by Nicholas Abi-Samra, Claude Godin and Curt D. Puckett, DNV GL
Big data analytics promises to improve grid performance and customer engagement, yet securing and protecting the immense data generated from monitoring, control and transactional energy operations requires new ways of doing business and revamped analytical solutions. But don’t confuse the hype with your organization’s capability to implement solutions and realize a positive return on investment (ROI). If you do, you will end up floundering or making piecemeal progress at best.
It is important to understand and align your capabilities with your information and operations technology functions. Creating a comprehensive road map focused on high-return predictive analytics with clear destinations and achievable milestones is the starting point for better understanding customers, their behavior and their impact on grid operations.
The Enticement of Tomorrow
Much excitement surrounds the benefits to utilities from manipulating data collected from the grid and customers, such as new operational efficiencies, new products, novel ways to serve, and fresh revenue streams. For executives and engineers, many of the operational benefits are easy to define and evaluate; improved data analytics lead to improved outage management, voltage optimization and predictive asset management, which all lead to reduced truck rolls and other quantifiable results. Customer benefits, however, where the greatest ROI might come from revenue protection, improved load forecasting and detailed customer segmentation leading to increased demand response program enrollments, are enticing but harder to quantify.
On the operational side, utilities often face little reinvention. The analytical processes and algorithms for data have been in place for some time, but now the information is better and available in real or near-real time, so better decisions can be made more quickly.
On the customer behavior side, it’s a bit fuzzier. Customers are harder to predict. Although there is a lot of buildup in the marketplace, most utilities have a long way to go before realizing the data analytics dream the way, say, Google does. Now it’s more like a nightmare: seeing enormous dumps of data but not knowing what to do with them.
In many instances, the operational side drives the utility business case for data analytics investment, with customer analytics’ providing additional, smaller benefits and supplemental ROI. Some utilities embrace the analytics as they look for ways to engage customers in better grid management. Many believe they will get something from all that customer data, but they aren’t sure what-perhaps improved customer engagement strategies or new ways to increase customer satisfaction. A more compelling question is how much customer engagement do utilities want? On the level of Google or a retail store? Probably not. Some regulated utilities are risk-averse and others shy away from the customer side of the meter because of increased liabilities and negative press if results don’t materialize as envisioned. Conversely, public power utilities, which generally have closer customer-member owner relationships, use data analytics for more engagement.
The Systems Problem
To apply analytics effectively, a utility’s information technology (IT) and operations technology (OT) functions require integration. The interdependency of those groups is central to the utility’s ability to maintain power quality, security and reliability and to achieve acceptable ROI on smart grid investments. Not aligning the two functions might result in inefficiencies and a negative impact on ROI.
Historically, utility OT and IT organizations have existed as separate groups with different structures, goals and mandates. Power systems engineers who deal with the grid’s physical assets traditionally have run OT systems. In addition, many of the transmission and distribution engineering applications often are proprietary and stand-alone.
IT systems have been the realm of computer and software personnel. Further, the current utility IT architecture still serves the basic meter-to-cash model. As such, it is a meter-centric system in which the billing system wants the data from the meter, and customers are not involved, except for billing purposes and questions or complaints. Most utility IT architectures have grown incrementally with point-to-point interfaces. The unfortunate drawbacks of that are information bottlenecks that restrict the free flow of data and access to information within the enterprise.
When IT created its systems, the operations group, as well as the marketing and strategy teams, were kept in abeyance. Now those teams want to use the increased knowledge behind the data but they often can’t access the system directly because of operational constraints such as design limitations to the utility’s IT architecture and its associated capacity. In addition, utilities are overwhelmed by the new automated meter information produced by the system. Most of the data go into a data store where they sit or are not used effectively. Although plenty of groups might want access to the data, the question is becoming whether any privacy restrictions on access exist.
Integrating and aligning IT and OT can be synergetic and lead to increasing service reliability and improved customer experiences. One of the greatest needs for closer collaboration is in data storage and management. With smart analytics on the IT side, the utility can mine this data to provide actionable information. The data may fall into key IT categories for use by OT (for the operation of the network) or to serve enterprise needs. Even incremental improvements with new software versions or technology can lead to organizational changes-even step changes-if IT and OT are aligned.
Building the Road Map
Each utility has a different value proposition for grid and customer analytics, depending on the IT structure, age and supporting systems. Some systems are fast, some slow; some are easily scaled while others can’t be scaled.
