In one of our original blog posts (What is Dynamo Series, pt 3), we alluded to the innovation being developed here at Dynamo—that we were developing a platform technology, based on turbo-machinery, to revolutionize small power products. We are going to spend the next few months showing the strengths of this platform, but first we will give you a little background on where this technology came from.
Dynamo was founded by two turbine engineers who had looked long and hard at the status quo for building turbines. We had firsthand experience working on the assembly lines for aircraft jet engines (in one of the first factories in the US to build jet engines, we might add). And we can confirm that all of your assumptions around building these turbines are probably true. Modern manufacturers work with super metals with esoteric names, like Inconel, Rene, and Waspaloy. They have machines that are 20 feet tall and can cut complex dovetails into solid disks of nickel with greater than 0.000,01” of accuracy (we call that tenths in the industry), and they have measurement tools to match. When you are pushing the envelope of engine performance, you need every tenth you can get. There is significant technology innovation being developed as well to improve manufacturability and product quality, from a machine that would friction weld shafts at high speed to novel ceramic composite matrix forming technology. A lot of work goes into building these parts; it’s not uncommon for a part to have a buy-to-fly ratio of over 90% (that means from the raw stock metal, only 10% is left over in the finished part).
As amazing this sounds, we also learned how 20th century the manufacturing process was. For a lean assembly line, there was not much of a line. Assemblies were put together by hand on mobile carts; the carts were moved around the factory floor to stations, where one type of work or another would be performed (e.g. welding, fastening, plumbing, etc). As often as not, engines would move back and forth between stations depending on the exact engine that was being built. The average time to assemble a small engine was two months.
On the parts level of manufacturing, there were other things that didn’t strike us as terribly modern. We called our business a “lean pull” manufacturing business, but the reality was that we built components in batches, and “lean pull” just meant we kept inventory in a holding pattern depending on what the assembly team told us to deliver in the next two weeks. We also did not have entirely fungible labor, and would spend a good deal of our planning time figuring out which machinist could make which parts on the given machines we had working that day. This combined with a metrics-driven culture resulted in some creative accounting. Sometimes we would build extra inventory when times were slow, just to keep labor working; I remember a few times we would “hold” unsalvageable components that didn’t pass their drawings check for a few weeks until we could “hide” the single reject when a large batch of inventory came through so it wouldn’t impact our metrics for that week. A large part of this seemed to be the fact that one out of ten of any batch would need to be re-worked at some point because the tolerances required by the parts were not met by the manufacturing process.
When there wasn’t a standard way to tell if a part was not conforming to the manufacturing requirements, we had to take the specimen to Al. Al was a living library with 30+ years’ experience making components for turbines—not an engineer by training, but a master manufacturer. His workspace, on the second floor, was filled with rejected components. Every one or two weeks I would bring a component to Al, show him the drawings and we would describe why we thought there was a problem. Al would gnaw his pen (which he also used to mark up the drawings), rub his brow and ask you to leave the part on his desk. You were to return the next day to hear his verdict on whether the part should be kept, reworked, or scrapped—and you took his word as gospel.
By contrast, I want to describe another engine factory for you; our founding team had the opportunity to tour a truck engine factory in North Carolina that was similar in scope to the turbine factory we worked at. This factory converted raw inputs to fully built, tested, and shipped engine in a week; and it did it at a rate of an engine every 5 minutes. While we did not have the same hands-on experience as we had at the turbine manufacturer, the differences were immediately clear. There was, for one, an assembly line! Engines would move down a conveyor belt; each station had a 5 minute step before an engine would move to the next station.
Even with this strict timing and specialized stations, each engine was built-to-order, with seamless inventory management in the background operation. Be it a different cam-cover or turbo-charger, the inventory was pulled to the specific station, and refilled as local supply ran low. Part of this was achieved with crude robots, where parts were delivered by following a set of colored lines on the ground from one side of the factory to the next.
What really inspired us, however, was that the diesel company was also building tens of thousands of small turbines as part of this process. Turbochargers are not the same as aircraft jet engines by any stretch of the imagination, but they do have a lot of technology in a small package. They have high speed bearings that must survive the constant loading and unloading of a diesel engine; and they have many little features that contribute to performance and life. When we compared the diesel manufacturer to our experience with turbines, we realized something. The products these two companies were building were for very different markets. By necessity, the turbine had to be built with critical alloys, exacting requirements, and a high rejection rate—partly because they are high performance products, and partly because so few were built a year (<500). In some ways, each engine was its own special production. The diesel units on the other hand are built in a cost competitive market, and where over 60,000 engines would be built—the manufacturing learning curve is also much faster with many more samples to work with.
But this also opened our eye at Dynamo. After seeing these two models, we asked the question “What if we built turbines the way they build diesel engines?” The result is a new way of thinking about the supply chain, of how the engine is built and assembled. It’s a new way to think about what the final product will cost, and how many we can build in a year. The other challenge is a market challenge; if we want to build 60,000 turbines, we have to find someone who wants to buy them. Luckily in the small power market, there are always people looking for something more reliable, more fuel flexible, and smaller than what they have today. In order to access all these customers, however, we had to also think of our product as a platform that could be easily adapted as needed for unique applications.