REVIEW NEGATIVE POWER FACTOR TEST RESULTS AND CASE STUDY ANALYSIS Long Pong Doble Engineering Company

1. INTRODUCTION

The negative watt loss or power factor is often considered as a strange test result and can lead the tester to confusion in the interpretation. This negative measurement can occur in any apparatus testing when the UST or GST-Guard test mode is used, and had been known and described since 1948 in several Doble conference papers 1, 2, 3 . In addition to what has already been reported, this paper provides the basic theory with the mathematical model and basic electrical circuit in order to fully understand the phenomena and to interpret the test results. Also the case studies will be presented and analyzed accordingly to the basic theory. 2. BASIC THEORY

In a parallel circuit as shown in the figure 2.1, each element (Z1 and Z2 ) is subjected to the same voltage in amplitude and phase. So the current depends only on the element impedance. VT I1 Z1 I2 Z2 I2 = VT / Z2 (b) I1 = VT / Z1 (a) VT I1 = I2 Z1 V2 Z2 V2 = VT Z2 Z1 + Z2 I2 = V2 / Z2 = VT Z1 + Z2 (c) (b) (a)

Equation 2.1
Basic electrical parallel circuits FIGURE 2.1 Basic electrical serial circuits FIGURE 2.2

Equation 2.2

In a serial circuit as shown in figure 2.2, the voltage across each element can be different in amplitude and in phase if the two elements are of different nature (resistor, capacitor or inductor). The voltage across Z2 is V2 and in function of the two elements Z1 and Z2, as shown in the equation 2.2b. Each element impedance Z can consist of resistive R or reactive X components (either inductance XL and capacitance XC) and can be represented by a vector in the vector diagram, see Figure 2.3. The resistance is a real number and in phase with voltage. The two types of reactive components are a complex number with 90 degrees out of phase with voltage, but opposites to each other:

Z = R + XC (or XL ) XL R Reference Voltage, VT Where: XC = 1 /(j2fC) XL = j2fL (b)

(a)

(c)

j is complex number, j2 = -1 XC Vector diagram FIGURE 2.3 Z Equation 2.3

If another element ZG in parallel (//) with Z2 exist as shown in figure 2.4, then the voltage V2 and its current I2 in function of three elements become more complex as shown in equation 2.4b and 2.4c. However the only mathematical difference from the previous model is the term Z1 Z2 /ZG . VT I1 Z1 IG
ZG
G

I1 = I2 + IG V2 V2 = VT x

V2 = VT x Z2

Z2 //ZG Z1 + Z2 // ZG

(a) (b)

I2 Z2

Z1 Z2 + (Z1 + Z2 ) ZG I2 = V2 / Z2 = ZG VT Z1 Z2 + (Z1 + Z2 ) (c)

Basic serial and parallel circuit

Equation 2.4

In insulation system the elements Z1 and Z2 are mainly capacitive XC, and the element ZG (shunting path) is mainly resistive, RG . The figure 2.4 circuit and the equation 2.4c can become:

VT I1 XC1 V2 IG RG I2 XC2 XC2 V2 = VT x (a) XC1 XC2 + (XC1 + XC2 ) RG I2 = V2 / XC2 = VT XC1 XC2 + (XC1 + XC2 ) RG An insulation circuit with leakage path to ground FIGURE 2.5 From the Equation 2.5, we can conclude the following points: 1) The shifting phase angle to the negative real value of I2 in figure 2.5 is due to the term (XC1 XC2 /RG ). The product of two complex numbers (XC1 and XC2 ) is a negative real value or the product of two vectors (XC1 and XC2 ) results in a vector (X) with phase angle shifted by the sum of the phase angles of the two multiplier vectors, as shown in figure 2.6. In an insulation system, the term (XC1 + XC2 ) represents the total insulation system impedance (XC). Its value should stay unchanged over time for a good insulation; If the construction of insulation system is known, we can use the terms XC1 XC2 /RG and XC to interpret the decreasing power factor, as suggested in the case studies; Equation 2.5 (b)

X= XC2 XC2

2 XC1

XC2 XC = XC2 + XC2

2)

Vector diagram of capacitive reactance calculation FIGURE 2.6

3)

4)

Equation 2.5a shows that the V2 is out of VT phase by the angle because of the term (XC1 XC2 /RG ). The measured watt loss through XC2 is negative only when the test voltage VT is used as the reference for current measurement, -R in the figure 2.7, while the voltage across XC2 is indeed V2. If V2 was used as reference then the current would be positive value (r). Therefore physically no real negative watt exist.

