Analysis system for transformer oil gas test results (DGA) with email alerts, for power transformer care in factories

How the brief came about

The task in this case was to move the reading of transformer oil gas test results (DGA) from a separate Excel report for each transformer into a single web app. The customer's existing reports contained historical test result tables, triangle charts, trend charts, status and recommendations for the next oil sampling interval.

In transformer care generally, abnormalities inside a transformer cannot be seen from the outside, so oil test results are the main information used for maintenance decisions. When each transformer's test results sit in a separate file, getting an overview of all the transformers is difficult.

Requirements and constraints

The system had to do what the existing Excel reports could do, and had to sit entirely within the customer's network. The main requirements were as follows.

• Accept test results both by Excel file upload and by entering, adding, deleting and editing values on screen
• Display triangle charts, bar charts and trend charts with descriptions of the abnormality zones, in the same form the team reads in the existing reports
• Assess status and sampling interval from Condition and TDCG Rate criteria tables that users can edit
• Send email alerts when test results are abnormal
• Run on the customer's server in a closed network with no internet connection
• The customer specified PHP and a Microsoft SQL Server database

Why this system is a web app on the customer's server

This system is a web app on the customer's server because the customer required a web app installed on the customer's server in a closed network, using PHP and Microsoft SQL Server. Every test result is therefore stored in a database within the organisation. Users open the system through a web browser from a machine on the internal network and access it with a user account.

All users see status calculated from the same set of criteria tables in the system, and email alerts go to the configured recipients without waiting for anyone to open a file.

How the system assesses transformer status from DGA results

The system assesses status by comparing gas values with the Condition and TDCG Rate criteria tables, which can be edited in the system, then displays Normal / Warning / Alarm status with sampling interval recommendations and action steps. The system only displays results, makes recommendations and sends emails. Decisions on sampling or maintenance still rest with the transformer care team. The workflow is as follows.

1. 1

   Import oil gas test results

   Upload test reports from Excel files, or enter, add, delete and edit values on screen. The report header details for each transformer can be edited on both the add-data page and the report page.

2. 2

   Calculate gas proportions and draw charts

   The system calculates the proportion of each gas, then plots a point on a triangle chart with a choice of 3 types, with text stating which abnormality zone the point falls in. The system displays a bar chart of the proportions of 6 key gases and a line chart of gas value trends by date, in which each gas can be shown or hidden individually.

3. 3

   Assess status against the criteria tables

   The system compares gas values with the criteria table (Condition). If a result exceeds the normal criteria, the system compares the total combustible gas value (TDCG: Total Dissolved Combustible Gas) and its rate of increase (TDCG Rate) with another table, to display the status, a recommendation for the next sampling interval and the action steps.

4. 4

   Send alerts and produce reports

   When the status is Warning or Alarm, the system sends an email immediately, and sends each recipient a summary email listing items according to the selected data types. Users can produce reports as PDF files from the report page.

Deliverables

We delivered a DGA analysis web app installed on the customer's server, together with system design documents and user manuals. In addition to the test result import page and report page in the steps above, the system has the following sections for administrators.

• Criteria settings page: edit the Condition criteria table and the TDCG Rate table; changes must be saved each time
• Email page: add recipients, turn sending on or off, and choose whether each recipient receives Normal, Warning or Alarm status items
• Access: log in with a user account through a web browser on the internal network

Confirmed results from acceptance testing and delivery

We delivered all items of the DGA analysis web app, with 3 types of triangle chart and a bar chart of 6 key gases, in 2019. Acceptance testing in April 2019 found that the scale of the triangle chart image did not match the original, and that when gas values were entered incompletely, the point on the chart fell outside the triangle, so users asked for points outside the chart to be hidden. We fixed both issues and retested before delivery.

1. 2018: project planning and requirements gathering began
2. April 2019: system acceptance testing, and recording of items that needed fixing
3. 2019: fixes, retesting and delivery of all items, together with system design documents and user manuals

This case has no measured business figures and no post-delivery usage report, so we report only the results backed by documents.

What we learned

This work offers 3 principles for systems that read test results and raise alerts against criteria.

• Test with incomplete data before acceptance testing. Real test results may be missing some values, so test data sets should include both complete and missing values, and how the system must display results when data is incomplete should be defined in advance.
• The charts a team uses to read results must match the originals the team is familiar with. Use original chart images with the correct scale, and confirm the screen designs with users before development starts.
• Criteria and alert recipients should be information the team that owns the work can edit itself. In this work, the criteria tables can be edited on the settings page and must be saved each time, while email recipients can be added and data types chosen per person.

How you can apply this approach to your transformer care work

If you have several transformers and already send oil for DGA testing at regular intervals, you can use the same approach as this case: bring test results into one database, compare them with the criteria tables your team uses, and send alerts by email. The parts that need adapting are the test result file format, the criteria your organisation uses and the IT systems that must be connected.

• Prepare sample test reports from the laboratory, both with complete values and with missing values
• Identify the criteria tables your organisation uses as a reference, and who is authorised to edit the criteria
• Check whether you have a server, a database and an internal email server for a system that runs on a closed network
• Agree clearly who receives email alerts and who decides the next step when the status is Warning or Alarm