Monty Python fans might recall the “Bring out your dead!” scene at the beginning of the “Quest for the Holy Grail.” Turns out “not quite dead yet” is a pretty apt description of my career. I debated for a bit about whether to discuss this publicly but thought it might be a good lead in for an over-due project update… Since I started flying in 1979, I’ve had five engines fail in flight: three Lycomings, 1 General Electric J-79 and 1 Pratt and Whitney F-100. While every fighter pilot dreams about being an ace, killing five engines isn’t exactly they way I’d want to go about that given my druthers. Seems more like a self-inflected gunshot wound. The most recent iteration of this bad habit pattern occurred in April at the conclusion of an AOA test hop.
Back in January, during the RV-4 condition inspection, I decided to send the carburetor jet out for flow testing and enlargement. It had run reliably for almost 800 hours since the engine was overhauled. For the test program we installed a fuel totalizer last year, and after all those hours, I realized that takeoff fuel flow was just under 12 GPH, but Lycoming calls for 13.5 GPH. I had vacillated so as not to violate the “if it ain’t broke, don’t fix it” decision matrix; but decided the FF calibration was accurate enough that it merited further investigation. As is typical in RV’s, modification of the primary jet in the carburetor may be required to get sufficient fuel flow on takeoff. I wasn’t comfortable drilling the jet myself, so I decided to send it to a carb overhaul facility for testing and any required boring. When I disassembled the carb, I decided to order a replacement “blue epoxy” float kit to replace the brass float. The metal float was serviceable; but I thought the improved design might be cheap insurance. An overhaul kit rounded out the order from Spruce. The carb shop tested the jet and bored it to increase fuel flow by 1 GPH, which we figured was a reasonable starting point. Since the carb is a single point failure item, I enlisted the help of a local AI to help with the re-assembly. Float alignment is critical; so, we were careful to follow instructions to ensure dimensions were within the required 1/32”. At least we THINK we did.
Initial tests showed takeoff fuel flow to be 13.5-13.7 GPH after modification. Touchdown. The airplane produced perceptibly more power during takeoff roll and initial climb segment. Excellent--more thrusties is always good. After a successful functional check flight after the condition inspection, we jumped right back into test ops. Most of the winter was spent with tweaking system calibration and installing the new Vector Nav VN-300 GNS/INS system in the airplane. Lenny did his standard awesome job tweaking code, and we spent more time testing in the hangar than in flight. We've also enlisted the help of a new team member: a local flight test instrumentation guru, Bob Baggerman. Not only is Bob one of the finest aircraft electricians I've ever met, he's uber talented in the analysis department and has automated a good deal of our data crunching. One of the fringe benefits of living in one of the Air Force's biggest test communities is meeting some really cool, smart folks! Beginning in March, we got back to a heavy flight schedule re-baselining all the test systems in the airplane, including the V2 (ONSPEED) box. Things proceeded swimmingly, despite my standard technique of “screw up, do over, think about it some more, try it again” accomplishment matrix. Like a lieutenant, I generally do the right thing; but I usually try some other things first…
By mid-April we were looking at “dynamic” vs “static” calibration (more on that in a later update). We designed a sortie with a series of Vmax to Vstall runs, followed by capturing a set of trim shots under identical conditions for comparison purposes. To minimize airmass effects, I do 95% or more of our test flying at dawn when the air is smooth and temps are lowest on the Gulf Coast. 12 April was our last opportunity to put some data in the can as I had to leave for training and a long airline trip on the 13th. I was at the hangar just prior to dawn and got a quick ground check of all the test systems. I really miss having a flight test engineer, crew chief and data systems specialist; because I’m not particularly good at any of those jobs. That lack of familiarity likely has a lot to do with my high “do-over” ratio. But if I build enough time in to the pre-flight schedule, I can get everything done without rushing (which never ends well).
After about an hour of pre-flight prep, I pushed the airplane out of the hangar as the sky was getting bright in the east. After strapping in, start was odd. The mighty-mighty usually starts smoothly after a few blades and settles down to IDLE, when I lean like a fiend to be kind to my plugs. This start was different. The engine barked twice, fired first on two, then three and finally all four cylinders. Then. It. Ran. Perfectly. Normal. Hmmm. I decided to modify run up to include a MIL power throttle chop and mixture cut-off pull to check control rigging as well as exercising the off position on the mag switch. Double checked primer locked. Everything was nominal and engine was running just fine. So, off for 8 deceleration runs and 12 trim shots…
I fly in USAF controlled airspace under positive control for all test ops. I like having a radar assist since I do all test work single-seat. We use a standard pressure altitude range of 5500-6000 feet for most of our testing. OAT was about 60 deg F. Air was smooth. Perfect. A deceleration run is exactly what it sounds like: accelerate to Vmax in MIL power (WOT), pull the power to idle, maintain altitude and decelerate to a stall. I typically run best power mixture for test points (easy to replicate using a known EGT); but for these runs, I decided to use a full rich mixture since I was letting the big dog eat. I was more concerned with not exceeding red-line RPM. Since almost all of my flight is maneuvering (the only steady state flight I’ve done in the RV-4 in the past couple of years are gathering data points), I usually end up moving my mixture control as much as I move the throttle. In this case, I figured I’d just eliminate it from the cross-check so I could concentrate on smooth pitch control and not over-speeding my expensive engine. That may not have been the most brilliant game plan.
