n g Nay 18, 1990.

Manuscript

to the proper analysis of power systems under degraded conditions following major disturbances.

MODELS The UEL models addressed in this paper have not been programmed

for digital computers and made commercially available. Figure 1 shows a block diagram for the model UEL-1 added to a modified IEEE type AC2 exciter model suitable for representation of high initial response (HIR) rotating brushless type excitation systems. The Volts/Hz limiter shown in figure 1 had to be also represented because of its significant interaction with the UEL. Figure 2 shows the UEL-STl block diagram added to a modified IEEE type ST1 excitation system model suitable for representation of static HIR systems. The equations described in the block diagrams were programmed into a dynamic simulation program package currently in use at Florida Power & Light Co.(FPL). Figure 3 shows a normalized capability curve with the steady state stability (SSS) limit of generator B , of figure 4 superimposed. The SSS limit was calculated following the derivation in the appendix of reference [l]. The current practice is to set the UEL limit with a 15% margin of the capability curve or the SSS manual control limit whichever is more restrictive. Whether the l i i i t is from the generator capability or from stability, the characteristic is described by the equation of a circle. Both UEL models compare the radius of the l i i i t circle with an off-set center in the Q- axis of the P-Q plane with a radius calculated from the instantaneous values of real power and reactive electric power. These values after some algebraic manipulation are expressed by V, in figure 1 and V, in figure 2 as a function of terminal voltage and current. In figure 1, the limit radius V, is compared with the instantaneous calculated radius at the adder. Stabilization, V,, is provided by a take-off from the exciter feedback loop multiplied by a constant and coming to the same adder with V, and V,. The same comparison scheme is shown in figure 2, except that stabilization for the model shown in fig. 2 is provided by the last block in the UEL-STl block diagram.

The seven machine eleven bus system shown in figure 4 was used for all simulations. Most of the analysis could be done with a two- machine system, but using a more complex system helps assure that multi-machine effects would not be ignored. This network is not representative of a normal power system but the generator or generators and their connection to bus 1 could be representative of a power system in a highly deteriorated state at some time during or after a Severe disturbance that caused loss of several lines. All units were represented by round rotor generator models with two circuits in the d-axis and two circuits in the q-axis. The analysis focused on machines labeled A and B. Machine labeled "W" has a very large MVA rating (30,000) to represent the rest of the interconnected world. Data for the lines (R,X,B) is shown in fig. 4. Data for the computer simulation models are given in the appendix. The load was represented as:

P= P4(O.1+0.9v) Q= Qv'

Where v is the bus voltage as a function of time and the subscript i indicates the initial nominal value of the load. This is a common representation for USA systems.

RESULTS For all the results being discussed the generator or generators being

tested are connected through a step up transformer to bus 1 in figure 4. We will first discuss the cases were the generator is initially operating beyond the steady state stability limits calculated according to reference [l]. Then we will discuss the cases were the generator is initially within limits set at the UEL. Two generator sizes were tested. The static excitation system is associated with a 500 MVA unit(generator B) and the rotating brushless exciters are associated with a 1000 MVA unit(generator A). Since the calculated stability limit is only slightly tighter for the 1000 MVA unit, the same setting was used for both units. According to current practice, a radius and center would have to be calculated such that the limiting curve would be no less than 15% more restrictive than the S S S limit curve and the generator underexcited capability of figure 3. For simplicity and for the purposes of this paper it was adequate to set the limits equal to the radius and center of the S S S limit curve shown in the figure.

1

- Underexcitation Limiter rype U E L l with modified IEEE type AC2 exciter model.

EPD

F& - Underexcitation Limiter type UEL-STI with modified IEEE type STI exciter model.

- Normalized typical generator capability curve showing also the Steady State Stability Limit for Generator B .

_ _ - BUS TRANSFORHER -- .r GENERATOR -- 6 CRPRClTOA -- A

m F a - One line diagram for the network used in the simrrlations.

Table 1 below lists initial conditions for the groups of simulations

Table I- Initial Conditions for the Simulations.

performed. ...................................................

?5FTJ a ANGLE FLOW REF.FIGS. B U S 4 - 1

D.U MW MW Mvar 1.57 0.95 450 -174 33.0 588 211 5 thru 10 1.73 0.95 900 -366 34.2 635 256 11 thru 14 1.61 0.95 375 4 44.4 153 39 15 thru 17 1.79 0.95 750 25 46.4 153 42 18 thru 22 1.83 0.95 750 53 38.7 231 58 23 thru 24 1.68 0.95 375 27 35.2 " 11 1 , *I

..................................................... Generator Initiallv Outside of UEL Limits.

For this series of simulations the generator tested is initially operating at 0.9-j0.34 clearly outside of the calculated stability limit circle shown in figure 3. Generator B,with static HIR system, in figure 4 is connected through a step-up transformer and a line to the rest of the system,

enerator A(brushless HIR) is off line for the first three conditions d. iscussed below. Then two simulations are discussed were generator A is on line while generator B is off line.

