Mining the Corporate Data Domain Safely and Securely Across the Enterprise

Part 5: EMS Data Mining Success

Enabling Success

The storage, retrieval and audit capabilities of the system are field proven with over three years of on-line service. The ability to pull massive sets of reliable, complete data from the system is being increasingly recognized and used by the planning and maintenance organizations at Entergy. Some of the successes include:

• Automated audit tracking of the SCADA database changes
• Reducing the need for SCADA support personnel to be involved when T&D planning, maintenance or management organizations request archived SCADA data across a wide time range. Compared with the legacy system, which could only piece together data on a day-by-day basis, a reliable and flexible interface to data on daily, monthly or longer time frames is now readily available.
• Reliable and easy access to data has allowed various departments to:
  • Start fine tuning alarm levels,
  • Determine which transducers are out of calibration,
  • Obtain data to support load balancing decisions that have a direct effect on reliability,
  • More effectively support the system design team,
  • And, identify “chattering” alarms to speed up repairs.

Next Steps

At this point development is focusing on data mining applications that take advantage of the tight integration of the system with Entergy’s SCADA system to provide reports and automated analysis of how well the T&D system is performing. Examples of data mining reports already developed and going into production in Q1-2008 include:

• “Sticky” circuit breakers (i.e., breakers not responding properly to operator commands). One click on this report brings up the list of breakers with problems, and selecting any of these brings up the OPEN/CLOSE history of that device for the month, showing each status change and any related operator command. [Refer to Figure 3]
• Devices changing state most often, sorted by frequency. This allows a quick drill down to see which field points require maintenance.
• Alarm limits most frequently violated, also sorted by frequency. This report can be used to check that alarm limits are appropriate, or quickly see which devices are most often operating beyond preset limits.

Figure 3: The system provides numerous data mining reports. This example shows a drill down (detail) from the ‘No Response’ report, flagging a problem by showing that Device J0434 required multiple close commands before the breaker operation was completed.

These reports also allow users to drag and drop columns so they can organize the data any way they want; for example: by substation; by device; by date; by area of responsibility; or any other column.

Additional sophisticated reporting providing mathematical analysis of analog values is under development. These reports include:

• Analogs that are mathematically erratic using statistical methods
• How long individual devices such as transformers operate beyond normal and emergency limits, and by how much

Further analysis based on pattern recognition, applied to the power system data store, will be used to look for patterns, statistical advantages, and inference between variables. This area provides an exciting chance to break new ground in real-time analysis of the Entergy power system in the coming years.

The solution provided to Entergy is a unique, very low maintenance, industrial quality platform that enables:

• Ready access to a vast storage of historical data and events
• Forensic analysis of major electrical grid events
• Preventive maintenance insights and benefits
• High-integrity automated auditing

Being able to specify real world requirements, in addition to functional requirements, and getting a solution designed and implemented around those requirements has made this project a success for Entergy, and it’s done this while helping to alleviate Entergy’s resource constraints.