3D printed Stirling engine

This article is about a low-temperature Stirling engine I designed to be mostly built using 3D printing. I designed the engine to use Stratasys’s fused deposition modeling technology, an additive manufacturing process. You can see a video about the engine here:

This article will go through much of the same material that is in the video, but with more details.

My goal in designing this engine was to use 3D printed parts for simplicity everywhere I could without compromising performance. The 170 degree F (77C) temperature limit of the ABS material would limit the engine to low temperatures. Other (more expensive) printed plastics have heat deflection temperatures up to 372F (189C) that could be used in the future on the two temperature-critical parts.

A secondary goal of this engine was to serve as a test bed for experimenting with different regenerator material and to test design methods to improve the heat transfer from the operating gas to the hot and cold plates with increased heat transfer area and small flow channels. This first design is a fairly ordinary low-temperature Stirling engine design to use as a baseline.

The 3D printed parts

The version of the engine in this article uses 11 printed parts, all of them different. The next 5 photos show the printed parts as received from Stratasys.

The cylinder body houses the displacer that forces the working gas back and forth. It also contains the 12 regenerator channels that will contain the regenerator material through which the gas flows to exchange heat each cycle. 6 sets of holes on each end are for attaching the hot and cold plates. The design allows separate screws to be used to attach the two plates minimizing thermal shorting between the hot and cold plates. The low heat conductivity of the ABS material is beneficial for this application.

The displacer requires a relatively close fit as it oscillates back and forth in the cylinder body to force the operating gas back and forth through the regenerator. A low-temperature Stirling engine such as this cannot tolerate the friction of having the displacer rub against the cylinder body so maintaining accurate alignment is critical.

The 6 screw holes in the displacer are for attaching polyurethane foam disks that complete the displacer. They are visible in a later photo. These 6 holes were the only holes in the plastic that required threading for assembly. I could have molded-in nut holders as you’ll see I’ve done for many of the parts, but I did not want the additional weight of nuts on the oscillating displacer.

The flywheel support holds the split bearing tube which in turn holds the ball bearings and the crankshaft. I wasn’t sure if the printed plastic parts would provide a suitably smooth and aligned surface for mounting the ball bearings, so I designed the part to use the split bearing tube which I also planned to make from aluminum if needed.

In the future I may eliminate the split bearing tube (saving one printed part) and design the bearings to fit directly into the bearing holder.

The flywheel support performs a variety of functions in addition to supporting the flywheel. It also provides the displacer shaft guide, holds the power cylinder in place, and secures the two O-rings on the bottom that seal the power cylinder and displacer shaft guide against gas leaks. Combining the above functions takes care of most of the precise alignment problems so that engine assembly is simplified. These functions are visible in some other photos in this article.

The flywheel was designed to be built from both ABS and aluminum. For the flywheel you’d like a dense material like steel. The low density of ABS is not very helpful for a flywheel but it has turned out to be adequate for this engine so far. If I’m able to improve the power output in the future as I hope to, I’ll probably need to switch to the aluminum flywheel.

The four lightening holes are there mostly for better visibility to see the flywheel spinning in videos.

This photo of the displacer crank disk shows the side that mates with the flywheel. For convenience and to avoid tapping holes wherever possible, I designed built-in nut holders. For the four mounting holes the nut holders are on the other side of this part and are visible in the assembled engine. The setscrew hole shown in the photo makes use of a captive nut so that threading could be avoided even for set screws.

The connecting rod in the photo shows the as-manufactured holes for the pins. I designed them to be .005 inches undersized so that I could drill them out for accurate slip fits for the 1/16 inch pins. The undersized holes act as pilot holes that make drilling out the holes very simple. Even though the holes are undersize, the location and alignment of the holes perpendicular to the surface seems excellent.

The power piston crank was designed to use a clamping action on the crankshaft rather than a setscrew just to test this other mode of attachment. It doesn’t grip as tightly but performs adequately.

You can see on the crank a variety of crank pin holes. These holes let me modify the compression ratio for different operating conditions. For low temperature ratios you want low compression to operate the engine at minimum power. For higher temperature ratios the engine will support higher compression and higher power output. Moving the crank pin lets me test performance at different compression ratios.