I V2 -R r VT

5)

The location of the leakage path to ground will determine the degree of its Measured current and real current influence in the insulation system. If it occurs at the end of insulation system, either XC1 or XC2 will be extremely small, so the product of FIGURE 2.7 XC1 XC2 will be also very small. This product increases gradually when the 2 location moves toward the middle. The maximum product will be (XC /4) at the middle of the insulation system and decreases when the location moves further toward the other end. This means that the leakage path to ground will have: a) A minimum influence at either end of the insulation system and;

b) A maximum influence at the middle of the insulation system (XC/2); 6) The resistance RG value of leakage path to ground will determine whether the path can have the influence or not: a) If RG is much greater than (XC2 / 4), then the leakage path to ground has little influence in the insulation system regardless the location.

b) On other hand if RG is extremely small (dead short-circuit), the leakage path to ground can have an influence anywhere in the insulation system. c) Between the two extreme cases (a and b), the influence degree of leakage path to ground depends on the relative value of (XC1 XC2 /RG ) comparing to the total insulation system impedance (XC). The higher negative value indicates that the leakage path to ground has higher influence, see case 3.1.

CONCLUSION Now we can apply the basic theory above in the interpretation of the power factor test results in two different types of insulation system: Insulation system initially without a shunting path and Insulation system initially with a shunting path. Insulation system initially without a shunting path: For this type of insulation, the initial test result is positive. The power factor or watt loss decrease will indicate an introduction of a shunting path which is brought out of meter. This indicates the change of insulation condition. The decreasing rate will depend on the position relative and resistance value of shunting path within the insulation circuit. For the insulation system that has the immobile shunting path location, for example the winding of a cast coil transformer (see case study 3.4a), the decreasing of power factor indicates the increase of the shunting path conductivity or the deterioration of the insulation itself. For the insulation that has a mobile shunting path, the decreasing of power factor can indicate the progression of the shunting path toward the middle or the increase of shunting path conductivity (see case study 3.3a). Insulation system initially with a shunting path: For this type of insulation, the initial test results are negative. Most of this insulation type has a permanent solid grounded shunting path for example the inter-winding insulation of a two winding transformer with a grounded shield between the two windings (see case study 3.3c) or the interphase insulation of a stator winding (see case study 3.3d). The power factor or watt loss tends to increase toward positive value and this indicates the deterioration of either the shunting path or the insulation itself.

3.

CASE STUDIES 3.1. Experimentation Using the circuit of ten TTR capacitors in series with resistor shunted to ground to simulate the leakage path to ground as shown in the figure 3.1. This experiment helps to visualize the equation 2.5b, specifically the influence of term (XC1 XC2 /RG ) , and the negative measurement phenomenon. XC1 11 10 9 8 7 6 5 XC2 4 3 2 1

RG

Experimentation circuit FIGURE 3.1 Test Results Node # 1 2 3 4 5 6 7 8 9 10 11 mA 3.66 3.65 3.64 3.63 3.63 1.88 3.62 3.63 3.65 3.66 3.67 RG = 5 M ohms Watts %PF 0.02 0.05 -1.77 -4.84 -3.11 -8.54 -4.07 -11.18 -4.62 -12.72 -15.88 -84.47 -4.61 -12.71 -4.06 -11.18 -3.12 -8.55 -1.77 -4.83 0.02 0.05 Cap. 971 968 962 958 954 265 953 957 964 970 972 mA 3.66 3.63 3.60 3.60 3.60 3.58 3.58 3.60 3.63 3.65 3.67 RG = 11.5 Watts 0.02 -0.31 -0.62 -0.87 -1.04 -1.23 -1.32 -1.27 -1.03 -0.77 0.03 M ohms %PF 0.07 -0.87 -1.72 -2.41 -2.9 -3.44 -3.67 -3.53 -2.83 -2.12 0.07 Cap. 972 962 955 953 954 948 950 953 961 968 974

