The Stuart Twin Launch Engine
I have admired the Stuart Models Twin Launch desktop steam model for many years. It is a lovely classical 19th century marine engine design that is self-starting and reversible with Stephenson valve gear. It runs nicely on compressed air, has a pleasing busy running sound, and is fascinating to watch in operation. The continuous motion of crossheads jumping, eccentrics wobbling, and valve gear flapping gives it an entertaining Rube Goldberg look that brings a smile to anyone who sees it. Plus, it fascinates the grandkids.
Alas, this wonderful engine seemed out of reach to me. I had built 3 relatively simple desktop engines previously a PM Research mill engine, a Stuart Models beam engine, and one of Jan Ridders ingenious Stirling engines. Altogether I had perhaps 2500 hours of modeling experience. Common sense would say that my chances of succeeding in building the twin launch were small.
However, once I discovered John Bentleys Constructing a Marine Compound Launch Engine on his web page at http://www.compound.modelengines.info I was hooked. Although these web pages are dedicated to the Stuart Models Compound Launch, the two engines are identical except for the cylinder diameters, corresponding steam chest dimensions, and piping. Soleplate, crankshaft, crosshead guides and bracket, eccentrics, and Stephenson valve gear are identical. Anyone desiring to build either the Twin Launch or Compound Launch must visit John Bentleys web site. The information and techniques shown cover every detail of the engines construction, and Mr. Bentley has generously given an amazing amount of his time to completely document engine construction.
I cant possibly improve on Mr. Bentleys detailed construction article, but I will offer a few observations of my own that might help other marginally experienced model builders who have passed up this beautiful model because of its apparent complexity. Believe me, it can be done.
A few additional observations.
Fasteners: I chose at the outset to use American UNC stainless steel machine screws and small pattern nuts rather than the British Association screws and nuts of the original design. In general this means that 7 BA was replaced UNC 4-40 and 5BA was replaced by UNC 6-32. I also used hex head machine screws rather than studs and nuts for the cylinder end caps and steam chest covers. Studs and nuts are so much trouble and not really necessary when tightening down a fixed cover. I like the look of the hex head machine screws although I will admit to facing off a few thousandths from each screw head to give a nice circular machined pattern rather than a stamped out appearance. Of course I continued to use studs and nuts for the piston rod and valve rod packing glands where adjustment in and out may be necessary.
Drawings: Honesty compels me to say that the drawings were disappointing. Drawings with this kit were faint, half size, second generation 10 by 14 inch Xerox copies of original drawings. Dimensions on these drawings are so tiny and muddled that they are often hard to read. Dimension numbers are sometimes confounding, as it is nearly impossible to see the difference between a 3 and a 5, for example, even with a magnifying lens. Dimension arrow markings are spidery and sometimes difficult to see. In general, all the information needed is on the drawings, but it takes a bit of extra concentration to extract it. I never cut any metal without double-checking that the dimensions actually added up correctly across the part. I may have misread an important dimension.
Also, I believe there is an error in the drawings with regard to the pistons. The drawings show pistons with a ¼ inch groove, whereas the supplied piston rings are 1/16 inch in width. Surely this cannot be correct. I machined the groves in the pistons to be a few thousandths greater than 1/16 inch for freely moving rings, and all seems to be well.
Drawings in my previous Stuart Models beam engine kit, as well as drawings I had purchased separately from Stuart Models were good black and white or sepia prints in 20 by 28 inch (or larger) size. So I am convinced that the drawings here are a temporary measure and not an indication of the usual quality of Stuart drawings.
Castings Sole Plate: The sole plate is relatively straightforward with two main areas to concentrate on accurate positioning of the support columns, and accurate alignment of the main shaft bearing with respect to the support columns. Because the engine block sits on 5 support columns above the sole plate and main shaft, it is vitally important that the support columns be accurately placed with respect to the main shaft bearings and the crank openings in the sole plate. I designated the left front and left back column location points on the sole plate as reference points, and thereafter referenced EVERYTHING on the sole plate to them (other column locations, main shaft position, crank openings).
Drilling the main shaft bearings after mounting the bearing caps on the sole plate will go smoothly by following John Bentleys procedure of first milling a guide groove in in both the sole plate and in the underside of the bearing caps. Im embarrassed to say that I attempted to drill this long bearing without the guide grooves and found out just how far astray a drill bit can go in three inches of drilling in gunmetal. Fortunately, I could mill out the bearings, solder in a patch, and try again.
