Winds of change: CenterPoint Energy enlists IoT for smart Houston power grid
Remember Hurricane Ike? A Category 4 hurricane, Ike slammed into Galveston, Texas, on Sept. 13, 2008. The storm obliterated the seacoast community of Gilchrist and then plowed inland, shattering office tower windows in downtown Houston and casting Houston's power grid into darkness.
“All our customers lost power. It was the worst event in our history,” says Dr. Steve Pratt, chief technology officer at CenterPoint Energy, which provides electrical power to 2.4 million customers across 5,000 square miles in the Houston metropolitan area. “Some customers lost power for three weeks. We could not assess the extent of the damage as quickly as we would have liked.”
Ike’s fury prompted a top-to-bottom rethink of CenterPoint’s power delivery and metering system, spurring the 140-year-old company to turn itself into a digital utility. The centerpiece of CenterPoint’s transformation is a massive Internet of Things initiative that includes intelligent meters at all customer locations and IoT sensors at key points along the power grid.
CenterPoint's decision to implement IoT places the utility in the vanguard of an industry wave that is growing steadily. By 2020, IoT spending will total $1.3 trillion globally, according to IDC. Utilities, which are expected to receive $69 million in investments by 2020, are among the industries that will receive the largest influx of spending.
Sensors + analytics
The CenterPoint system consists of several million IoT meter sensors feeding data to an analytics engine that can assess damage to the grid infrastructure, study power delivery patterns, and head off outages before they occur. The system tells CenterPoint managers whether power is being delivered and, if not, what the problem might be and whether it’s possible to route around it. It can also spot trends that may point to security breaches.
Sensors on meters at customer locations and on devices such as transformers on utility poles and at power substations gather information on the health of the grid—whether a piece of equipment is on and working properly—as well as how much power is being consumed at any given moment. The data is sent over a private radio frequency mesh network to four interconnected data centers, where structured and unstructured data is studied with the help of in-memory analytics on a SAP HANA database running Hewlett Packard Enterprise converged infrastructure hardware. The vast amount of sensor data collected is kept in Microsoft Azure cloud-based storage.
Costs down, satisfaction up
Seven years after intelligent metering was brought online, Pratt says availability is greater and customer satisfaction is higher than ever, despite serious weather events. “We have had massive storms that have taken out hundreds of thousands of customers since Ike. But now we can assess the damage and resolve issues exponentially faster than in the past,” says Pratt. By remotely rerouting the flow of power, outages have been minimized and recoveries have been faster. As such, fewer repair trucks and crews are on the road, enabling significant financial savings as well as reducing carbon dioxide emissions.
The benefits of the intelligent grid are not limited to weathering disasters. By tying information about the power grid into the customer service center, 7 million calls—ranging from inquiries about service availability and billing to reports of outages—have been addressed automatically, without human intervention, according to Pratt. In addition, service provisioning and billing have been streamlined. Previously, if a customer moved from one dwelling to another in the Houston area, it would take several days to shut off service in one location and bring it up at the new address. In the meantime, the antiquated billing system would issue a customer two bills. Now, such a transition can be initiated online, with no interruption in service and no billing overlap, Pratt says.
Analytics challenges and opportunities
Predictive maintenance is a key analytics initiative, and to that end, CenterPoint built an asset health engine on SAP HANA. The first implementation was to study the health of CenterPoint’s underground power distribution network. The system studies a number of factors and then creates an overall health score. “The higher the score, the more likely things are to fail. It gives us the information we need to repair or replace a piece of equipment,” says Pratt.
Applying analytics to better understand the impact of weather events on the grid is also highly beneficial. The system predicts where weather is most likely to damage assets as well as which assets, based on age and condition, are most vulnerable. “We can ultimately predict within a square kilometer where weather damage might occur,” says Pratt. Such knowledge enables CenterPoint to dispatch repair crews with a high degree of efficiency, placing them near likely break points and keeping them off the road when they are not needed. The result is both higher uptime and lower costs.
CenterPoint is also applying HANA analytics to its financial health, another important focus for the utility. By studying factors related to the organization’s financial picture—ranging from the cost of an outage to the revenue and service calls generated by an influx of new rate payers in a just-completed office park—trends can be spotted as they develop. That means top executives don’t have to wait until the end of each month to gain critical financial information before they can act.
In CenterPoint’s digital transformation, implementing analytics presented the greatest challenges. “It takes a lot of work to reap the analytics benefits,” says Pratt, who built an analytics team of a dozen members and brought in third-party experts as needed. Even so, he says, “it’s not inexpensive to develop sophisticated analytics, so there are some limitations on the time it takes to deliver those capabilities.”
Next, CenterPoint will build on its analytics accomplishments to implement more advanced artificial intelligence (AI) and cognitive technologies, with the ultimate goal of creating an entirely self-healing power grid by 2025. According to IDC, these efforts are right in line with where IoT implementations are heading. By 2019, IDC predicts that 100 percent of IoT initiatives will implement AI.
“We’ll deploy AI, machine-to-machine learning, and heuristic technology to guess the potential root causes of issues before they occur,” says Pratt, who asserts that CenterPoint will be one of the first utilities in the world to create a self-healing grid. Implementing a smart grid puts CenterPoint well on its way to its ultimate goal.
IoT and smart power grids: Lessons for leaders
- Predictive maintenance is a key analytics initiative.
- Applying analytics helps determine where weather damage might occur within a square kilometer.
- Tying information about the power grid into the customer service center enabled 7 million calls to be addressed without human intervention.
- Reaping the benefits of analytics takes a lot of work, says CenterPoint Energy's Pratt, who built an analytics team of a dozen members.