Automatic Voltage RecrulatoHAVR) on Manual:- This simulation is intended to show that the generator is actually beyond its steady state stability limit assuming constant excitation. A 3-phase 4-cycle duration short circuit at bus 1 was simulated at time 4 . 1 s (seconds). Figure 5 shows a time plot of P (real power), Q(reactive power), ET (generator terminal voltage) and EFD (Generator Field Voltage). For this and all other similar plots, the power is expressed in p.u. of 100 MVA, generator terminal voltage in p.u. of rated generator voltage and the Field Voltage in p.u. of field voltage required to generate 1.0 p.u in the air gap with the generator open circuited. The variable range is indicated in parenthesis next to the variable symbol. The figure shows that by approximately 8 s the generator looses synchronism indicated by the collapse in power and voltage followed by the slip frequency oscillations. By definition of steady state instability the loss of synchronism has to occur for any minor disturbance as in fact it would for this case. The short-circuit simulation was selected to expedite matters.

AVR on Auto, UEL OffStatic HIR Exciter : Figure 6 shows the same disturbance as figure 5 except that the voltage regulator is on automatic operation. The power and voltages behavior indicate that the generator is stable with a comfortable margin. This is an expected result but clearly contradicts the need to resuict the lead reactive power output to about 0 p.u. as implied by figure 3.

AVR on Auto, UEL on. Static HIR Exciter: Figure 7 shows the same disturbance as figures 5 and 6, but for this simulation the AVR is on automatic mode and the UEL is turned on at time zero-plus with limits set as shown in figure 5. This is a somewhat contrived situation designed to illustrate the effect of the UEL since the unit could not be operating with the same initial conditions if the UEL had been on all the time. The plots show that the system is now not as well behaved as in the simulation with the UEL off. Further, the type of pulses seen at about 9.5 s of the simulation illustrate what happens if an operator tries

31

,

to drive the generator into a region beyond the set limits. These pulses are occurring when the radius calculated from instantaneous terminal conditions equals the limit radius and a quick transition from UEL control to voltage control ensues. This fact is portrayed in figure 8 showing time plots of the calculated radius times terminal voltage (VJ, the limit radius times terminal voltage (V3 and the UEL output The figure shows clearly that at about 9.5 s the Calculated and the Limit Radius are equal for a fraction of a second. During that time the UEL is switched from a positive value to a minimum negative value. When +e UEL output goes to minimum control is switched to the voltage error function. That is reflected by a short pulse on EFD , ET, P and Q as shown in figure 7.

Figure 9 will provide additional insight on the generator behavior. It shows the apparent impedance trajectory (Z-locus) in the Z plane where three circles of interest are shown. From the one with smaller radius to the larger the circles represent respectively the boundaries of the Zone 1 Loss of Field (LOW protection, Zone 2 LOF and the UEL set limit corresponding to the one shown in the S-plane of figure 3. Displaying the trajectory on the Z-plane has the advantage that the limit circles do not have to be changed for different terminal voltages. Conversion from the S-plane to the Z-plane can be quickly accomplished by the equations shown in [SI if one does not feel l i e deriving i t Here it can easily be seen when the apparent impedance goes outside the limits (enters the circle.) In the S/vz-plane a multitude of circles would have to be drawn since the limit radius and center are a function of the square of the voltage. The figure shows that at t =O s the apparent impedance is inside the largest circle (UEL limits) and that it is settling near the circle as indicated by the higher concentration of points there. Figure 10, shows the corresponding plot for the simulation where the UEL is off. It shows that even without the UEL the apparent impedance settles outside of the LOF circles.

AVR on Auto. UEL on, Brushless HIR Exciter: Generator labeled A on figure 4 is on and generator B is off line for this simulation. The operating point is the same in per unit of generator MVA rating as for the corresponding simulations of the static exciter. Figure 11 shows the time plots of P,Q,ET and EFD for simulation of a short circuit as described in association with figure 7, the only difference being that generator A is equipped with a rotating brushless HIR excitation system as shown in figure 1. In figure 11 we detect a persistent 1.7 Hz oscillation which was not present in figure 7. Though the oscillation might at first be attributed to dynamic instability (power system stabilizers were not represented), the actual cause of the oscillation is the interaction of the V/Hz function and the UEL. Figure 12 shows the time plots of the V/Hz output, the UEL output as well as the CALC-RAD and LUIIT-RAD where one can verify this phenomena. Because the V/Hz is set for 1.05 p.u voltage for this sirnulation it will take conmI whenever voltage goes over this value. The phenomena does not occur with the static HIR because the particular V/Hz limiter of the static HIR is inoperative for frequencies above about 56 Hz. For further verification of this fact we run an aditional simulation setting the V/Hz to an arbitrarily high value (4.0 P.u.) The resulting behavior, shown in figure 13, clearly verifies this hypothesis.

