Digital Utility: CESC uses data analytics for efficient network planning

CESC uses data analytics for efficient network planning

In the past couple of years, data analytics has emerged as one of the key focus areas of discoms. It is aimed at improving their operational performance and enhance consumer satisfaction. Utilities are increasingly undertaking data analytics to predict load growth for efficient network planning, and assess the health of the network for predictive maintenance. Further, data analytics plays a key role in revenue protection and theft detection. By analysing consumption data, utilities can identify the anomalies in energy consumption.

Data analytics in distribution

Data analytics involves interpreting data elements and communicating their meaning to the relevant stakeholders for deriving actionable insights. It leverages the volume, variety, velocity, variability and veracity of big data to deliver value. Big data refers to very large and complex data sets that require analysis to provide insights for modelling, monitoring and controlling purposes.

In a distribution utility, multi-variate data is analysed to derive benefits. Data pertaining to distribution transformer loading (three-phase current, voltage, active and reactive power data is available for 30-minute intervals), distribution station and feeder loading (current, voltage and power data is available for 15-minute intervals), consumer meter reading (one tab-based reading per month is available in electronic meters whereas 15-minute interval reading of V, amp, kW and kWh is available in smart meters), fault and fusing is analysed. In addition, utilities analyse data from call centres and complaint management systems (including the reason, type and resolution timeline of complaints), as well as data pertaining to bill payment and bill payment mode (online versus offline). It helps improve customer service, revenue protection, cost efficiency and manpower utilisation.

CESC’s initiatives

CESC has formed a cross-functional team with both information technology and domain expertise to perform data digital analytics functions. Any exception in performance is reported for course correction purposes. The discom also undertakes predictive analytics in order to maintain the health of the distribution system. Further, feedback is used to fine-tune the prediction model, that is, enhance the precision and the recall level.

In order to identify the key network elements for capex planning, the discom analyses a host of performance parameters. One of the parameters is network loading, wherein loading data for 1,400 feeders and 8,000 transformers is analysed to identify the areas that require intervention. Potential load growth in terms of consumption growth and new consumers is estimated based on past trends, region-wise load patterns and new load applications. Some of the other factors that are analysed are network performance in terms of fault frequency, that is, feeder/transformer failure frequency data; customer sensitivity; and yearly capex budget provisioning.

In order to reduce commercial losses, CESC has prepared a priority list for undertaking meter replacement based on year of purchase, meter make/model, sales increment history after meter exchange, average consumption, and budget sanctioned for meter exchange. Three to four years back, around 50 per cent (around 1.5 million) of consumer meters were electromechanical while 50 per cent were electronic meters. Now, many of the electromechanical meters have been replaced by electronic meters. At present, there are around 2.1 million electronic meters, while only 1 million are electromechanical.

The utility has deployed data analytics for the detection of non-technical losses to ensure revenue protection. The company analyses the data of 3 million consumers, the pilferage history of suspect consumers, the loading history of distribution transformers in theft-prone pockets and the creditworthiness of customers. While the loading history of the transformer in a pilferage-prone area is analysed to identify phase imbalance and undertake load curve pattern recognition, the creditworthiness of customers is assessed based on instances of delays in bill payment. The company has undertaken the segmentation of pilferers based on the neighbourhood, tariff category, activity type and meter type. With a thorough analysis of these parameters, suspect neighbourhoods/consumer groups can be identified for undertaking inspection drive.

Further, using data analytics, CESC predicts faults in the network in order to reduce operations and maintenance (O&M) cost, and improve system reliability. The company analyses historical data of distribution transformers and high tension (HT) cables (frequency, type of failure, root cause analysis of faults), loading history/pattern of HT cables/distribution transformers (aggregated hours of overloading and peak overloading), ambient condition (temperature, rainfall and humidity) and asset history (age, manufacturer/model, quality of installation, maintenance history) to predict a probable fault scenario.

Besides, CESC is undertaking fault reduction and proactive health monitoring through big data analytics. It uses analytics tools to redistribute the load and reduce overloading of equipment. Further, it is undertaking condition monitoring using non-invasive online technologies such as an online partial discharge monitor, a remote temperature monitor for distribution transformers and infrared thermography. The company is also deploying smart meters for better energy management. These meters help consumers track their daily energy consumption and plan their energy cost. Further, automated meter reading and the integration of meters with the new billing system has resulted in zero non-actual billing. Smart metering also helps in integrating renewable energy sources into the grid with net metering options. Another area where the company has deployed analytics is proactive compliance management. Through data analytics, the company identifies prospective areas of complaint and sends proactive communication to consumers regarding this. Besides, the discom addresses consumer complaints through multiple channels working round the clock such as call centres, multi-tasking counters, social media and website, and the managing director’s chat corner. It is also using data analytics for smart outage management. In case of an outage, the smart meter directly logs a supply complaint on to the complaint management system and sends an SMS to the consumer with an expected restoration time. Further, a GPS-enabled fleet is sent for power restoration. It reaches the consumer faster with the help of navigation.

The way forward

Going forward, data analytics will be one of the key trends driving the digital utilities of the future. The use of artificial intelligence (AI) and machine learning (ML) is creating value by enhancing operational efficiency. The utilities are fast adopting AI and ML solutions for HT fault prediction, reduction in outage duration and optimisation of O&M costs. Further, internet of things and analytics-based solutions are being deployed by utilities for predictive asset maintenance. Meanwhile, for better consumer complaint management, utilities are resorting to automating call centre operations using voicebots and resolving customer queries using chatbots.

To conclude, it is essential for the distribution utilities to undertake data analytics in order to assess their operational performance and identify anomalies, if any. Further, the distribution utilities must take steps to plug the gaps in network performance.

Based on a presentation by Kirit Rana, Deputy General Manager, CESC, at a recent Power Line conference