Analysis: These test results clearly demonstrated how the shunting path out of the meter influents the insulation test data: upon the location and the resistance of the shunting path. Regarding the location, the shunting path has a minimum influence at either end (node 1 and 11) and a maximum influence at the middle (node 6). In between the two points, the power factor changes accordingly to the term (XC1 XC2 /RG ), as shown in figure 3.1b. Regarding the shunting path resistance, it determines the degree of the influence: the lower resistance is the higher influence it has. The experimentation had also demonstrated another important point regarding the AC and DC resistance measurement. Indeed the RG=11.5M ohms resistor consisted of ten resistors in series with 10 M ohm each. In DC measurement, an ohmmeter indicated about 95 M ohms, but in AC, the resistance was 11.5 M ohm. The resistors were ceramic types and laid on the lab floor, so the capacitive coupling could happen making the difference in DC and AC resistance measurement. This phenomena was described in detail by the paper [4] in 1998 Doble Client Conference. Therefore it would not be sufficient to judge a contaminant just by performing the DC testing.

Figure 3.1b Shunting Path Influence

-20 -40 -60 -80

10

12-1
-2 -3 -4

-100 Location (Node #)

3.2. Capacitor Suspending a capacitor in air for testing as shown in figure 3.2 could result in lowering the power factor. In fact the grading capacitor for Alsthom, 300kV, type FX 22, SF6 circuit breaker vintage 1991 was tested with negative power factor in this configuration.

Crane Sling

RG XC 1
M

XC2

Mid-point hanged capacitor and test circuit FIGURE 3.2 Test Results (10kV equivalent) Test connection kV MA UST- hanged 10 4.5 UST- mounted 10 4.504

Watt -0.014 0.038

PF -0.03 0.08

Cap (pF) 1193 1194

Nameplate Cap (pF) 1200 1200

Analysis: The position of mid-point hanging divides the capacitance in two equal values that maximizes the term XC1 XC2 /RG . Therefore it is an ideal position for having negative watt loss when the sling becomes more conductive and the coupling effect occurs. When the capacitor was mounted on the circuit breaker, the sling no longer existed. So the mid-point path was eliminated and the power factor became positive.

3.3. Bushing a) Surface Contamination: The oozing problem, as shown in figure 3.3a1, resulting in the negative power factor of the ABB O plus C bushings was reported in 2000 and discussed in Doble Bushing Committee. Test Results (10kV equivalent) Bushing H1 S/N 3718620391 C1 -0.23% Inverted C1 0.07% C2 0.34%

Analysis: The core of condenser type bushing consists of multiple layers of insulation as shown in the figure 3.3a2. The surface contamination of low porcelain provided to C1 a ground path which shunted the C1 circuit at the point where the contamination had reached. In the early stage (1), the shunting point was near the ground sleeve, this results in a very small value of XC12 and so do the term XC11XC12 /RG . With the progression of the contamination down to the middle of the porcelain sleeve, stage (2), the shunting point was also moved toward the middle of the C1. Consequently the XC12 increased and the term XC11XC12/RG become important, this results in lowering the power factor relatively from the stage 1, see the figure 3.3a3 a). When the inverted C1 test was performed, the test voltage stressed more the first coupling link which was near the ground point, see the figure 3.3a3 b). This resulted in smaller value of XC12 , so the term XC11XC12 /RG became smaller also. Consequently the power factor increased even though the bushing defect was the same.

Bushing with S/N 3718620391 Bushing with low porcelain contamination FIGURE 3.3a1

C E N T E R C O N D U C T O R

CA CB CC C1 CD CE CF CG CH CI CJ

C2
Tap Electrode

CK

Grounded Layer/Flange

Surface Contamination

Simplified test circuit of low porcelain contamination at early stage FIGURE 3.3a2