AVR on auto.UEL on BrusMess Conventional Exciter: The exciter model of figure 1 was made to represent a conventional response excitation system by use of proper data as shown in the appendix. Then the same short circuit simulation was performed with generator A on line as described in the previous section. A time plot of powers and voltages for this simulation is shown in figure 14. When compared with the Brushless HIR response we notice that the 1.7 Hz oscillation vanishes after t=5 s. The reason is that the voltage forcing function is not strong enough to drive the terminal voltage above 1.05 p.u. However the phenomena of conflicting objectives of the V/Hz and UEL is still present and could surface under different conditions.

Generator Initially Within UEL Limit - Loss of Load For this series of simulations the generator is operating initially at

0.75 real power and 0 reactive power in p.u. of generator M V A rating. At time = 0.1 s a shunt capacitor of 0.74 p.u of the generator rating is switched on at bus 1. This simulation is representative of a condition that the system may be at some time after islanding and associated load shedding. The resulting unloading of many transmission lines is associated with a net excess of capacitive reactive power seen by the generator. Without UEL, if the excess capacitive power is large enough the voltage regulator will drive the field voltage to zero, or negative values, depending on the type of equipment The exciters discussed in this paper will not produce negative voltages. With field voltage very low loss of synchronism may occur or the LOF relay may operate. The UEL

I I I I I I I I I

1u.

F&.J - Voltage regulator(VR) on manual control. Short circuit on bus 1. Gen. A off line.

I I I I I I I I I I

13.u I * O M 2 1 w 1 1 . # I O * T,me,t L * W $ l l l l l I I l I l I I J 0.0

- VR on auto. Skort circuit on bus 1 . Static HlR exciter. UEL on. Gen. A off line.

acts to limit the minimum value of excitation. Depending on the amount of excess reactive power the UEL action can cause sustained high voltages that may also be intolerable and cause over voltage pmtedon schemes to trip the generator and arresters to exceed their thermal capabilities. For the type of excitation system in figure 1 -Rotating

32

Brushless Exciters- we will see that the V/Hz limiter tends to annul the effect of the UEL. Transformer saturation effects were not represented, but the size of the capacitor switched was chosen such that final voltages would not much higher than the knee of typical transformer saturation curves thereby making saturation effects negligible. The following discussion and figures will better illustrate these points. For all cases that follow the AVR is on automatic operation.

Static HIR: Figure 15 shows P,Q,ET and EFD time plots for the simulation with generator B on line (generator A off line). The response is very well damped but voltages are settling around 1.15 p.u. due to UEL action. Figure 16 shows result of the same simulation except that the UEL limit radius was set to 2.04 p.u and the center to 1.54 p.u which are the settings appropriate to prevent operation with excessive rotor end iron heating. It is seen that even here the settling terminal voltage is not acceptable. If no UEL was present for this disturbance figure 17 shows that field voltages would stay too long at zero and the generator was on the verge of voltage collapse. Although we are not showing a plot here, the apparent impedance trajectory for this case penetrates zone 2 of the LOF relay for a few cycles and it is clear that a slightly more stressing condition would cause generator trip.

Rotating HIR: Figure 18 shows P,Q,ET and EFD time plots for the simulation with Generator A (lo00 MVA) on line. Here we notice an approximately 0.5 Hz oscillation which was not seen with the static exciter in figure 15. This oscillation exists even when setting the UEL limits to a larger radius and center (2.04 and 1.56 p.u. respectively) as would be recommended by the manufacturer. Again this is caused by the interaction of the V/Hz limiter with the UEL limits as can be verified by comparison with figure 19 which shows the same simulation as in figure 18 with the exception that the V/Hz limiter was made inoperative. However, in both simulations the terminal voltage is settling at unacceptably high values. Figure 20 shows the same variables for the simulation where both the UEL and V/Hz limiters are made inoperative. The generator terminal voltage collapses and pole slipping occurs as evidenced by the high frequency oscillations around t ime6 s.

Rotating Conventional Excitation: A similar set of simulations to the ones desc.ribed in the two previous paragraphs were run with Generator A equipped with a conventional response (ASA response of 2.0) brushless excitation system. The same model shown in figure 1 was used with constants modified to represent the non-HIR exciter (see appendix for values.) Generally comparable overall results were observed as for the simulations with the Rotating HIR excitation, as shown in figures 21 and 22.

Multide Units at the Same Station: Similar Simulations to the ones described in the previous paragraphs were run for the following conditions:

a) Generator A (lo00 MVA) on line with HIR rotating exciter and Generator B (500 MVA) on l i e with Static HIR exciter. b) Generator A (lo00 MVA) on line with conventional rotating exciter and Generator B (500 MVA) on line with Static HIR exciter.

