Modeling Oncor's distribution network: a "reality check" that improves reliability

Modeling the network allows tools such as distribution network applications

Few utilities have a larger or more complex distribution network than Oncor Electric Delivery -- serving more than 7 million customers in Texas, with over 102,000 miles of distribution lines and 4 million network nodes. To help ensure that power keeps flowing, Oncor has created a detailed, dynamic, near-real-time model of its entire distribution network.

On Feb. 5 at DistribuTECH, Larry Kohrmann, Senior Distribution Operations Manager for Oncor, will explain how this model was created, how it works, and how it has enhanced Oncor's operations.

According to Kohrmann, modeling a distribution network is a major undertaking -- but it offers substantial advantages over mere monitoring.

Modeling the network allows tools such as distribution network applications or an outage management system (OMS) to provide a clearer "reality check" of what's happening on an network -- enabling more accurate deployment of resources when restoring power.

"For example, if we have an outage, or need to reconfigure some feeders, a model would tell us with reasonable accuracy how much load we are moving between switches," said Kohrmann.

Also, modeling provides more accurate fault detection and deployment of repair teams. "It's a reality check on the outage management system," Kohrmann said. "If a customer experiencing an outage has been connected to the wrong transformer, that can cause the OMS to incorrectly show where an outage is. With our model, it's less likely that we'd send a crew out on a wild goose chase."

A distribution network model also can provide more accurate information for customers, especially estimates of power restoration times. This, in turn, tends to improve customer satisfaction indices -- important to both utilities and regulators.

Furthermore, modeling supports simulation -- allowing utility staff to virtually "try out" certain measures (such as repairing equipment, rerouting power, switching or replacing assets) to predict likely effects.

Data support from OMNETRIC Group

Oncor built its own distribution network model, with assistance from OMNETRIC Group (a joint venture between Siemens and Accenture).

Jayapal Parakkuth, OMNETRIC's Vice President for Grid Operations, clarified the practical daily value of simulation. "If I'm about to open a switch, I'd want to study what might happen, what ripple effects might arise on the network. You can test that in the model," he said.

Kohrmann agrees. "This isn't just about outage management," he said. "Every day we have planned outages and planned switching. We would like to be able to simulate these tasks in our model as part of our routine operations."

Large numbers of Oncor staff across many departments "touch" the model on a daily basis. Also, the model continuously receives and analyzes data from many sources across the network (smart meters, grid sensors and more) -- as well as from interconnected IT systems, such as outage management. With so many variables, maintaining the model's accuracy is a paramount priority and challenge.

Kohrmann explained that Oncor uses advanced analytics and data cleanup tools for every aspect of the model. They also perform ongoing field verification, and run error-detecting scripts.

"This database is big, really big," said Kohrmann. "Every time the model gets touched -- by a developer or operator -- those changes get tracked and resolved to keep the model accurate."

Besides responding to outages and keeping customers informed, a distribution network model can also support network planning -- an increasingly challenging task, with the fast rise of distributed generation. Vasan Krishnaswamy, OMNETRIC's Vice President of Analytics, observed that such modeling can help utilities make both daily and long-term decisions about system loading. It can also support predictive maintenance and network expansion -- by analyzing trend data for, say, how often certain transformers have been overloaded.

Many utilities don't have robust in-house capacities to handle big data and advanced modeling. OMNETRIC Group works with utilities to provide tools and expertise for data analytics and cleansing.

"It's not practical to construct a model for a large network feeder by feeder," said Parakkuth. "Using the right data tools significantly speeds up the process."

What's next for Oncor's distribution network model? Simply to keep it running, and keep it accurate. "This will always be a work in progress," said Kohrmann. "A model like this is never really finished."