XC11

XC12

RG

XC 2

C E N T E R C O N D U C T O R

CA CB CC CD CE CF CG CH CI CJ

CK

CC A E CB N CC CD T CE E CF CG R CH C CI O CJ N CK D U C T O R
2

XC11

XC12

XC11

XC12

RG

XC 2

RG

a) Standard C1 Test Circuit b) Inverted C1 Test Circuit Simplified test circuit of low porcelain contamination at advanced stage FIGURE 3.3a3

b) Bushing with internal insulation problem [5] During the transformer routine test on September 16, 2001, Bonneville Power Administration (BPA) obtained the negative C1 power factor from LAPP, POC-A, 34.5kV bushing vintage 1968 mounted on Y2 phase of a single-phase 300MVA, 550/230/34.5kV autotransformer. The C2 test was not able to be performed at any voltage. Test Results (10kV equivalent) Test connection kV mA Watt PF (%) UST- C1 test 10 0.551 -0.18 -3.3 UST- C1 test 2 0.540 -0.16 -2.96 Guard-C2 test 0.5 Test set tripped off Analysis After the bushing tear down , BPA found that the binding tape of the last insulation layer was peeled off and came to contact with the internal surface of bushing flange, as shown in the figure 3.3b. The bounding tape was a heat cured epoxy material (non-conductive material). Therefore, some sort of contamination was deposited on this bounding tape bridged to the bushing flange. As a result, this contamination deposited on the surface of paper created a leakage path to ground. For the last 30 years in service this contamination could have increased and consequently increased tracking to ground. This case illustrated very well the influence of the leakage path to ground conductivity (RG ) on the power factor measurement, while the XCXC stayed constant in the process. The bounding tape bridge provided the leakage path to ground for the C1 insulation in the figure 3.3b 3. At the early stage, the bounding tape was not contaminated and good insulation, so the leakage path to ground had a high resistance (RG ). Thus the term XCXC /RG was insignificant, because of RG >> XCXC. However in service this link was subjected to high electrical stress that can cause the partial discharge and contamination. Eventually it became more conductive, then RG decreased. So the term XCXC /RG has became important decreasing the C1 power factor as predicted. It is also interesting to notice that this problem was discovered by C1 test after 30 years of service, because the C2 test was not performed as the routine test. However we believe that the C2 test would be more effective and would detect this problem earlier.

Cap (pF) 146 143.1

Nameplate Cap (pF) 360 360

1) Bushing core

2) Bushing tap schematic

Bridging between last layer near electrode and flange

C1

XC
XC is small and RG low (through contaminant) So the term XCXC RG

XC RG

is still significant

3) Insulation circuit schematic

4) Internal flange with bounding tape left

5) Bounding tape on bushing core Bushing with internal insulation problem FIGURE 3.3b

10

3.4.

Transformer

a) Cast coil two-winding transformer National three-phase 2250kVA , 34/.48kV transformer vintage 1984 was found with the negative CHL power factor. # Insul. kV mA Watts %PF Cap(pF) ------------------------------------------------1 CH + CHL 10 4.772 0.092 1265 2 CH 10 1.522 0.120 0.79 403.50 3 CHL(UST) 10 3.252 -0.020 -0.06 862.50 4 CHL 3.250 -0.028 -0.09 861.50 5 CL + CHL .5 15.230 3.059 4040 6 CL .5 11.990 3.051 2.54 3179 7 CHL(UST) .5 3.254 0 0 863.20 8 CHL 3.240 0.008 0.02 861 Analysis: The cast coil construction divides the CHL insulation system in three sections, see figure 3.4a1: 1. 2. 3. CHa is the insulation from the high voltage coil conductor to the solid insulation surface; Ca is the air gap from high to low voltage solid insulation surfaces. This element has very low watt loss and the smallest capacitance, so dominates the circuit; CLa is the insulation from the low voltage coil conductor to the solid insulation surface.

When the surfaces of the solid insulation is contaminated, example condensation mixed with dirt, the surface conductivity will increase. This will add the fourth element which are mainly resistive elements (RHg or RLg ) and shunting the CHL circuit to ground, see the figure 3.4a1. Because the air gap dominating rule in the circuit, the CHL test voltage will drop significantly across the air gap. The surface nearest to the ground has little influence, so it can be negligible. We can visualize the negative power factor phenomena by simplifying the test circuit to the figure 3.4a2. The two circuits suggest that the CHL test at high voltage winding will be more sensitive to detect the HV solid surface condition and the CHL test at LV winding will be more sensitive to detect the LV solid surface condition. The contaminated surface can cause the superficial partial discharge and eventually lead to permanent surface damage. The test results should be compared with the initial test data on the unit prior to commissioning, with the previous test data or with the test data for unit with similar insulation system. Any change, either higher or lower, power factor values should be investigated.