In both sets of simulations the limit radius and center for the UEL setting were the same for both generators in p.u. of generator rating. It was observed that the results were similar to simulation of single machine at the station. However the conditions are aggravated by the interaction of the V/Hz limiter and UEL of Generator A which propagate the severe oscillations to Generator B. This behavior is illustrated in Figure 23. The disrupting oscillations are not as significant when the V/Hz is made inoperative as shown in figure 24.

In all of the simulations discussed the generators connected to busses 2,3 and 4 showed negligible response. This was expected since the disturbances simulated were designed to have localized effects.

CONCLUSION Two UEL models were programmed and merged with a power

system simulator package commercially available. The effect of the added models was illustrated in several different scenarios.

From the facts presented above the following additional observations and conclusions can be formulated:

a) Using the UEL to prevent the generator from operating in the steady state instability region unnecessarily restricts the generator operating range. The restriction is unnecessary because the current utility practice for calculation of steady state stability assumes manual control of voltage. As illustrated in this paper a condition that is unstable under manual control is comfortably stable when the automatic control of voltage is active.

b) When a generator or group of generators is, for whatever reason, isolated with an excess of capacitive load the effect of the UEL is to drive the terminal voltage higher than it would go without the UEL.

Depending on the amount of excess capacitive load the voltages might be intolerably high causing further system disruptions. Moreover for some excitation systems, it would not take much of an excess capacitive load to drive the voltage to a point where the V/Hz limiter in the excitation system becomes active and interacts with the UEL in a disruptive way. The important message in this paper is that in simulation studies of islanding conditions or other degraded state scenarios the UEL and V/Hz must be represented in order to obtain a more realistic evaluation of the system. That these effects are not often represented is attested by the non-existence, until now, of commercially available simulation models of two relatively common types of underexcitation limiter as the ones discussed in this paper.

c) The results showed that relying on the corrective action of the UEL to protect the generator during islanding conditions is unwarranted because if such situation arises it is most likely that the UEL will force the generator to unacceptably high voltages that will cause other control or protection to become active and eventually nullify the UEL action. Therefore, the only uue protection would be to fmd other solutions to prevent the generator from facing these stressing system conditions.

In view of these observations it is proposed that the role of the underexcitation limiter be reevaluated. It seems that the only unquestionable need for the underexcitation limiter is to prevent stator end iron heating when the rotor field is low. This being a thermal effect it most likely does not require action in the transienvdynamic time frame. The subject will certainly require further analysis from the power industry community and this paper will have achieve its most important objective if the community is challenged to additional investigations in this subject.

1 j I I I I I I I J ".$om T,me,S , 1010 ' m*n lu.u 2 OPOI u.0

Fip.8.- Same conditions as in figure 7: Vk, Vc and I UEL 4' output. I l l 1

X-oworen1 t 1

t L

i-

I I I

F& -Same as figure 7. Apparent impedance trajectory(2-locus.) LOF = loss of field relay.

I I I I I I I I I : f

X-OoDorent 0 .32 - i

Fin.lQ - Z-Locus. VR on auto. Short circuit on bus 1. Static HIR exciter. UEL turned off. Gen. A off line.

, , !, :: , .

L I IjP.(# I 1 Tlrne,s I I I ! .oo'o I :1u.u 1 '1.1100 u.u Fia.lJ - Same description as jig.11 except here VIHz was made inactive.

F M - Rotating HIR exciter. VR on auto. 3-phase short circuit on bus 1. UEL and VIHz on. Generator B off line.

- Rotating Conventional exciter. VR on auto. 3-phase short circuit on bus 1. UEL and VIHz on. Gen.B off line.

1 t 1

Fin.lZ - Same description as Fig.11. UEL and VlHz output variables. F M - Static HIR. VR on auto. UEL on. 370 Mvar capacitor switched on at t=.l s. Gen. A off line.

i-

1 1

/ j

/ I ?.,, / / / j

, : ; . . . . .,,. ....... ................ !!!.lor ...... * .................... , I : : ".-A 1 : ~., 1 i ETlU.5,1.51

F M - Same description as fig.15. UEL is on with manufacturer recommended setting (R=2.0,C=I 5 P.u.)

- Same description as figure 18, except VIHz is off:

1

-7 - Same description asfig.15 except UEL is off. F M - Same description as figure 18, except UEL and VIHz are off.

I I I I I I I I I 1

, EFU(-1,9)

0.u I 1100 %OW* Tlw,~ b W P I I . W W 10.

Fin.lS - Rotating HIR. VR on auto. UEL and V I " on. 740 Mvar capacitor on at t=0.1 s. GenB off line. Fin21 - Rotating conventional exciter. VR on auto.UEL and VIHz on. 740 Mvar capacitor on at t=0.1 s. GenB off line.

