I guess what I try is to have a high zone created by the prop blast and velocity pressure and a low pressure zone created normally by a lip in front the exit opening. Then force the air to travel across the cooling fins while minimizing leakage.
The first photo is the Roto 85 FS installed in the BUSA 1/3 scale Super Cub, the firewall is the boundary between the high and low pressure zones. High pressure due to velocity pressure of the aircraft and prop blast, the low pressure caused by the scale sized lip on the bottom of the cowl and cowl protrusions on the sides of the. Since the cylinders are a few inches ahead of the pressure boundary, the fiberglass ducts are included. Over the past two years, one alarm over 300 degrees and this was on very hot day at the NATS this year. The alarm occurred while performing a Chandelle maneuver. The maneuver was almost complete so no action was taken. There is some leakage across the pressure boundary since there are a few gaps between the firewall and the cowl. At the last fun fly of the year, I had to idle on the runway for 6 to 7 minutes due to a photo opp, I was monitoring the temperature and during takeoff the engine was operating at normal inflight temperatures, flew an eight minute scale sequence without an alarm. The left cylinder runs 40 to 50 degrees F hotter that the right, not sure why, nothing seems different right to left. Works OK on all but the hottest days, the only reason I was flying on the day of the alarm because it was at the NATS contest, I would have much more liked sipping a Miller Lite.
I have the alarms set at 300 F with a sensor under each spark plug and on the top of each cylinder,four total.
This next photo is the 1/3 scale Champ with the Roto 85 FS. Here the pressure boundary is located in front of the engine more like what you used. Added the baffles behind the boundary plate to keep the air close to the cooling fins. The low pressure is caused by a lip on the bottom of the cowl with and exit area about equal to entrance area. There are two sensors on each cylinder near the top. This is an older engine and the plug caps were different, I don't believe a sensor could be fitted under the plug. The temperatures measured are closer to Super Cub at the top of the cylinder. Have the alarms set at 290F and over 5 years had a few alarms that were corrected with a mixture change. Guessing no alarms over the last 2 years. I would say it works about the same as the Super Cub.
The sensors are positioned in the same place on each cylinder, left side runs about 30 degrees apart while the right side match. I think the plog temp is more reliable.
The final photo brings us to this build. There's a fellow over on RC Scale Builder that has a wonderful BUSA Ercoupe, great and amazing build. He had overheating and added a baffle that solved his problem, he provided the file that was used to cut the baffle here. The pressure boundary on this plane is much better than the previous two and I expect it to work as planned or should I say hoping? Again the exit and entry area are very close in size. I can measure 6 temperatures on this plane.