11

Surface contamination

HV winding

LV winding
CORE

Air gap

CHL CHa Ca CLa

RG1

RG2

Air gap = Very low loss and smallest capacitance

Ground Potential

Typical cast coil winding and test circuit FIGURE 3.4a1

CHa

Ca

CLa

CHa

Ca

CLa

RG1

RG2= Negligible

RG1= RG2 Negligible

Simplified test circuits: at HV and LV windings FIGURE 3.4a2

12

b) Three-phase three-winding transformer ABB 345/161/13.2 kV, 248MVA transformer vintage 1977 was tested with the negative CHT power factor. Test results # Insul. kV mA Watts %PF Cap(pF) ------------------------------------------------1 CH + CHL 10 157.70 5.148 41848 2 CH 10 157.30 5.218 0.33 41736 3 CHT(UST) 10 0.417 -0.09 -0.22 110.5 4 CHT 0.400 -0.070 1.8 112 5 CT + CHT 10 191 7.048 50667 6 CT 10 190.5 7.071 0.37 50556 7 CHT(UST) 10 0.404 -0.010 0.3 107 8 CHT 0.500 -0.023 0.5 111 Analysis: This transformer has three windings intercalated in the same leg in the way that when the interwinding test between the outer and the inner windings is performed, the middle winding acts as a shield and provides a shunting path out of the meter as shown in the figure 3.4b. This situation results in the low or negative power factor.

HV

LV

CORE

Ter. HV winding CHT HV LV Ter. LV winding Tertiary winding

RG

Three-phase three-winding Transformer FIGURE 3.4b

13

c)

Single-phase three winding transformer

ABB 174kV, 132MVA transformer vintage 2001 with oil-water cooling system was tested in June 20, 2001 with negative power factor. Test results # Insul. kV mA Watts %PF Cap(pF) ------------------------------------------------1 CH + CHL 10 16.640 0.332 4413 2 CH 10 8.566 0.224 0.25 2272 3 CHL(UST) 10 8.073 0.105 0.13 2141 4 CHL 8.074 0.108 0.13 2141 5 CL + CLT 10 24.100 0.483 6394 6 CL 10 24.090 0.562 0.22 6391 7 CLT(UST) 10 0.013 -0.080 59 2.863 8 CLT 0.010 -0.079 76 3 9 CT + CHT 10 32.090 0.672 8512 10 CT 10 24 0.570 0.23 6368 11 CHT(UST) 10 8.068 0.110 0.14 2140 12 CHT 8.090 0.102 0.13 2144

Analysis: Normally this transformer has three windings (HV, LV and Tertiary) and two legs. The LV and Tertiary windings are physically in separate legs. The HV winding is split in two wounding around each leg in the way when CLT test is performed, the HV winding becomes the middle winding between the LV and tertiary windings as shown in figure 3.4c. This configuration results in the low watt loss or negative power factor in the CLT test. Single-phase three-winding Transformer FIGURE 3.4c

CORE

Tertiary
LV

HV CORE CORE

CLT LV HV Ter.

RG

14

d) Two-winding transformer with grounded shield between the windings McGraw Edison, 20.9/2.3kV, 1250kVA transformer vintage 1984 was tested October 3, 2001 with negative power factor: Test results # Insul. kV mA Watts %PF Cap(pF) ------------------------------------------------1 CH + CHL 10 13.040 0.477 3460 2 CH 10 7.854 0.752 0.88 2083 3 CHL(UST) 10 5.189 0.280 -0.5 1376 4 CHL 5.186 0.275 -0.5 1377 5 CL + CHL 10 16.850 2.024 4470 6 CL 10 11.670 2.322 1.83 3095 7 CHL(UST) 10 5.186 -0.290 -0.5 1375 Analysis: This transformer has two windings with a grounded shield between the two windings, as shown in the figure 3.4d. This arrangement results in a negative power factor as explained in basic theory. Because of the term (XC1 XC2 /RG ) is greater than the insulation watt loss. A power factor change will indicate the change of the shield ground condition (broken, burn away, etc.) or of the insulation watt loss. Generally the change tend to increase the power factor toward a positive value.