35

1

k -1

I I I I 1 . 1 I I I J '-4684 TlmC,S '.U" I.- 10. I.",, 0.u

- Same description as f ig21 except VIHz is off.

b.u I ! .nw I *.*U I me,s I &.*e* I I A** I 10. ' Flp2j - GenA wizh rotating HIR, UEL and VIHz 0n.Gen.B with Static HIR, UEL on. VR on auto for both generators. 1110 Mvar capacitor on at e0.1 s.

I.

- Same description as figure 23 except VIHz is turned off.

REFERENCES [l] W.G.Heffmn, R.A.Phillips,"Effect of a Modem Amplidyne Voltage Regulator on Undemxcited Operation of Large Turbine Generators", AIEE Transactions, V01.71,August 1952.pp.692-697,Appendix In. [21 A.S.Rubensteii, M. Temoshok, "Undemxcited Reactive Ampere Limit for Modem Amplidyne Voltage Regulator.", AIEE Transactions PAS,

P I R.A.F'hillips,A.S.Rubenstein, "Operation of Large Synchronous Generators in the Dynamic Stability Region with a Modem Amplydine Voltage Regulator." Part I and II. AIEE Transactions PAS, Vol 75,

[4] I. Nagy,"Analysis of Minimum Excitation Limits of Synchronous Machines.", IEEE Transactions on PAS, Vol. PAS-89, N.6, pp.100~-1008, July/August 1970. [5] Applied Protective Relaying (book), Westinghouse Electric Corporation, Relay Instrument Division,Cod Springs=. [6] S.B.Famam,RW.Swanout, "Field Excitation In Relation to Machine and System Operation" AIEE paper 53-387, AIEE Fall General MeetingSamas City&lO,November 24,1953 [7] Benjamin C. Kuo, "Automatic Contml Systems",Pmtice-Hall hG Englewood Cliffs"., 1967

VO1.73, pp.869-874, August 1954.

pp.762-771, August 1956.

APPENDIX - DATA

Generators A,C,E: Rated lo00 MVA. Equipped with Rotating Brushless Exciters

T 'W T"W T'QO T"QO H D XD 6.54 0.06 0.73 0.070 3.3 0.0 1.8

X'D X'Q X"D XL S(l) S(1.2) XQ 1.75 0.44 0.57 0.29 0.22 0.08 0.459

-HIR Rotating BNSMWS With UEL-1 Undemxcitation Limiter and V/Hz limiter. TRTB TC KA TA VAMAX VAMIN KB

VRMAXVRMIN 100.0 -90.0 TE K L K H K F T F K C K D K E

1 .o 15.0 1.0 0.03 1.0 0.487 0.31 1.00 VLR E l S(E1) E2 SG2) 9.95 3.14 0.028 4.19 0.101

0 0 0 400.0 0.05 8.0 -8.0 30.0

. .. .

-UEL-1 data: CL RL CC K U L K F L T U L W M X V U M N 0.34 0.966 1.0 4.4 3.3 0.050 8.3 -8.3 -v/Hz data:

K1 K2 K3 1.05 82.2 0.93

-Conventional Rotating Brushless (ASA response about 2.0) With UEL-1 Underexcitation Limiter and V/Hz limiter.

TRTB TC KA TA VAMAX VAMIN KB

VRMAXVRMIN 10.0 -9.0

0 0 0 250.0 0.050 10.0 -9.0 1 .o

TE K L K H K F T F K C K D KE 1 .o 15.0 0 0.030 1.0 0.487 0.311 1 .o VLR E l S(E1) E2 S(E2) lo00 4.9 0.05 6.5 0.39 -uEL-1 data: CL RL CC K U L K F L T U L W W 0.34 0.966 1.0 100.0 1.0 0.050 8.30 -8.30 _ _ -v/Hz data:

K1 K2 K3 1.05 82.2 0.93

Generators B,D.F: Rated 500 MVA

T 'W T " W T'QO T"QO H D XD XQ X'D 3.70 0.060 0.46 0.060 2.85 0 1.60 1.53 0.25 X'Q X"D XL S(1) S(1.2) 0.45 0.2 0.145 0.093 0.43

36

Static HIR Excitation System (Generators B,D,F) With UEL-STl Underexcitation Limiter. TRVIMAX VIMINTC TB KA TA 0 0.20 -0.20 1.0 5.0 200.0 0.010 VRMAX VRMIN KC KF TF 7.0 0 0.12 0 1.Ooo

KRU KCU KF'U KIU T1 T2 T3 0.966 0.34 0.007 10.0 6.4 0.8 0.640 VKMAX VCMAX VUMAX VUMIN KIM

-UEL-ST1 Data:

4.0 4.0 0.2 -0.2 -0.012 Note: V/Hz for Static HIR system modifies the voltage reference set

point of the voltage regulator. The frequency setting is adjustab1e.Manufacturer recommends setting to about 56 Hz. With this setting the V/Hz would be inactive for all the simulations discussed and therefore was not represented.

