Unlocking the power of analytics with the right data in the right place
As the Internet of Things becomes a reality, billions of connected devices are generating massive amounts of data.
In fact, by 2020, the number of internet-connected things is estimated to reach 50 billion. All this data has a lot of potential for utilities. However, massive amounts of data are meaningless if utilities are unable or unwilling to sift through it all to analyse and use it.
For utilities who want to optimise their operations and enhance their revenue, it’s not about big data and crunching millions of data points; it’s about leveraging the right data in the right place to achieve the desired outcome. This is where analytics come into play.
Analytics are essential to helping utilities understand their distribution grids and optimise their investments and quality of service. Utilities can leverage the power of data to reduce operational costs, improve efficiency and reliability, increase customer satisfaction and satisfy regulatory requirements – all by allowing analytics and corrective actions to take place where the best outcome occurs. The key is enabling operational data to be analysed where it will create the most valuable result for a specific utility issue or need. Sometimes this is in the back office, sometimes at the network level and sometimes at the local device level. Operational issues that need to be analysed, such as voltage fluctuations, hot spots, impedances or sources of outages, are examples of how and where analytics can play a role in resolution and prevention.
The right data in the right place
Data typically flows from end devices through a network to be stored, processed and analysed in a traditional utility environment. Big data refers to the tsunami of data that can result from using only this approach and solely pulling data into the back office to be analysed.
The way to address this big data dilemma is with the right data in the right place.
In some instances, an analysis should take place within the endpoint, like the meter or sensor, where it makes more sense to take immediate, local action. This is called distributed analytics, and this type of analysis is complementary – not counter – to back-office analytics. Analysis occurs continuously at the edge of the network, sending insights rather than droves of data to the back office. Intelligent devices communicate and collaborate directly with each other and make decisions in real time.
In some instances, computing at the edge is the best place for analysis to take place, while other types of analysis are better conducted in the back office. The following guidelines can help determine the best place for analysis:
• The type of data needed for the problem or situation.
Is the problem or issue a system-wide issue, a network level issue or a device level issue? Does having insight and SCADA-like awareness into the low voltage network enhance the ability to make a better decision to correct it? If the answer is yes, then distributed intelligence at the device level can be used to analyse the data and take local action for these types of scenarios.
• The accuracy and granularity of the data needed.
In some cases, high-resolution, granular, one-second data is the best data to address a problem. This type of data is only available at the local device level due to latency and data volume constraints. As such, distributed analytics are necessary. Action and decisions using the most current and most accurate data will produce better outcomes, such as faster response times, which translate into customer and operational savings. On the other hand, if only hourly data is needed to adequately solve an operational issue, back-office analytics could be a better option.
• The response time for the action needed.
If action needs to be taken quickly, distributed analytics makes the most sense. Whereas issues that are less time sensitive, like transformer connectivity, can be handled in the back office.
Back-office analytics are typically provided in a software-as-a-service format via a public or private cloud service so they can be accessed anywhere by utility staff. Examples of analytic applications that are best performed in the back office include load forecasting, reliability reporting and grid connectivity.
• Load forecasting aggregates AMI data to ensure tighter operations, reduce uncertainty and instil confidence that operating and investment decisions are based on a solid understanding of current business fundamentals, which strengthen both forecasting and backward-looking analysis.
• Reliability analysis allows utilities to improve overall system reliability by conducting a detailed analysis of all outage data generated from a smart grid network over a period.
• Grid connectivity service helps utilities quickly and accurately identify meter to transformer and meter-to-phase connectivity using hourly interval voltage data and sophisticated machine learning algorithms, without the need for expensive hardware or field labour. This can be done surgically for a specific group of equipment or system wide.Read full article at Metering and Smart Energy