Fixed leaks and started to do temperature testing. I attached the heat exchangers to the cold cylinder heads and circulating ice water using a pump. Also heated the hot cylinder heads with a torch. It seems that the cam is too eccentric, so the engine would need a higher temperature difference then I feel comfortable applying. The polyethylene bags have max working temperature of ~160 degrees F. One pair of cylinders lost pressure and the other pair has a slow leak. See follow up section below for temperature measurements and failure analysis.
I found one causes of the gas leakage, the tire valve stem was loose. Tightening it stopped most of the leakage. I also replaced another seal bag. It split near the head gasket and tore at a wrinkle in the bag near the piston. After replacing the bag, the 3-4 cylinder held pressure.
Cooling the cold cylinders
I made 2 heat exchanges out of 1" square aluminum tube, by sealing the ends of 5" sections with scrap acrylic squares. Drilled and tapped 1/4" NPT treaded holes into the side of the tube at both ends and screwed in hose barb adapters.
I was intending on using a rectangular shaped tube instead, to provide more surface area to transfer heat the flat head. But the square section tube was more available.
The heat exchanger is attached to the cold cylinder heads using plastic straps. A small 12V bulge pump is connected to both cylinder heat exchanges by 1/4" hose through a tee. The other side of the exchangers are connected to another tee and a return hose. The container is filled with ice and water.
Heating the hot cylinders
I used a propane torch with a very small flame to gently heat just the head, avoiding the rubber hose connections. I heated them until hot to the touch. Since I was in a hurry, I didn't take any temperature measurements. I could have used more ice and gotten a lower cold temperature.
Cylinders 3 and 4 spring a leak
After a few minutes into the test. Luckily 1 and 2 still had a pretty good seal, so I could continue.
One of the advantages of this engine configuration is that each pair of cylinders are essentially separate engines. Because of the cam drive, when a pair looses pressure, the pistons don't press down on the cam and just stay mostly out of the way.
This was a quick test performed, before I was to BBQ fish for dinner, so I didn't have time to hook temperature probes.
I could feel the engine want to run a little, even with only the 2 cylinders. When I calculated the cam ratio, I assumed shorter push rods then I currently have. I wanted to do a real test before committing to shortening them. I still have to check out the leaks that have developed in both cylinder pairs, just one is much worse then the other. I bought a 0-30 psi presure gage, so to make leak detection more sensitive, old gauge was 0-200. I am only pressurizing the cylinders to 10 psi right now. If I go higher it makes it to hard to turn the engine over. The plastic shaft starts to deform. I am also afraid the case my come appart from the forces. After I get the cam size nailed down, I will switch to steel. The 10 psi translates to ~30 lb of force for each piston, pushing on the cam. The pressure increases even more when the both pistons are near the top of their travel. I also need to add a gauge to the pair when turning it over, to compare the actual pressure against my calculations, which ignore the volume of the hoses.
Pistons 3 and 4 still have the original wrist orientation that is parallel to the main shaft. So I will shorten their push rods and drill the wrist pin hole perpendicular to the main shaft. This way push rod can slide if there is lateral forces on the push rod and not push against the bags as much. The other 2 pistons already have this enhancement, but will also need shortening. I am only going to shorten 3 and 4 for now and see if it is enough so engine can run on the current temperature difference 35 to 140 degrees F.
I will redo the cam/temperature difference calculations based the current configuration and see if it agrees with my observations.
Do an other temperature test on just the pair with the shortened push rods and take accurate temperature measurements.
I still have to find a better sealing bag material, one that is tougher and has a higher working temperature. Still trying to find a plastic supplier for nylon bags. The polyethylene bags keep breaking.
I measured the temperature of the head using a thermocouple probe that came with my multimeter. Without a thermal pastes the head was 8 degrees C. Cylinder #3 was the one that sprung the leak. A hole in the bag as usual. I used the new 0-30 psi pressure gauge to check #4 and it held 10 psi for over 2 hrs, while I watched Chuck and Heroes.
I am going to change the push rod length and wrist pin orientation. Also the piston core had some cracks where the set screw is threaded through to hold the piston shell in place. Luckily no damage to the shell. I am going to flip the core over and drill and tap two new holes.
The new push rod length from center of wrist pin hole to cam end will be 3". Full length will be 3.5" That should give a 1/4" clearance from the piston inner cap. The bottom of the stroke will have the bottom of the piston a little less then 1/4" from case side. The greater volume will reduce the required temperature difference without having to change the cam. Right now the cam is using the same 1-3/4" diameter Plexiglas solid rod stock as the piston caps. These solid rods near 2" diameter are quite pricey so sharing the stock for both pieces keeps the material cost lower. The circumference of these rods are smooth as glass, so it beats cutting circles form sheet stock. My band saw is out of alignment so I am having trouble cutting in a straight line, so most of the piston cap pieces have a tapered thickness. I try to arrange the high side of the inner cap with the low side of the outer cap, which presses the end of the bag into the piston cavity. This is quite snug, so I haven't had one pop out yet. The pressure should hold them in place anyway.