SteamSpeed - Turbo Performance Testing
Step 1: do math
Building a great turbos starts with great engineering. Before we build anything, we do the math considering the platform and engine characteristics. The basic idea is to build a conceptual map of the engines' flow and and compute optimal turbo wheel geometries for the platform, and how our turbos are likely to be used.
We won't go into the details here, but Corky Bell provides a good primer on how to do this in Maximum Boost (pdf). It is an algebra-level overview on how to design turbo systems, so it is approachable even with basic high school math if you want to learn the basics.
Actually, this is an approach we recommend everyone takes before buying parts for your car. You can design, plan, and buy once, and get the result you were looking for instead of buying the part over and over again when it doesn't work. This avoids problems where you find out after the fact that your turbo's turbine housing is too small because your car has bigger cams, and it will never be efficient and make good power, or your turbo is too big and is unacceptably laggy so your car is not fun to drive any more.
Of course, we have already done this for the products we sell, so you can be sure that all of our turbos will work well for your car. We also offer different turbo sizes for most applications to tailor the turbo to your goals.
Step 2. CFD & computer modeling
Once we have completed the basic designs, we further refine the designs with computer modeling to project how the design will perform with more realistic computer models. When we design new wheels, we also use numerical methods to compute harmonics, potential weak spots, etc. to refine their design. These details of how we do this, what our results are, and so on, are not public, but they hep direct our designs.
Step 3. build prototypes and road-test
After, we've done the math, we have a good idea what turbo design will work well, and it is time to manufacture prototypes for real-world testing. We install our prototype turbos on real cars, and see if they met our design goals. Most of the time, they produce the result we predicted in step 1, but sometimes we need to make adjustments. In either case, we use this phase to refine the part manufacturing, and we prepare to go into production.
Step 4. gas-bench testing
Years ago when we first started manufacturing our own turbos, something unexpected happened, our turbos were exceeding our predications by like 10-20% on actual cars on the dyno. Our first units were for the STI, JB hybrids based custom 9-blade turbine wheels in Mitsubishi geometries, and custom billet compressors milled to Garrett GTX 2867 and 3071 sizes. The turbine wheel was a lot more modern than the Garrett GT and MHI style wheels, but the compressor wheel was not bigger than a GTX wheel, it just a few more modern features like extended blade tips, slimmer hub, etc. We wanted to know how it was possible that it was flowing more 10-20% more than Garrett's GTX wheel.
We set out to understand why, so we gas-bench tested our turbos. As you can imagine, it is expensive, like $6k just for the natural gas to run the machine for 1 test, so we can one test analytically estimate maps of similar sized turbos which is a common practice in the industry.
The gas bench reveled why our turbos were making more power than we expected. the compressor wheel was flowing 10-20% more than Garrett's GTX wheel with a similar geometry. Our turbine wheel flowed a lot more than a similar GTX wheel with less efficiency, however our geometry was larger. We were really surprised with the compressor wheel results. We expected our turbine wheel to flow more with less efficiency, but now we had an actual detailed measurement.
Were we somehow smarter and did a better job than Garrett's engineers? Simply put, no. Garrett had different design goals, specifically OE-level reliability, so their designs and wheels are very conservative. They also often manufacture to lower tolerances and use more mainstream components. We've used a lot of Garrett Japan GT BB CHRAs in our own turbos, before we take them apart and rebuild them, many Garrett CHRAs are balanced below our standard for our products, and they use lower grade ball bearings. Its fine and reliable as they are engineered and designed for it, but this gets back to our design goals being different. We balance to higher tolerances and use more expensive bearing sections so we can be more aggressive in our designs to flow more.
Here is the compressor map for the STX 71's compressor wheel.
This is what it looks like superimposed with the GTX3071R map:
Here is our turbine map for our custom 9-blade TD06SL2 sized wheel:
Compared to a Garrett GTX3071R: