Revision C57.104 – TF on Data Analysis

Report
IEEE C57.104 Revision (Nov 2011)
TF Database Analysis
Luiz Cheim and Lan Lin
ABB Inc.
Luiz Cheim, Lan Lin - IEEE Fall Meeting - Milwaukee 10/23/2012
Building the Database
+ HQ
= 521,700 records
Transformers Age Groups
Transformers Power Ratings
Transformers Voltage Classes
Transformers Oil Preservation Type
Reason for DGA
Percentiles of Hydrocarbons (all data)
ppm
%
Percentiles of CO,CO2 and TDCG (all data)
ppm
%
Percentiles of H2, C2H2 and CO vs. Age Groups
ppm
%
Percentiles of H2, C2H2 and CO vs. Voltage Class
ppm
%
Percentiles of H2, C2H2 and CO vs. Power Ratings
ppm
%
Percentiles of H2, C2H2 and CO vs. Source of Data
ppm
%
Percentiles of All Gases (Suspicious Units)
ppm
%
Percentiles of H2, C2H2 and CO vs. Power Ratings
(Suspicious Units)
ppm
%
Percentiles of H2, C2H2 and CO vs. Voltage Class
(Suspicious Units)
ppm
%
Percentiles of H2, C2H2 and CO vs. Oil Preservation
ppm
%
Annual Rate of Gas Formation (ppm/year)
Percentiles of All Gases
ppm
%
After that…
1. HQ Data incorporated
2. Discussed unknown oil preservation system
3. Question of Rate of Gas Formation vs Levels
Montreal Meeting (May 2o12)
Revisited Percentiles with Following Condition:
O2/ (N2 + O2) < 5% and N2 > 70,000 => N2 Blanket
5 – 15% Membrane/Rubber Bag
> 15%
Open Breather
Legend (next slides)
IEEE
Cigre
AFMo
AFSi
Condition 1 Limit
Typical Value (Brochure SCD1.32, 2010)
Arc Furnace, 90th Percentile (mineral oil)
Arc Furnace, 90th Percentile (Silicon oil)
H2 – All Data
AFSi
IEEE Cigre
AFMo
CH4 – All Data
IEEE
AFSi
Cigre
AFMo
C2H2 – All Data
Cigre AFMo
IEEE
AFSi
C2H4 – All Data
Cigre
AFMo
IEEE
AFSi
C2H6 – All Data
Cigre
IEEE
AFMo
AFSi
AFSi = 1936
CO – All Data

Cigre AFMo
IEEE
AFSi = 25,387
CO2 – All Data

Cigre
AFMo
IEEE
TDCG – All Data
AFSi = 2,160

Cigre
IEEE
AFMo
Further Breakdown…
Given a type of oil preservation
(say, N2 Blanket)
What happens to
Gas Levels vs Voltage Class, MVA, Age?
H2, 90th Percentile – Per Voltage Class
AFSi
IEEE Cigre
AFMo
H2, 90
th
Percentile – Per MVA Rating
AFSi
Cigre
IEEE
AFMo
H2, 90
th
Percentile – Per Age Group
AFSi
Cigre
IEEE
AFMo
C2H6, 90th Percentile – Per Voltage Class
Cigre
Revision C57.104 – TF on Data Analysis
Gas Rate Subject
1. Should we calculate rates percentiles for all gases?
2. Should we associate rates to levels? How?
3. Cigre recent experience
Revision C57.104 – TF on Data Analysis
Cigre Levels vs Sampling Intervals
Cigre SCD1.32 Brochure, December 2010
Revision C57.104 – TF on Data Analysis
Cigre Rates vs Sampling Intervals
Cigre SCD1.32 Brochure, December 2010
Revision C57.104 – TF on Data Analysis
Cigre SCD1.32 Brochure December 2010
Revision C57.104 – TF on Data Analysis
Proposal Under Discussion – Level vs Rate
Frequency of DGA
Revision C57.104 – TF on Data Analysis
Proposal Under Discussion – Level vs Rate
Severity of Fault (?)
Revision C57.104 – TF on Data Analysis
Gas Rates (IEEE Database)
 ppm
t
Revision C57.104 – TF on Data Analysis
Gas Rates (IEEE Database)
Revision C57.104 – TF on Data Analysis
Gas Rate Issues
 ppm
t

115  72
 0 . 323 ppm / day
133
115
72
133 days
Revision C57.104 – TF on Data Analysis
Gas Rate Issues
 ppm
t

115  72
 118 ppm / year
133
115
72
133 days
Revision C57.104 – TF on Data Analysis
Gas Rate Issues  ± 15% Cigre Typical Lab Accuracy
 ppm
t

98  83
 41 ppm / year
133
133 days
Revision C57.104 – TF on Data Analysis
Gas Rate Issues
 ppm
t

68  115
  50 ppm / year
343
115
?
72
68
343 days
Revision C57.104 – TF on Data Analysis
Gas Rate Issues
 ppm
t
 13 . 5 ppm / year
Revision C57.104 – TF on Data Analysis
Another Example
Revision C57.104 – TF on Data Analysis
Another Example
500 ppm/year
27 ppm/year
Revision C57.104 – TF on Data Analysis
Inherent Uncertainty in Trend Calculation
1. Multiple Laboratories
2. Lab Variations (repeatability, reproducibility, accuracy)
3. Seasonal variations (temperature, etc)
4. Sampling errors (humans, sample contamination, etc)
5. Type of Transformer (N2, etc…)
6. Sampling interval??
Revision C57.104 – TF on Data Analysis
Distribution of DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Distribution of DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Distribution of DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
Gas Rates Distribution vs DGA Sampling Interval
38%
Revision C57.104 – TF on Data Analysis
Rates Distribution vs DGA Sampling Interval
Revision C57.104 – TF on Data Analysis
204
53
Revision C57.104 – TF on Data Analysis
Revision C57.104 – TF on Data Analysis
Revision C57.104 – TF on Data Analysis
Revision C57.104 – TF on Data Analysis
Revision C57.104 – TF on Data Analysis
Food for Thought!
Revision C57.104 – TF on Data Analysis
Proposal for Discussion – Level vs Rate
Severity of Fault (N2 Blanket, etc)
90th | t
95th | t
…
Table conditioned to t = Sampling Interval
Revision C57.104 – TF on Data Analysis
Rate of Gas Increase vs Sampling Interval???
Revision C57.104 – TF on Data Analysis
Rate of Gas Increase vs Sampling Interval???
Revision C57.104 – TF on Data Analysis
Rate of Gas Increase vs Sampling Interval???
Food for Thought…

similar documents