ACKNOWLEDGEMENT The author wishes to acknowledge the co-operation of Messrs.,

M.L.Crenshaw of the General Electric CO and J.D.Hudey of the Westinghouse Corp. in supplying block diagrams and data for the underexcitation limiters as well as reviewing some results, and Messrs. J.W.Shaffer and R.Rey of FPL for helping in the initial debugging and for estimulating discussions.

j. R. Ribeiro (M'73) was born in Sa0 Paulo, Brazil. He received the B.S.E.E. degree from New York University, New York, in 1973 and the M.S.E.E. degree from Union College, Schenectady, New York in 1974. He has been employed by the Florida Power & Light Company since March 1984 where he is a Principal Engineer in the System Planning Department. Previously he worked in different capacities in the area of power system planning for American Electric Power Service Corp., New York, New York (1969-1973). Power Technologies, Inc., Schenectady, New York (1973-1977) and Niagm Mohawk Power Corp., Syracuse, New York (1977-1984).His fields of interest inciude power system modeling for dynamic and transient analysis, as well as transmission and generation planning.

Mr. Ribeiro has co-authored several technical papers and is a member of Tau Beta Pi and Eta Kappa Nu. He serves on IEEE working groups on Excitation Systems and Switching Surges.

He is a registered Professional Engineer in the State of New York.

3 1 associtated with that function operates. Thus, a MEL should be set to coordinate properly with the loss of field protection. If the loss of field protection is set based on the steady state stability limit, then an approapriate MEL setting would also be based on the steady state stability limit, with a reasonable amount of separation. If the MEL setting is selected based on the generator underexcited capability curve, it should be checked for proper coordination with the loss of field relay.

Manuscript received August 10, 1990.

Discussion

F. P. de Mello, B. K. Johnson and L. N. Hannett (Power Technolo- gies, Inc., Schenectady, New York): The discussers would like to com- mend the author on a well written and interesting paper.

We agree that representation of the MEL may be vital when operation in the underexcited region is simulated. We have studied a disturbance where a loss of excitation relay is known to have tripped a critical unit even though the steady state characteristic of the MEL suggested that the relay would not operate. Dynamic simulation of the disturbance with the MEL represented however indicated that the loss of excitation relay operated on a swing in spite of the MEL. Representation for maximum excitation limiters and volts per hertz limiters may also be vital for some simulations. Fortunately for most simulations these limiters come into play only transiently and thus do not significantly impact the results. However, in cases of islanding as shown in the paper, they have significant influence on the systems performance.

This paper shows the importance of letting a stability program user customize dynamic models for particular simulation needs. As the author points out commercially available stability programs do not provide repre- sentation of minimum excitation limiters for all types of excitation sys- tems. There are however standard models available in the simulation package supplied by the discussers company which represent both maxi- mum and minimum excitation limiters for several excitation system types. There is an industry need for models of minimum and maximum excitation limiters and it is hoped that the E E E Excitation Performance and Model- ing WG will address this need.

The simulated interaction between the minimum excitation limiter and the volts per hertz limiter is quite interesting. Have such oscillations been observed for actual equipment?

Finally we agree that in many situations it is preferable to have the minimum excitation limiter be an alarm rather than a control override function. This is particularly the case where load rejection can occur leaving units charging high voltage lines. Manuscript received August 10, 1990.

JOSEPH D. H U R L E Y (Westinghouse Electric Corp., Orlando, FL): This paper provides some interesting concepts regarding the behavior of minimum excitation limiters (MELs) during severe dynamic conditions. I strongly agree with the authors conclusion that simulation studies representing such limiters are important when unit dynamic o p e r a t i o n f a l l s w i t h i n t h e l i m i t e r c h a r a c t e r i s t i c s . T h i s p a p e r s h o u l d encourage others to do s o when applicable.

The simulations presented in this paper include a very high impedance transmission system in various abnormal operating modes. In such cases it is important to consider the possibe adverse effects of the MEL and its setting in response to these severe conditions.

H o w e v e r , in most power systems the transmission system is stronger than this, and as a result the MEL setting is much less restrictive than that shown in Figure 3 . In such cases, dynamic operation within the MEL characteristic may not be a major concern. Does the author know of MEL applications on his system where the MEL setting is s o restrictive?

One important benefit that a MEL provides is that it prevents the inadvertent lowering of excitation (either by the operator or voltage regulator misoperation) below a level which will cause the generator to either lose synchronism or trip due to loss of field protection. This consideration makes the MEL a valuable device on most large generators.