Utilities climb through four stages of analytic readiness and experience (see Figure 1): nascent analytics, siloed analytics, analytics orientation, and analytically enabled high performance. Utilities at the first stage make limited use of analytics in operational and customer-related decision-making. Utilities in the second use analytics in a limited way: in large transformer maintenance management, for example. Stage-three utilities have begun to look for a more integrated processes, metrics and ways to improve. At the fourth stage, utilities have adopted an enterprisewide analytics perspective.
Market studies show the ROI for data analytics is anywhere from two to 12 times the investment if you implement everything that could be implemented. It probably is impossible to put every function into action. Each utility’s stage of analytics readiness, maturity level and systems dictate the outcome. Executives might want the information; IT might want to control it. IT is good at the how, whereas the business team is good at the what.
To cut through this, the utility needs a clear road map of how to satisfy the business team’s needs using the data controlled by IT. The necessary changes cannot be made at once, so see what makes sense at a stage level, depending on the maturity of the networks and systems.
Beyond the Hype
Whether regarding grid or customer analytics, the highest returns for a utility come from predictive analytics (see Figure 2). A road map for predictive analytics starts with segmenting each opportunity (16 are listed here) into functional components a utility can evaluate on their own merit, taking into account the impacts to current business processes and IT systems (see Figure 3). Then, the utility can group the opportunities into feasible sets based on ROI, ease of implementation, impact on business processes, and investments in IT systems. The utility can assemble these sets and associated investments into a comprehensive road map that lays the opportunities along short-, mid- and long-term horizons for funding and implementation.
Beyond the Hype
Transformer load management (TLM). TLM can create and store virtual transformer load profiles using advanced metering infrastructure (AMI) load data with meter-to-transformer relationship data. You also can capture other information, such as direct transformer load measurements, oil temperature and hourly weather data. With this information, you can evaluate transformer loadings continuously against nameplate ratings and initiate alarms at predetermined thresholds. A TLM application also can generate targeted demand response actions, optimize transformer sizing, as well as monitor maintenance and replacement programs.
Augmented distribution management system (DMS) in near-real time. Getting real-time meter and sensor data from renewable assets to augment DMS enables utilities to enhance performance in such areas as conservation voltage regulation, dynamic voltage regulation, load balancing, flow analysis and back-feed information. A utility must develop or acquire dynamic load flow tools that can capture real- or near-real time data from AMI systems.
Asset use for predictive line, asset maintenance. This function stores, processes and trends momentary outage event data and provides a report or alarm system that identifies customers who experience a high number of momentary outages attributed to vegetation or maintenance issues. The utility could display the information in a geographical information system (GIS) to identify potential problems in specific areas. In addition, the utility could warehouse AMI, supervisory control and data acquisition and distribution automation information and use it to drive advanced asset management.
Near real-time renewable and microgrid monitoring. This leverages the AMI data received from renewables or microgrids to enable the system operator (either directly or through a DMS) to monitor load activity, status and conditions. This enables greater prediction for balancing and control.
Enhanced outage management (OMS). The OMS market is evolving, and OMS functionality is now part of advanced distribution management systems (ADMS). Most expect the market for stand-alone OMS to decline with utilities’ opting to install more robust ADMS. Given the importance and ROI for enhanced OMS functionality, from both an operational and a customer analytics perspective, it will be paramount to have a clear road map for this.
Revenue protection. Energy theft in the U.S. exceeds an estimated $6 billion annually. A utility can capture data that reports locations with possible energy theft using standardized tamper events and sophisticated algorithms. In addition, technical losses from component failures and equipment malfunctions also contribute to revenue loss. A more comprehensive offering provides customizable dashboards that display summary-level statistics with drill-down capabilities linked to a GIS system and other tools and overlays for further analysis. An interface to the work management system also can initiate site inspection work orders.
Revenue reporting. This function generates revenue tracking and prediction reports based on energy and demand readings including settlement validation and is unavailable on a daily basis to most utilities. But combining this capability with weather-normalized estimates for revenue prediction would help utilities better manage cash flow and credit requirements.
Meter asset management. A utility can store and track meter configuration information, daily-read and data-quality statistics and other information over the lifespan of an AMI device. A dashboard allows users to assess the performance of their entire production system of AMI devices organized by user-defined variables-geographic, utility service area, feeder, meter type, meter configuration type, etc.-for any period.
