Many management experts are obsessed with “startup culture,” the particular mix of vision, energy and nimbleness that allows companies to take an idea, rapidly prototype it and get it to market in short order. But could the same spirit live inside a large industrial company? GE thinks so, with the help of FastWorks, its speedy product-development initiative. In fact, engineers at GE Power just used it to improve a popular gas turbine.
“In the past, we would make products that we thought the customer wanted, then we would check in with them,” says Anna Tam, an engineering platform manager at GE Power, the GE business that develops technology for power plants. “This time, we went to them well before we started.”
Tam and her team wanted to upgrade the 9E platform, one of GE’s most popular turbines. Not surprisingly, the 11 customers the team talked to came back with a list of clashing requests. Some wanted the machine to generate more power. Others wanted a turbine that burned fuel more efficiently. Another group wanted a model that they could directly swap out for the earlier turbine module, the 9E.03. All of them wanted to have the same high reliability and operational flexibility. And they all agreed that they wanted the new machine fast.
Rather than pick and choose between these competing wishes, the team deployed FastWorks and decided that the new development, called 9EMax, would have to fulfill all of them. Applied properly, the approach would allow the team to shave more than a year off the development process, which typically takes three years.
If this seems like a tall order, consider that the development team’s 300 members were spread around the world. There were turbine engineers located in the U.S. and Poland. Their colleagues in India were performing configuration analyses, and the machine components would be made in France.
The team ran into a major challenge pretty much immediately. Turbines use a system of “nozzles and buckets” to harness the energy of heated gas. They are arrayed like circular fans in stages along a rotating shaft. The 9E.03 had three stages, but to make the 9EMax work, they would need to add another one. The four-stage model, however, had to fit into the same exact space as its predecessor.
Having no time to spare, one group focused on streamlining the process and approvals. They started eliminating potential bottlenecks by empowering engineering experts who were further down the ladder to make important calls. “We made decisions when we needed with the experts,” Tam says.
Tam says that at the beginning, the team focused more on speed, releasing models for parts to be cast and machined from one group to another as soon as possible. In the old system, some of these steps could take up to a year. “Now, we release a lot of things when they are ready for the next process,” Tam explains. “If we think it’s ready to go, we’ll pass it along to our manufacturing counterpart, so they can start working on it.” She says this “concurrent engineering” enabled the groups to work in parallel, rather than waiting on one another.
These changes helped the team to “pivot,” quickly responding to new information, ideas or pitfalls. One place where this made a major difference was in the cooling system. The buckets in the turbine need to be cooled constantly; otherwise, the temperature from the hot gas will cause them to deform. The team was deep into the development process when they realized that the traditional radial cooling method, which uses channels that go from the bottom to the top of each bucket, wasn’t strong enough for their needs. Instead, they switched to another method, serpentine cooling, that used channels flowing up and down the buckets. It was the first time a serpentine system was ever used on a 9E platform. “It’s a more highly effective cooling scheme than we’ve ever applied before to E-class machines,” Tam says.
The global nature of the team also turned out to be a blessing. It allowed them to work around the clock, passing jobs from one time zone to the next.
Using FastWorks ultimately reduced the development time by 25 percent, from four years to three years. The first 9EMax, which was brought online in March, achieves all the customers’ performance goals: Its four-stage path uses an estimated $5 million less in fuel per year than the 9E.03, while increasing revenue by up to $6 million per year. “It was a huge jump in performance over the 9E.03,” Tam says.
Ultimately, though, the biggest advance might be in the way the 9EMax was developed — and the way it provided a test case for mixing fast, flexible startup culture with GE’s size and resources. “We really took advantage of the global nature of GE and our FastWorks culture,” Tam says. “It was amazing.”