Skip to main content

Take A Load Off: How Software Is Cutting The Power Bill Of An Iconic French Plant

September 12, 2018
The city of Belfort in northeastern France knows all about speed and power. The hardworking town, which lies between the Vosges and Jura mountains, has made railway locomotives for nearly 150 years. The first TGVs, France’s iconic intercity high-speed trains, also rolled out of Belfort.
GE is also part of Belfort’s rich industrial history. Nearly 60 years ago, GE gave a local company called Alstom the license to manufacture gas turbines in Belfort. Alstom, which GE would acquire 50 years later, sent the first turbine to roll off the production line to Algeria, a newly independent nation.

The same plant has since produced more than 1,800 units, including a huge 212-megawatt turbine for French state-run giant EDF in 1986, which was, at the time, the largest such machine in the world.

GE’s Belfort plants keep cranking out massive turbines — one of them powers the most efficient combined-cycle power plant, according to Guinness World Records. But GE also is now mastering the fine art of energy efficiency inside its own towering production halls by harnessing the latest data science and smart algorithms. “The facility’s managers knew there were huge untapped efficiency gains throughout the site,” says Jean-Luc Roy, business development director at GE Power Distributed Energy Solutions. “GE decided one and a half years ago to take action.”

GE’s production lines in Belfort nearly doubled in size after the GE-Alstom tie-up in 2014. Its plants here now employ around 3,000 workers making and testing giant steam and gas turbines and using processes that require huge amounts of energy.

For example, the plant’s average power demand jumps from around 7 megawatts (MW) to approximately 20 MW as test crews put the newly manufactured turbines are put through their paces. The site consumed a whopping 14.4 gigawatt-hours of power in 2017, which is enough to power all of Paris for one morning.

 width= Above: GE Power's massive 9HA.01 gas turbine in Belfort, France. Top image: GE's Belfort plants consumed a whopping 14.4 gigawatt-hours of power in 2017, which is enough to power all of Paris for one morning. Image credit: GE Power.

However, the Belfort site is not as energy-efficient as it could be. That’s partly because some of its buildings and their red-brick facades date back to the dawn of the power industry more than a century ago.

But that’s now changing. “We now measure, interpret and analyze how energy is procured and consumed on the site,” Roy says. “Based on this data, we agree on an improvement plan that focuses on changing human behavior — not changing large assets.”

Engineers at the site knew they needed better data before they started to tinker with the process. Some of the steps were pretty straightforward. For example, the Belfort plant today has only a handful of electricity meters installed to measure the power consumed by its 46 buildings. GE engineers are now planning to install 130 submeters over the next months and collect data from existing gas and water gauges to build up a more detailed picture of energy consumption.

The submeters will be pumping consumption data into the cloud every 15 minutes. The wealth of readings from the metering devices will allow the team to analyze the data — so they can quickly zero in on, say, the energy efficiency of lighting in a certain workshop, or the inefficiency of a ventilator in another. “We can quickly see if one ventilator is consuming more energy than the other — and replace it,” Roy says.

But it is about more than just pure data gathering. Energy tech startup Evolution Energie — which makes software for energy monitoring and environmental reporting as well as commodities trading and risk management — is building the system on Predix, GE’s platform for the Industrial Internet.

Predix apps allow the team to analyze the reams of data it harvests. The software continually analyzes the input, looking for insights that can make the plants’ industrial processes more efficient.

For example, apps built on Predix could determine that a certain workshop testing newly manufactured turbines — which causes the site’s power demand to peak — is using a lot of energy. The workers could reschedule engine tests and run them during the grid’s off-peak hours, when wholesale electricity prices are lower. “We can break it down into parts like profits and losses on a balance sheet,” says Roy. “We can see who’s spending too much and vice versa.”

The target is to reduce energy consumption by 13 percent from current levels within the next five years, which would bring savings of around €400,000 per year by 2023. That is a considerable benefit for a site whose annual energy bill is around €3.6 million.

Now imagine implementing the same energy-efficiency system across all energy-intensive manufacturing sites: It translates into millions in savings — and that’s before counting any environmental benefits.