Weather analytics help forecast, prevent and mitigate outages

Big Data Evangelist, IBM

Weather is a never-ending source of exploitable energy. It’s also a major reason why companies that specialize in delivering energy to consumers can’t rest, even for a minute. As long as the sun beats down and the wind blows, utilities that harvest solar and wind power are assured of steady supplies of energy.

But during a sweltering summer or a winter freeze, not only some but all utilities—including those that burn fossil fuels—must support customers’ urgent demands for air conditioning or heating. And when storms sever transmission wires, blacking out whole cities, utilities must be able to jump into emergency response mode without losing a beat. Consequently, every energy or utility firm, whether doing energy exploration and extraction or specializing in energy production and distribution, needs weather data analytics, not least for several foundational applications:

  • Outage prediction, prevention and remediation
    Outages caused by adverse ice, wind, tornado or hurricane events—even by lightning strikes—can cost utilities as much as $60 million for a single event. But weather data analysts can leverage granular meteorological, geospatial and infrastructure data to forecast the likelihood of outages, even at the neighborhood and street level. Just as important, they can identify measures that can help prevent likely outages—as well as evaluate the logistical requirements for responding promptly to outages that do occur.
  • Peak load forecasting
    Weather data analysts can forecast peak loads under likely meteorological scenarios, based on temperature, humidity and dewpoint predictions. This data, together with historical usage levels recorded under various weather conditions, can help utilities manage power consumption and load factor. It can also help them provision electricity effectively to areas having particularly high likelihoods of power failure.
  • Predictive asset failure mitigation
    Weather is a contributing factor to asset wear, tear and failure in the energy and utilities sector. Weather data analysts can identify equipment that is more likely to fail under forecasted scenarios. By using geocoded weather data in combination with equipment data—including maintenance and in-service histories—utilities can mitigate risk of asset failure. Failure prediction is an integral part of ensuring uninterrupted power delivery to all customers under all likely circumstances.

Among the energy and utilities stakeholders who stand to benefit from data-driven weather analysis are those responsible for transmission, distribution, generation, maintenance, fleet monitoring and diagnostics. Such analysis should incorporate both weather data and a utility’s own transactional and operational data, enabling comprehensive modeling of relevant scenarios. energy and utilities companies have leveraged weather data analytics to produce favorable business outcomes:

  • A large utility company using analytics and weather information to predict outages and allocate resources built an analytics solution that uses predictive modeling to identify situations that might lead to equipment outages, allowing it to proactively dispatch repair units. Drawing on current and historical data, the solution also provides business users with a resource plan helping them respond to outages quickly and effectively.
    The solution has demonstrated 79 percent accuracy in predicting when and where outages will occur, as many as five days in advance. In so doing, it has helped enhance planning and response for outage restoration while also taking into consideration resources and constraints. Moreover, it has provided insight into, among other important problem areas, resource constraints and the effects of weather on costs.
  • A national electricity grid operator uses predictive modeling and big data and analytics to implement condition-based maintenance. Using a cloud-based big data and analytics solution provides the operator a 360° view of its assets, from transformers to the entire grid. Predictive modeling and advanced analytics not only provide asset status in nearly real time, but also give long-term projections of maintenance requirements, helping the company plan future preventive maintenance.
    The company can now plan maintenance for each asset on an as-needed basis, rather than scheduling simultaneous maintenance for all assets of one type—thus cutting costs. The solution, which has produced a 23 percent reduction in operating expenses thanks to condition-based maintenance, provides alerts to facilitate proactive rather than reactive responses and uses cloud-based hosting to do away with the costs of implementing or replacing infrastructure.
  • A oil and gas producer in Australia using predictive modeling to anticipate and avoid costly equipment failures built an analytics solution that uses predictive modeling to identify situations that might lead to equipment outages, allowing it to proactively dispatch repair units. The solution draws on current and historical data to provide insight into production potential, allowing business users to create enhanced forecasts and make highly intelligent purchasing decisions. In addition to demonstrating 87 percent accuracy in predicting potential equipment failures with 48 hours’ warning, the solution has saved millions of dollars because of the insight it gives into purchasing decisions.

Energy and utilities companies can learn how to do weather-driven analysis for themselves at IBM Insight 2015, scheduled for 25–29 October in Las Vegas. At Insight, IBM and The Weather Company will demonstrate how to use weather data packages and IBM data science tools—such as Apache Spark as a Service in the Bluemix cloud—to address urgent weather-related business challenges by building powerful smart data applications for meteorological forecasting. Additionally, discover how to use Apache Spark as a Service to hack your analytics-related business challenges.