In selecting the setting of any limiter, it is important that the limiter be allowed to operate before the protective trip device

D.C. LEE, R.E. BEAULIEU, P. KUNDUR, G.J. ROGERS, Ontario Hydro, Toronto, Ontario, Canada. This paper illustrates the difficulties which may be encountered when normal and protective controls interact to the detriment of power system stability. It is important to design sufficient control and protective features to ensure that no damage is done to generating units. However, at the same time these controls should not unnecessarily restrict operation or adversely affect the performance of the interconnected power system.

If an Underexcitation Limiter P E L ) is required only to respect thermal operating limits, then a slow-acting control is all that is required, preferably one which acts to modify the lower limit of the voltage setpoint. The same argument applies if the UEL is designed to prevent an operator from lowering the setpoint into an unstable operating region when operating on manual control. The problems with the UEL arise when attempting to apply it as a fast- acting controller with the voltage regulator in service. A high- speed voltage regulator, frequently with a power system stabilizer, normally allows stable operation in the underexcited region well beyond thermal limits. Consequently, if the limiter is adjusted to respect thermal and manual control stability limits, it will unnecessarily restrict operation with the voltage regulator in service, as shown to a limited degree in this paper. Also, in many cases if the UEL is designed to prevent the unit from going out of step during large transients, a much more restrictive limit start curve may be required. In some cases that we have examined and tested (albeit with UEL signals that were summed rather than being gated), it would have been necessary to set the limit to start near unity power factor at rated power in order to meet the above conditions and have a damped response when driving hard into the limit. Some of our field tests have clearly shown oscillations in terminal voltage, power, etc., due to inter-actions between the AVR and the limiter, without a V/Hz limiter to compound the problem.

Because the UEL limits the ability of the excitation system to keep the terminal voltage from rising excessively, the overvoltage experienced under some emergency system conditions may cause other system problems, some of which are mentioned in the paper. Could the authors comment on the use of a terminal voltage limiter set at 1.05 pu? Is it the use of the UEL that dictates the selection of such a restrictive terminal voltage limit with the unit on line? Does this not cause operating problems under normal operating conditions? Are there conditions under which the voltage ratings of unit breaker could be exceeded if the UEL were employed without the terminal voltage limiter? We have examined the use of these controllers several times but have generally ended up leaving them out of service, particularly if the excitation system incorporates a power system stabilizer. We would agree with the author that if high-speed UELs are to be employed, they should be modelled carefully, with their excitation

38 systems and examined in simulation studies. To the extent possible models should be validated by field testing.

Manuscript received August 3, 1990.

M. L. Crenshaw and M. Cardinal (GE Company, Schenectady, NY): As observed by Mr. Ribeiro in his opening remarks, scant attention indeed has been paid to either the steady-state settings or the dynamic perfor- mance of the control known by many terms but generically as the under excitation limiter (UEL); even less to the relatively newer V/Hz limiter controls. During final commissioning at the plant, frantic activity fre- quently occurs as settings must be selected. Occasionally control instabil- ity has been encountered, often solved by trial and error until recent (unpublished) analytical studies have provided a more scientific basis. This paper should help spur some useful dialogue and industry activity in this

Historically, the UEL was considered as a control of last resort whose function was to prevent operator manipulation or power system demands from:

area.

1. exceeding assigned general operational guidelines. 2. exceeding generator thermal limits. 3. allowing operation in a region of conditions which would result in

loss of stability if the voltage regulator were to be removed from service.

Considerations 1 and 2 above are obviously warranted for unattended stations and units which, due to their rating or location, can become readily overloaded when attempting to hold voltage levels constant. In light of the results presented in the paper, Item 3 above deserves further consideration. Can the author comment on the dilemma facing the plant operator, i.e., the potential for loss of steady state stability if the automatic voltage regulator is removed from service?

The author has also illustrated an often voiced concern-a power configuration which results in excess capacitance beyond the UEL setting of the unit. Unacceptably high voltages can occur as the control changes from the normal voltage regulating mode.

Considering the multitude of types of UEL controls of different vintages and from various manufacturers, as well as difficulty in obtaining accurate simulation data, does the author recommend extensive system studies incorporating detailed control models? It would seem that recognitation of UEL operation from P, Q or R, X plots (where UEL control is not simulated), leading to subsequent studies and development of relaying and switching strategies might eliminate, in most cases, the need for detailed UEL analysis.

Both UEL and V/Hz control performance specification requirements are suggested as appropriate working group or task force activities.

Manuscript received August 13, 1990.

J. R. Ribeiro, FPL Co., Miami, FL: I would like to thank each of the discussers for their enlightening contributions. I will address explicit questions from each discusser and end with comments of a more general nature.