Castings Engine Block: As with the sole plate, I used the left front and left rear column location points on the bottom of the engine block as reference points and located everything else (all block exterior dimensions in the horizontal plane, other column locations, cylinder centers, etc.) with reference to these points. However, there is another constraint to consider with the cylinder block. Stuart castings have the valve openings into the steam chest cast into the engine block. These openings are cast to within a few thousandths of their final dimensions, which means that only a little clean up with a miniature end mill is required. It also means that extra care is need to be certain that these openings come out on vertical and horizontal centerlines with respect to the engine block. In other words, if one is not careful, the valve openings can come out slanted or off center with respect to the engine block. I found that the critical time to address this is when initially setting up the two reference column locations and then in setting up the milling operation to establish the top and bottom surfaces of the engine block. It was a bit tricky but not impossible to satisfy all of the interacting constraints. I actually made up a template from a piece of 3/8 inch aluminum bar with four corners accurately representing the four outside column locations, and with marks on each end showing the vertical centerline. I could then maneuver this template by hand around the bottom of the raw engine block to find by eyeball the best location for the two initial column location points to ensure that the valve openings were centered. Then, when milling the top and bottom surfaces of the engine block, I took care to see that the valve openings were parallel to and equidistant from the top and bottom of the engine block.
Below is a picture of the valve openings in the cylinder block with vertical and horizontal center lines for reference. As you can see, my alignment is close but not perfect. This is an area that deserves meticulous attention when milling the surfaces of the engine block.
Looking back on this, I think it would have been easier not to have the valve openings cast into the engine block. Milling the openings and drilling the passages on my own would have been easier. However, Stuart Models has done things this way forever, and I dont expect any changes because of my small difficulties.
Castings Crankshaft: A cast iron crankshaft casting is provided in the kit, and John Bentley does a nice job in describing the process to machine it into a final form. However, I chose what seemed to me to be an easier way out and built up my crankshaft from 5/16 stainless steel precision shafting. I built a jig to hold three lengths of shafting accurately in place in relation to each other, slipped on crank webs that were then fixed in place with pins and Loctite, and then milled away the unnecessary sections to create the cranks. I also took the opportunity to add counterweights with the intention of improving engine balance somewhat.
Castings Steam Chest Covers: One of the cast iron steam chest covers had a hard corner that was not annealed properly. I found out about this by turning a $12 end mill into a rounded nub at this hard spot. Fortunately I was able to anneal the part by heating it to red hot with a propane torch for about 10 minutes, and then allowing a slow cooling. An initial quick check with a file around the extremities of the cast iron castings might have identified the hard spot without the loss of an end mill.
Other Castings: Other large castings such as the crosshead guide bracket, top and bottom cylinder caps, and steam chests did not cause a particular problem. However, there are a slew of smaller castings that were a serious challenge to me. These include the two connecting rods and crossheads, the four eccentric straps and rods, and the two eccentrics.
This really gets to the heart of what makes this model so difficult for a marginally experienced builder like me. These 10 castings are of high quality, but they are somewhat delicate and are cast with breathtaking closeness to their final dimensions. They also have no mercy with respect to work holding stubs or extensions. They are close to final dimensions, and good luck in holding on to them for machining. AND, there are no longer any spares available for sale from Stuart Models. The modeler must succeed the first time and every time in machining each of these castings. There is no room for error.
My usual process with a complex part is to fail and learn a few times before succeeding. It was abundantly clear that I would not succeed the first time and every time with these castings, so I elected to make up these parts from bar stock instead. This allowed me to build in work holding provisions in advance and also to fail a few times without jeopardizing the entire project.
Below is a picture of my collection of parts machined out of bar stock. There are no castings here.
Cosmetic Design Changes Cylinder Cover Bolt Pattern: The original design has a somewhat peculiar asymmetrical bolt pattern for the cylinder end covers. There are five bolt holes, which would indicate an equal angular spacing of 72 degrees. However, one of these bolt holes is pulled out of position because of interference on the bottom cover with mounting screws for the crosshead guide. Its not clear why this spacing was carried up to the top cover where it becomes visible. At any rate, I found that the interference could be eliminated by using a counter sunk flat head screw at the original 72 degree location on the bottom cylinder cover. This screw is nearly invisible, and can only be seen by turning the engine upside down. I then also repeated the equal 72 degree spacing of five bolt holes on the top cover.
Cosmetic Design Changes Steam Chest Cover: The original design had four bolt holes in the steam chest cover. To my eyes, this is not enough to look realistic. The Stuart Compound Launch engine uses 8 bolts, even in the steam chest cover for the smaller cylinder. So my decision was to use 8 bolt holes. This also required adjusting the size of the steam input and exhaust output flanges so as not to interfere with the additional bolt holes.