Prepay tracking. This application monitors and reports on participants in prepay programs while providing a common interface to third-party prepayment solutions. Taken to its full build out, the solution could perform most operational components of a prepay offering except for the cash management functions and customer notifications.
Load forecasting. This analytics set includes a full suite of load forecasting and statistical estimation techniques that can model complex relationships among interval loads and such variables as weather, seasonal and time-dependent influences on power flows. Load forecasting can occur at various levels of aggregation including system, class, feeder, transformer, all the way down to the individual customer.
Load research and pricing. The load research function must meet the sample design, analysis and reporting requirements of a utility’s load research and program evaluation groups, as well as the information needs of various departments, such as rates and pricing, marketing, system planning and demand-side management. At a minimum, the system must develop efficient sample designs and estimate class and segment load profiles by combining interval load records through the AMI system, for example, with customer billing and other ancillary secondary data. This component is directed primarily to enhancing rate design.
Advanced energy efficiency/demand response (EE/DR) monitoring and evaluation. This function provides data and analysis tools to integrate billing with interval AMI load data and create tools to analyze EE measures and DR program offerings by utilities, load aggregators and evaluators. This component should couple tightly with other data analytics components-for instance, load forecasting, load disaggregation, customer behavior and auditing, TLM-to leverage other capabilities such as forecasting and estimation, data quality analysis, and customer characteristics from customer behavior and auditing. This component primarily is directed at tracking and estimating DR baselines, calculating demand and energy impacts and EE savings.
Load disaggregation. Understanding how customers use energy is essential to optimizing energy consumption. One of the early projects still active is the nonintrusive appliance load monitoring system known as NIALMS. This function provides direct estimation of customer loads by developing statistical relationships and leveraging information from the other modules using AMI interval load data. It can estimate accurately large customer loads, including baseload, weather-sensitive loads, heating, ventilation, air conditioning and water heating.
Data overlays for segmentation and program targeting. This component tailors messaging, program and product features to distinct customer segments and improves the performance of customer outreach. It must include a database of customer information, which could serve as the basis for developing predictions for program sign up, reduced consumption and load shifting. The utility could augment this database with such secondary sources as the census. Private vendors could supply another layer of information to identify subsegments. Finally, a utility’s primary research can be used to get customer feedback on programs and satisfaction.
Customer behavior, auditing and engagement platforms. Utilities are using with some customization customer behavior and engagement platforms as informational and mass market customer engagement opportunities. Generally these portals have an information page that summarizes customer account information and consumption history. The more sophisticated portals might provide load disaggregation capability, highlight the customer’s usage by major end-use appliance, and provide energy-saving recommendations. Additional offerings might include a profile of the home’s energy consumption relative to similar homes in their area, as well as to more energy-efficient homes.
Data visualization. With data visualization, utility staff and customers can explore, summarize and analyze time-series data associated with interval load information received from the AMI system. It allows the user to explore the data in alternative views: hourly graphs, load duration curves, daily, monthly and annual graphics, scatter plots, thermographs and more. These alternative views help identify patterns and irregularities in individual profiles, compare a single profile among time periods, summarize or compare several profiles, and identify relationships among profiles.
You Need a Road Map
The long list of predictive analytics applications for the grid and customers demonstrates the problem’s complexity. Building single-purpose systems for each component cannot be the answer; each component is closely linked with others, from a utility departmental view and an IT architectonical view with data and relational elements shared among components.
The process would benefit from an effective strategic and tactical implementation road map, which will allow a utility to take advantage of its own construct, needs and opportunities. Developing that road map will sharpen executive vision, spur executive sponsorship and inspire a team of business and technical analysts that can implement that map and achieve the data analytics dream.
Nicholas Abi-Samra is senior vice president of electricity transmission and distribution at DNV GL. He has served as general chair and technical program coordinator for the IEEE General Meeting of 2012 and is a professional engineer.
Claude Godin is director of energy data analytics at DNV GL and has 35 years of experience in the energy market, specializing in design and delivery of large-scale projects, meter data acquisition/management and energy analytics.
Curt D. Puckett is senior vice president of sustainable use consulting at DNV GL, where he is responsible for overseeing North America’s eastern operations, which includes offices in the Northeast, Midwest, Mid-Atlantic, Mid-South and Southeast regions.More PowerGrid International Issue Articles
PowerGrid International Articles Archives
View Power Generation Articles on PennEnergy.com