The power piston insert lets me make a very simple piston because this insert simplifies the wrist pin connection to the power piston.

Non-printed parts and assembly

Most of the engine parts that were not printed were off-the-shelf parts such as machine screws, nuts, washers, ball bearings, and O-rings. Many of them are visible in the following two partial assembly photos.

I wasn’t able to design the engine (at least not yet) to use only printed parts or inexpensive off-the-shelf parts.

The engine works best with a power piston and cylinder that provide both a good seal and low friction. These somewhat conflicting requirements were satisfied by an aluminum piston sliding in a brass tube. These were the only parts that required machining on a lathe to true up the ends of the brass tube and to turn the aluminum piston to a slip fit in the brass tube. The only other power tool I needed for other parts was a drill press.

If others are interested in building this engine I can redesign it to use telescoping brass tubes that will work for the piston and cylinder and will not require a lathe.

The piston insert shown earlier attaches to the power piston with a machine screw. The parts are assembled with the connecting rod and wrist pin already attached.

To minimize friction I needed the crankshaft turning on ball bearings. I also needed four pins for both ends of the two connecting rods and a shaft for the displacer. The crankshaft was cut from 1/8 inch stock precision ground stainless steel shafting for an easy press fit into the ball bearings. The other pins and shafts were cut from 1/16” music wire.

The ABS material in this engine has a thermal conductivity that is less than 1/500 that of aluminum. For many parts of the engine the insulating properties of ABS (and plastics in general) are either beneficial or it doesn’t matter. For some parts, such as the hot and cold plate, good thermal conductivity is important. Thermal conductivity is important not only to move heat into and out of the engine, but but also to distribute it evenly so that hot and cold spots are avoided.
The hot and cold plates were made from 1/8 inch thick aluminum although any material .080 inch thick or greater would be adequate to avoid flexing.

The hot plate requires 6 mounting holes.

The cold plate was also made from 1/8 inch thick aluminum. It has a larger diameter for two reasons. The first is to allow clearance for the 3 mounting holes used for inverted operation (visible in some of the video). The second is for increased surface area for better cooling.
The cold plate has three additional holes to accommodate the displacer shaft, a hole for the gas to the power cylinder, and the threaded hole near the center for mounting the flywheel support. This flywheel support mounting hole is the only metal hole that required threading.

An earlier photo showed the 3D printed part of the displacer. The rest of the displacer is made up from 4 polyurethane foam disks. The polyurethane foam is a low-density closed-cell foam that is easily cut and drilled to size and has good insulating properties.

The printed parts needed holes drilled to exact size for press fits and slip fits of shafts and pins. This photo shows the displacer shaft and the fork pin. In all cases the undersized holes were already in their correct location and just needed to be drill to size.

Engine performance

The engine operating can be viewed on the video. In the video the engine was always at least partially warm so it would start quickly. Although I don’t want to deceive anyone that this engine starts instantly, I also don’t want to waste the viewer’s time while it heats up. From a cold start the engine will take anywhere from a fraction of a minute to several minutes to start depending on the power of the heat source. The 1/8 inch thick hot plate requires more time to heat up than a thinner one would.

The unloaded engine can be started with a temperature differential of about 45 degF (25C) and will run at around 100-120 rpm . The starting actually depends on the absolute temperature ratio which is approximately 1.084 for T(hot)/T(cold).

A temperature ratio of 1.13 (about a 70F or 39C temperature differential at room temperature) will results in an unloaded engine speed of 300 rpm. I’ve measured a maximum speed of 468 rpm.

To measure the power output I apply a friction source to the flywheel and measure the torque exerted by the crankshaft on the friction source. By also measuring the rpm I can compute the power output. For this engine so far it is quite low at .015 watts at 175 rpm. This was measured with hot and cold plate temperatures of 170F and 80F (77C and 27C). Unloading the flywheel resulted in the 468 rpm speed.

My next step in this project is to build some additional parts to improve the heat transfer between the hot and cold plates and the operating gas. With these changes I’m hoping to double or possibly triple the power output, but even .050 watts is still not very much power. It’s very difficult to get much power from a small, low-temperature, Stirling engine.

STL files for the 3D printed parts are available at thingiverse

Engine power output measurements