1. Dayashankar Dubey (MTech)
Online Monitoring of
Dissipation Factor
Dayashankar Dubey (MTech)
Suhas P. Solanki, MTech
Guide: Prof PC Pandey
EE Dept, IIT Bombay

2. ABSTRACT
The insulation status in HV equipment can be monitored by dissipation factor measurement. Online monitoring of dissipation factor is based on dividing the actual power by apparent power. Sampling rate lower than the power line frequency results in aliased periodic waveforms which retain the original phase relationship, and these waveforms can be processed at a relatively low computational speed. Numerical simulation has been carried out to study the effect of quantization error with different number of bits, for finding the effect of different filters used in processing, effect of harmonics, and variation in power line frequency.

3. INTRODUCTION Lossy Capacitor Dissipation factor of lossy dielectric
Loss angle = , dissipation factor = tan
Parallel model Series model

4. Monitoring of dissipation factor
▪ Insulation deterioration → increase in 
▪ Monitoring of  or dissipation factor → safe & reliable operation of HV equipment
Online monitoring of dissipation factor
▪ HV equipment need not be removed from service
▪ Insulation deterioration between scheduled checks gets detected
▪ Monitoring under actual load & temperature conditions

5. METHODOLOGY
Signal Acquisition

6. METHODOLOGY(contd..)
For small , cos δ ≈ 1, s9 = sin δ ≈ tan δ

7. ERROR ANALYSIS Assumption for theoretical analysis:
Error caused by quantization noise,
Harmonics totally eliminated by LPF
RMS error in s9:
σ =2-L√(8γ/3)
where L = no. of quantization bits
& normalized cutoff frequency γ = fc/fs
For dissipation factor: 5 – 5010-3 & 1% resolution: σ = 50 10-6 L 8 12 γ
6.1410-5
1.57 10-2

8. Project objective: Verification of the technique
IMPLEMENTATION
Project objective: Verification of the technique
Through,
Numerical Simulation
Experimental setups
For two implementations
High Sampling rate for fast updates
Low sampling rate: low cost instrumentation for low update rate
(based on aliasing of periodic waveforms)

9. IMPLEMENTATION (contd..)
High sampling rate: Numerical simulation
fs = 450 sa/s,: , γ=10-3 (for 8-bit), γ=15.53*10-4
(for 12-bit) (without band pass filter at the input)
L bits
Filter
Freq (Hz) σ
(simulation)
(theoretical) 8
IIR Butterworth 50
1e-10 to 1e-4
2e-4
IIR Chebychev-I
1e-7 to 4e-4
IIR Chebychev-II
1e-6 to 3.3e-4 12
2e-5 to 7e-5
4.9724e-5
8e-7 to 1e-5
7e-5 to 1e-4

10. IMPLEMENTATION (contd..)
High sampling rate: Experimental setups
Low voltage setup (30 Vpp)
I/V converter & res. V-divider
Acquisition with 8-bit 2-channel DSO, 5 k record length
LPF: 1 k rect. FIR filter
For D.F. of  10-3,
Best fit line: slope = 1.044, offset =  10-4
High voltage setup (600 V)
Cap. divider for V & shunt resistor for I sensing
Acquisition with 8-bit 2-channel DSO, 50 k record length
LPF: 10 k rect. FIR filter
For D.F. of  10-3
Best fit line: slope = 1.051, offset = 5.7  10-4

11. IMPLEMENTATION (contd..)
Low sampling rate
Sampling with fs< fo: aliasing of periodic V and I signals with frequency f = fo- fs, retaining the original phase relationship
Advantages
Low cost data acquisition system
Distributed signal acquisition units can transmit data over a serial link to central unit for processing

12. IMPLEMENTATION (contd..)
Low sampling rate: Numerical simulation
fs = 45 sa/s, : , γ=10-4 (for 8-bit), γ=2.48*10-4
(for 12-bit) (without band pass filter at the input)
L bits
Filter
Freq (Hz) σ
(simulation)
(theoretical) 8
IIR Butterworth 50
5e-3 to 49e-3
0.6e-4
IIR Chebychev-I
1e-4 to 49e-3
6e-5
IIR Chebychev-II
9e-5 to 1e-4
1e-4 12
1e-5 to 4e-5
2e-5
8e-7 to 1e-5
1e-5

13. IMPLEMENTATION (contd..)
Low sampling rate: Signal acquisition card

14. IMPLEMENTATION (contd..)
Low sampling rate: Experimental setup

15. EFFECT OF FREQUENCY VARIATION
High sampling rate: Numerical simulation
fs = 450 sa/s,: , L=12bit, γ=15.53*10-3
(without band pass filter at the input)
Filter
Freq (Hz) σ
(theoretical)
(simulation)
IIR Chebychev-I 48
5e-5
6e-6 to 5e-4 50
6e-5 to 1e-4 52
5.8e-5 to 6.1e-5
IIR Chebychev-II
3e-5 to 5e-5
7e-5 to 1e-4
5e-4 to 5e-2

16. EFFECT OF FREQUENCY VARIATION (contd..)
Low sampling rate: Numerical simulation
fs = 45 sa/s, : , γ=10-5,L= 12-bit
(without band pass filter at the input)
Filter
Freq (Hz) σ
(theoretical)
(simulation)
IIR Chebychev-I 48
1e-5
4e-6 to 9e-4 50
9e-5 to 1e-4 52
3e-5 to 5e-5
IIR Chebychev-II
5e-5
3e-5 to 8e-5
2e-5 to 5e-5

17. EFFECT OF HARMONICS THD in power line is around 5%
Harmonics filtering through
Band pass filter (BPF) (Using IIR Chebychev-I)

18. EFFECT OF HARMONICS (contd..)
High sampling rate: Numerical simulation
fs = 450 sa/s, : , L=12bit, γ=15*10-3, σ = 5e-5
Freq
With BPF σ
Without BPF σ 48
1e-4 to 6e-5
5e-5 to 6e-5 50
1e-4 to 9e-5
9e-3 to 9e-5
Low sampling rate: Numerical simulation
fs = 45 sa/s, : , L= 12-bit, γ=10-5, σ = 1e-5
Freq
With BPF σ
Without BPF σ 48
1e-4 to 6e-5
5e-5 to 6e-5 50
1e-4 to 9e-5
9e-3 to 9e-5

19. SUMMARY AND CONCLUSIONS
Direct calculation algorithm for dissipation factor m/s verified for range with resolution 1% (i.e.510-5)
Implementation using high and low sampling rates
IIR Chebychev filters for m/s insensitivity to power line drift
Band pass filter for attenuating harmonics of power line freq.
High sampling rate implementation
For detecting incipient faults during tests/charging of HV equipment
Instrumentation for m/s with high update rate (~10 s):
DSP with two 12-bit simultaneous sampling ADCs
Low sampling rate implementation
For monitoring of HV equipment under normal aging process
Instrumentation for m/s with low update rate (~10 min.):
signal acquisition h/w with serial data link to central unit for pro.