Advancements in Dissolved Gas Analysis: Data Quality

By Dr. Zachary H Draper & Dr. James J Dukarm

There is more to DGA interpretation than comparing the latest gas concentrations to limits in a table or plotting them in a triangle or pentagon to identify the apparent fault type. We have found that the whole DGA history of a transformer must be considered when interpreting its most recent DGA results.
Trend evaluation and accurate assessment of short-term changes require accuracy and low measurement variability of gas data. Data quality problems must be recognized and dealt with before an interpretation is attempted. Below we point out some of the most common data quality issues.

Data management
As a result of the historical importance of DGA data, proper organization and preservation of DGA data are extremely important. In addition to archiving the lab reports, keep the data in tabular form in a database or, for small volumes of data, a spreadsheet. A well-organized database supports sorting and filtering for graphical and statistical analysis.
Use a unique and permanent ID to identify transformers, oil compartments, and the oil sample data belonging to them. Substation and unit number are not a suitable ID, for the same reason that the dentist doesn’t identify you by your department and job title. Large transformer fleets may require company-assigned asset numbers to avoid possible serial number duplication across manufacturers.
Disciplined chain-of-custody practices provide correct IDs of transformers and compartments to be sampled, ensure that oil samples are labeled correctly, and guarantee that analysis results returned by the lab are attributed to the right transformers and oil compartments.

Read full article in the Special POWER SYSTEM RELIABILITY Issue 2020

SOURCEElectricity Today Magazine, Vol.33, No.4
Previous articleCommunications, Data Hardening, and the Smart Grid
Next articleEFACEC case study
Dr. Zachary Draper is a scientist with Delta-X Research Inc.  His current research focuses on reliability statistics, multi-dimensional gas diagnostics for transformers and load tap changers, and seasonal variability and signal processing for online DGA monitors. He also provides training on how to interpret advanced analytics and reliability-based dissolved gas analysis. A member of both IEEE and CIGRE, Dr. Draper received a B.Sc. in Physics and Astronomy from the University of Washington, and a Ph.D. in Astrophysics from the University of Victoria.