Cloud-based 'digital twins' could make power plants more efficient

Big data and Internet of Things technology may offer particular advantages to power systems that are spread out across whole regions, have thousands of specialized parts and have to operate at all times. 

General Electric supplies a lot of the parts that go into power plants and grids, and it's been working on applying IoT to the field for several years. Now the company's GE Power & Water division is giving plant operators an identical, virtual power plant that runs in the cloud. 

The Digital Power Plant is a real-time simulation that models the present state of every asset in a gas power plant or wind farm. By knowing what kind of condition each part is in, for example, operators can do fixes and maintenance when they're needed to keep things running as efficiently as possible. GE announced the Digital Power Plant at an event in San Francisco on Tuesday.

The virtual plant also lets administrators simulate different conditions and find out how the real plant would respond to them. For example, demand may go up or down because of weather or events and power companies want to be prepared to respond. Supply can also fluctuate, partly due to renewable energy sources such as solar and wind that vary as natural conditions change.

Utilities need to deal with a growing volume of data as equipment gets more instrumentation and energy grids grow more complex. A typical gas power plant has more than 10,000 sensors, said Ganesh Bell, chief digital officer at GE Power & Water. Only about two percent of that data gets analyzed today, he said.

The Digital Power Plant is based on Predix, GE's software platform for industrial IoT, but isn't limited to GE's own components in a plant. It can manage any asset in a power plant, Bell said. 

Applied to an already built gas power plant, the digital twin can save as much as US$50 million over the remaining life of the plant, GE says. That includes lower fuel costs, lower emissions, increased performance and reduced unplanned downtime. For a new plant, it could be $230 million.