Stephenson Valve Gear: I built all of the valve gear out of 303 stainless steel using a rotary table and the techniques illustrated in John Bentleys article. There are many small parts to the valve gear, but none of them presented an insurmountable problem. My only modification to Bentleys technique is that I never fed parts by hand when creating a radius with an end mill. I always fabricated a jig of some sort to hold the part and allow use of the rotary table to control the motion of the part in creating the radius.
Assembly, Initial Adjustment, and Break In: Assembly went fairly smoothly for me although I did encounter a few unexpected interferences in the valve gear that had to be diagnosed and corrected by hand. Setting the valve timing was fairly simple, and the engine ran for me the first time I connected compressed air. To be sure it was stiff and noisy to begin with, but a break in run totaling about three hours smoothed out engine performance at both high and low speed. I was conscientious in lubricating everything liberally during break in.
I will say that the first ½ hour of break in produced an alarming amount of oily, gritty sludge at the top of each piston. I didnt see any sign of scoring on the cylinder walls, so where did this grit come from? My theory is that it was leftover particles from milling the valve openings in the engine block that blew through into the cylinders upon initial operation. I had tried to flush this grit out with mineral spirits before assembly, but apparently I left a fair amount.
Fine Tuning: Although assembly, adjustment, and initial break in went smoothly, fine tuning was another matter. It is easy to set the valve timing to its theoretical point once the valve rod is adjusted to give the slide valve equal throw above and below the inlet openings. The eccentrics are machined so that the forward and reverse eccentrics lead their respective crank angle by 70 degrees. Forward and reverse angular alignments are offset 70 degrees symmetrically about the vertical crank position. This corresponds to the top inlet valve just starting to open when the piston reaches top dead center, and setting the forward eccentric to this point automatically sets the reverse eccentric to its design position 140 degrees in the opposite direction. There is really only one theoretically correct setting.
However, there is theory and there is practice. I found that minute changes in valve position could have a drastic effect on the way the engine SOUNDS while running. Of course I expected a smoothly running engine without a lot of banging and clanking. However, tiny changes in valve position, while not affecting the apparent speed or air consumption of the engine, can create annoying clanking. There is a range of settings, all astonishingly close to theoretical, that will produce clanking in one or the other direction of rotation. Perhaps this is a reflection of small and accumulated errors in machining of the valve gear or eccentrics. Who knows? The best solution that Ive found is to search for a compromise that brings engine noise down to acceptable levels for both forward and reverse. I finally settled on valve settings that give reasonably quiet operation in both directions, perhaps somewhat better in forward. This fine tuning took more time than initial assembly, initial adjustment, and break in.
It is also quite possible that my fixation on the sound of the running engine is just another symptom of an obsessive mind. (Update based on a comment kindly provided by Hrcoleman66 (Hugh) - Setting the valves to open just PRIOR to TDC does significantly improve the smoothness of operation and the sound of the engine).
Set Screws or Not: I know that many modelers prefer not to use set screws for positioning eccentrics and valve gear. The rap on set screws is that they may not hold under extreme circumstances. However, set screws seem entirely suitable for this small engine. After all, these screws are only securing the position of the eccentrics and valve gear, which is an extremely light load in this case. I have used set screws for the eccentrics and the valve gear here and not had any failure to hold during hours of continuous running. Also, the valve settings are so touchy (see previous paragraph) that I seriously doubt that these parts could be permanently fixed into position by theoretical calculation. If pins or shaft dimples or Loctite are to be used, there must first be some method to experiment by trial and error to find the best valve settings before permanent lock down.
Compressed Air or Not: Purists will be horrified, but I only run my engines on compressed air. I am fascinated by the beauty of the running engine and not so much by the mechanics of steam generation. Also, running on air has the advantage of reducing corrosion due to unwanted condensed water in the cast iron cylinders and steam chests. There is a complication, however. Running on compressed air coming directly out of a warm compressor can also condense droplets of water inside of the engine. After running in this manner for several hours during break in, I was horrified to see drops of water at the end of the exhaust pipe. The best solution I have found is to run the compressed air through a long hose to give water an opportunity to condense out, and then to filter the compressed air just prior to the engine with a moisture trap. I now use a simple moisture trap designed for use with air brush paint sprayers which have a similar water problem.
Remaining To Be Done: The original design calls for cylinder lagging. However, I was so anxious to see the running engine that I have left the engine block bare except for a touch of paint to brighten things up. Eventually I hope to add proper cylinder lagging, although it appears to require tedious hand crafting based on John Bentleys article.
Final Reflection: I am very pleased with the completed engine. It is fascinating to watch and the grandkids love it. Perhaps more importantly, I learned an enormous amount during the 1000 or so hours required to build the engine. I highly recommend this model to anyone who is hooked on model engine building and who loves the design of these antique marine engines.