The first couple of runs went fine. I was getting better at mitigating minor pitch inputs at high speed/low alpha. Cool. So was the carb throat temp; but to know that, I would have to had flipped the switch on the antique UBG-16 engine monitor. It’s part of my cross-check to reference carb temp “in range,” i.e., the descent check prior to landing. During cruise (or in this case maneuvering flight), I generally monitor reference EGT for mixture control. The UBG-16 only shows 1 digital parameter at a time. Great instrument, just display challenged. During a couple of the test runs, the engine stumbled when I pulled the throttle to idle. I don’t specifically recall which, and I didn’t note this in the tape. Not important when it occurred, but the Lycoming on the RV-4 doesn’t stumble. At least it hasn’t since it’s been overhauled and I’ve owned the airplane. Hmmm.
After all of the trim shots were recorded, it was time to RTB. The benefit of flying a long time is that you develop some help rules of thumb to compensate for cognitive decline. One of mine is that I can chaff off one anomaly, but two of anything constitutes a trend. In this case, the odd start combined with the behavior of the engine during the decel runs. I decided that I’d keep sufficient altitude during RTB to be set up for an SFO. Since I fly on the Eglin range, there is no shortage of runways if I connect the dots properly. As I normally do, I turned the camera off when finish up the test runs. Minimizing the amount of time the video runs simplifies “debriefing” the tape post-flight. Kinda’ wish I would have left it on.
Things proceeded nominally until I noticed a fuel odor when I switched to the fullest tank for landing during the in-range check. Hmmm. Third data point noted. I moved the top of descent close enough to the runway to allow for either a power-off straight in or 3K’ AGL SFO at my home field. I had just gotten a hand-off from the USAF controller when I began an IDLE descent from the top of descent. Things got quieter than usual in the cockpit during the descent. Hmmm. Fuel on, mixture rich. Energy was good, so it was more important to fly the airplane than to trouble-shoot beyond those cursory steps. Uneventful flame-out pattern. Really, it was uneventful. No different than practice other than the fact there was no go-around option. Nothing to focus your attention like a single-option decision matrix. Easy to manage energy burn with tone and glide path with maneuvering. Glide path is slightly steeper with no thrust vs IDLE thrust; but the RV-4 is still a low-drag glider (just like the big airplane I fly at work that doesn’t like to come down). In the mighty mighty, a combination of “lift flaps” and ONSPEED is the sweet spot for energy maneuverability when you are maneuvering in reference to the ground--say for an engine-out landing. I describe this technique in the academic articles here on the web site; but for folks that haven’t read those pubs, “lift flap” is a flap setting that produces more lift than drag. As flaps are deployed, ONSPEED and L/Dmax begin to marry up. In most airplanes, this optimum maneuvering configuration will be half flaps or less. Obviously, if range is the ONLY glide consideration, a flaps up, L/Dmax (best glide speed) condition is warranted. So, my luck continues to hold. At least for now.
So, what caused the engine anomalies and subsequent failure? I don’t know for certain as I type this. After towing the airplane back to the hangar, the first thing I did was pull the cowling. I found evidence of excess fuel on the FAB (Van’s fiberglass air box) and carb assembly. I also found fuel stains on the lower cowling; so something was amiss in the carb itself. Obvious causes would be a stuck float or needle valve issue. Cursory examination of the carb prior to shipping to an overhaul facility didn’t present anything obvious when the bowl was split, unfortunately. The carb is still at the overhaul facility, but initial examination by the technician didn’t show any anomalies as well. I may find out more or simply get a “CND” (cannot duplicate malfunction) write-up. Using standard technique of not biasing an independent analysis with any input, I sent the UBG-16 data to Ray Eaker over at Saavy Maintenance for analysis without any amplifying information. Ray is a world-class maintenance guru and has been a regular with the ONSPEED project since we started with the Gen 1 system. He noted low carb temps and then described the order the cylinders flamed out—about all you can get from the data stream. I pulled carb temp and fuel consumption data for the previous 20 flights, and, sure enough, during three out of the last 4 flights, overall fuel consumption was up about 1.5 GPH above usual. Carb temps were nominal for Florida winter ops. Ray postulated carb ice may have been a factor. The fuel staining was indicative of excessively rich mixture. Post flight examination of the lower spark plugs showed oil staining, which was to be expected since the engine has not run since it flamed out. Hopefully, we’ll get the carb back from the shop, re-installed and tested so that we can resume AOA test soon.
…oh, one more thing. The boom data locked up half way through the sortie, so we didn’t get those comparison trim shot data points after all. First time we’ve every seen that anomaly. Just goes to show ‘ya that it definitely ain’t worth pushing a test program like this one—too many variables to get cocky. We’ll continue our “no rush, get the physics right approach.”