HV

Shld

CORE

LV HV winding Grounded Shield CHL HV Shld LV LV winding

RG

Three-phase three-winding Transformer Figure 3.4d

15

3.5. Stator Winding GE Generator, 3-phase, 2-pole, 306 MVA, 15.75 kV, 3000 RPM, 50 Hz, 0,85% PF, Hydrogen Pressurized cooling at 45 PSIG, vintage 2001, was tested on February 8, 2001 and the inter-phase test results for A phase were: Inter-phase test Results (CAB): Insulation kV mA 2.3 4.5 CAB 6.8 9.1 10.8

Watt 9.997 9.979 9.964 9.945 9.935 0.402 0.031 0 -0.09 -0.08

%PF 0.4 0.03 0 -0.09 -0.08

C(pF) 2650 2646 2642 2637 2634

TipUp

-0.49

Analysis: The inter-phase insulation of a stator consist mainly of insulation within the slot and outside the slot. Because the stator core is permanently grounded, the insulation within the slot has a permanent ground in the middle of its circuit, as shown in the figure 3.5a. This arrangement causes the negative watt value as discussed in the basic theory. The insulation within the slot is represented by CAg and CBg . The other insulation outside of the core can be the bracing between the coil ends, water hose (if water cooling) or the inter-phase insulating blocks, and 1 represented by C AB . This is truly ungrounded insulation and its watt loss is a positive value. Consequently the CAB current is equal to the sum of two paths as shown in the figure 3.5b. I CAB = I C AB + ICBg ICBg = VT XCAgXCBg + (XCAg + XCBg ) RG The inter-phase test result analysis is done by comparison between the test data (example: the current to to initial test data). For the good condition insulation, the power factor of inter-phase should be lower than the phase-to-grounds. It can be negative if the C1 AB is less important. But if C1 AB become more important such as water cooled stator, the power fator may stay positive. Because the water hose contribue more loss in the ungrouded insulation ( I C1 AB). Because the location of the ground point is unlikely to move, the change from the initial test result in power factor indicates the deterioration of the insulation, most likely in the insulation outside of the core. In some extreme case it can be caused by the deterioration of the insulation
1

C AB

Conductor

CAg

CBg

CORE Phase A Phase B CAB

Phase C
Typical stator winding FIGURE 3.5a
1

I C AB

CAg RG

CBg
I CBg

CORE
I C AB + I CBg
1

Test circuit of stator winding FIGURE 3.5b

16

Reference 1. Povey, E.H. Testing Bushings Equipped With Special Test Electrode, Minutes of the Fifteenth Annual Conference of Doble Clients, 1948, Sec. 3-101. Rickley, A.L. Low-Voltage Switch for Doble Type MH Test Set, Minutes of the Twenty-Fourth Annual Conference of Doble Clients, 1957, Sec. 3-201. Kopaczynski, D.J. and Manifase, S.J. Negative Power Factor of Doble Insulation Test Specimens (An Analysis), Minutes of the Fifty-Fourth Annual Conference of Doble Clients, 1987, Sec. 2-501. Cavigelli G.A., Kopaczynski D.J., Lachman M.F. and Levi R. AC Power Factor Versus DC Insulation Resistance Measurements, Minutes of the Sixty-Fifth Annual Conference of Doble Clients, 1998, Sec. 1-8.1 Ferreira, Marcos Suspect Lap type POC, 34.5kV bushing at Monroe Bank 1 phase A Doble Email Correspondence, sent on 14 January, 2002 12:09 PM

2.

3.

4.

5.

BIOGRAPHY Long Pong obtained his Bachelor of Electrical Engineering from the Ecole Polytechnique Montreal, Canada, in 1988. He currently holds the position of Principal Engineer with Doble Engineering Company in the Client Service Department. Before joining Doble in 2000, Mr. Pong had amassed 12 years of experience in electrical maintenance and project management for Generation, Transmission and Distribution apparatuses with ALCAN (Aluminum of Canada) and Hydro-Quebec. He has been a CEA (Canadian Electrical Association) member and a registered professional engineer in Quebec since 1988.

17