Messrs. Lee, Beaulieu, Kundur and Rogers indicated that as a result of their own analysis and field experience, they have carried the concepts of my paper to their logical conclusion-leaving the UEL out of service. This fact certainly promotes investigation of current applications in other systems. Concerning the use of V/Hz l i t e r set at 1.05 pu., I can say that this practice is being revised by our System Protection Section. It was based in the concept that the limiter was to be set lower than the corresponding protective tripping device. Only a few of our units are of the type modeled by figure 1 of the paper. These are the ones more likely to enter into undesirable interaction with the AVR, but in our system they also happen to be operated very close to 1.0 pu voltage. For this reason, I suspect, we have not experienced frequent operating problems. It is important to notice that for the model in figure 2, even when the V/Hz is active, it modifies the reference voltage in the automatic voltage regulator (AVR) providing a smooth transition. On the contrary, in the model of figure 1 the V/Hz limiter completely takes over the voltage control function as soon as it becomes active.

Messrs. Crenshaw and Cardinal voice very important factors in the evaluation of need for UEL. Let me attempt to answer the two explicit questions from these discussers. First, removing the AVR from service is

an absolute last resort measure. When that is done, operators can be read limits on a unit P-Q capability curve and follow them. As for the second question, I do recommend extensive system studies incorporating detailed control models for the UEL. That does not mean that every system study requires the detailed control model. Recognition of the UEL operation from R-X plots is not sufficient because it would not reveal the generator response to the UEL. The R-X plots could serve as a screener for cases requiring detailed simulation but that would require repeating some cases. Since the major effort is in collecting the data and setting up the models, one might as well have the models as part of the data base. The incremental effect of adding the UEL model in the running time of a stability simulation is negligible. Of course, at the present stage a great deal of effort would be required to define models and collect data for the different types of UELs in service. Therefore, I agree wholeheartedly with the discussers suggestion for a task force on UEL.

Messrs. de Mello, Johnson and Hannet describe an event that supports the need for detailed simulations. A simple R-X plot in that case would have indicated UFL operation, but could lead to the erroneous conclusion that UEL action would prevent operation of the loss of field (LOF) relay. The question of whether we have observed the oscillations caused by interaction of UEL with V/Hz I have partially answered when addressing Messrs. Lee et al., above. Id like to add that the oscillations may have occurred with their true cause unrecognized due to the unawareness of that potential interaction. The fact is that we have no records of oscillations having occurred specifically caused by the referred interaction. I agree that the ability to customize standard models in simulation programs is very helpful when conducting the type of investigations reported in the paper.

Mr. Hurley points out some relevant considerations in the evaluation of the UEL. In addition, he states that in most power systems the transmis- sion system is stronger than the one used in the paper and therefore the UEL setting would be less restrictive. While I agree with the first part of Mr. Hurleys statement the second part does not necessarily follows. The reason is that the generator connection to a strong transmission system is not always strong. There are many systems in this country that evolved from lower voltage level systems and for that reason many generators are weakly connected to the strong part of the system. In addition, the common method for setting UEL presented by Hefion et al., [l] specifies steady state stability calculations with voltage regulator on manual and strongest tie out of service. Using these conditions, several of the units in the FPL system would require even more restrictive limits than the one used in the paper. Having been involved in determining underexcited steady state stability limits for other systems I know that this is not a problem affecting only FPL. Of course, many engineers end up ignoring the limits calculated with this method resulting that different companies adopt different rules for setting the UEL. As for the UEL being effective in preventing generator trip due to loss of field protection, let me refer to the discussions of Messrs. de Mello et al., relating an incident where the UEL failed to prevent generator trip due to LOF. This agrees with the fact implied by Messrs. Lee et al., and also illustrated in the paper, that very restrictive limits are required if the UEL is to prevent LOF operation during large transients caused by excessive system reactive capacitance. Such limits would interfere with normal operation of the generator.

I will finalize my closure by strongly recommending that engineers reevaluate the use of UEL in their own companies. When doing so, I would suggest the following ideas to ponder:

1. Thermal protection of the generator does not require action during transients therefore the UEL could activate an alarm or be very slow acting.

2. Stability limits can be implemented by operating rules based on simulations with the AVR in service.

3. In the event of system overvoltages caused by excess reactive capacitance, as occurs in islanding conditions, the UEL can be effective in preventing loss of synchronism provided that the V/Hz limiter is of the type shown in figure 2 of the paper. If it is of the type shown in Figure 1 the V/Hz will take over and push the field to zero causing loss of synchronism. Even in the types where the V/Hz effect can be ignored, a UEL setting restrictive to normal operation of the system is required to be effective in preventing loss of synchronism. In this case the resulting excessive overvoltage generally leads to generator trip by the overvoltage relays. While it seems preferable to trip on overvoltage protection than on out of step protection, it is questionable whether it is economically advisable to restrict normal operation to achieve this objective, particu- larly given that islanding is an uncommon event.

Manuscript received November 2, 1990.