The two most rock solid facts concerning the LBG's last few minutes are uncontestable. First, the aircraft must have passed almost directly over the position at which Lt. Woravkas remains were found in 1960 at 26*54’N, 24*08'E. His parachute opened but became entangled, so he would have dropped nearly straight to the ground. Secondly, the aircraft impacted the ground some 14 miles southwest at 26*42’45.7”N,  24*01’27”E at an indeterminate time afterwards. As the crew apparently thought they were bailing out over the Mediterranean Sea there is a strong possibility that the bail out of all nine happened very quickly, within much less than a minute's time, so as to stay together as tightly as possible and attempt to rally at sea as a group in wait for rescue the next day. Our "hands free" portion of the simulation therefore would begin at Lt. Woravkas position. In addition to these two irrefutable facts are a series of other less clear factors that, if reasoned correctly, add a great deal of import as to how the flight model should be set up before attempting to run a simulation. These postulated "known" parameters, if correctly configured, should contribute much to the success of determining the "unknown" facts accurately. Some of the things that could hopefully be gleaned from this process are:

1. At what altitude did the bail-out occur?
2. At what speed was the LBG flying at bail out time?
3. What was the descent rate of the aircraft in feet per minute?
4. Could a Liberator running on the outboard starboard #4 engine be radically trimmed so as to arc to the right, when the torque would naturally pull it strongly to the left?
5. If such trim could be achieved, would the bank angle and general attitude of the aircraft at impact match the crash site evidence?
6. How long after abandonment did the LBG fly alone before crashing?
7. What was the speed of the aircraft at impact?
8. If the above questions could be determined with some degree of accuracy, what other questions or deductions could be inferred from the data?

Determining the Known

Extracting as much as we could from both the hard evidence and carefully reasoned judgment, we set up the aircraft for testing with the following considerations, flight parameters, and and expected results in mind:

One Engine Running:

When the wreck was first discovered, the propellers for engines number 1, 2, and 3 were found to have been feathered (i.e. the blades rotated to reduce wind resistance). Thus, it was clear that only engine #4 (the outboard right hand engine), was running at the time of the bail out. Because a B-24 cannot maintain altitude in any configuration on one engine alone this would have been the primary reason for abandoning ship shortly after the failure of the third engine due to fuel exhaustion. We would therefore run on only engine #4 from the point of bail out.

Trim Settings:

The WWII B-24 pilots manual states that if there is a failure of two engines on one wing, the rudder trim should be set to the full extent of the trim knob in the direction of the running engines to counteract the aircraft's natural tendency to pull in the direction of the failed engines. It further states that this configuration will assist in, but not eliminate, the need to apply pressure to the rudder pedals in order to maintain a consistent heading. Interestingly, there is no mention in either the 1942 or the 1944 versions of the pilots flight manual of any procedures pertaining to a three engine failure scenario. We can only assume this is because it was generally understood during pilots training that a Liberator cannot maintain it's altitude at any flight level with only one engine operating, and that abandoning ship then becomes the crew's priority.

We had hoped by studying the Army investigation photos of the center flight console (where the trim knobs are located), we would be able to match the settings as found at the crash site with those in the cockpit of our model. Unfortunately, it appears that these knobs were touched or altered sometime before the Army investigators got to the scene by someone one with Don Sheridan's party three months before. Sheridan apparently stated as much (either in his memoirs or to Mario Martinez while he was researching his book "Lady's Men") because Martinez relates that Sheridan was aware of them having been fiddled with. The Army investigators did not touch the trim knobs until they were completely photographed, but the photos clearly show that they had been touched fairly recently due to the 15 years of accumulated dust having been rubbed off on a few places on the knobs. We have some other questions about this matter which may be addressed in an update later.

For our purposes then, we set the rudder trim knobs to full right rudder (as suggested in the WWII manual for a two engine failure on one wing), and found that with only engine #4 running there is still a slight pull to the left, though interestingly, somewhat reduced from the torque imposed by having both right engines running. The only way to stop this leftward tendency, without applying rudder pedal pressure, is to turn the aileron trim tab to the right. Not far, but just a bit. At about 2.3 points to the right ("points" as read by the numbers on the aileron trim wheel), the compass stops drifting left. Though the aircraft steadily loses altitude on only one engine, it none the less flies relatively straight this way. However, it takes on a "wounded bird" kind of attitude because it's right wing remains lowered a few degrees and it's nose is yawed about 3 degrees to the right.

We decided that this must have been very close to the only way Hatton or Toner could have hastily set the aircraft trim so as to maintain relatively level flight for the time it would take to abandon ship. We therefore decided to run several parallel simulations, all with full right rudder trim, but varying aileron trim settings of  2.3, 2.4, and 2.5 points to the right to see how the Liberator "settled" into the unmanned portion of it's flight with each configuration.

Throttle Settings:

The B-24 Flight manual states the following RPM and manifold pressure settings are to be used to for the given circumstances listed:

Exactly how Hatton and/or Toner set the engine settings for the remaining engine in the moments after their third engine quit cannot be fully determined from the crash scene evidence because of the previously mentioned potential tampering with the flight controls and the fact that engine #4 was ripped from it's housing along with the ends of the cables connecting it to the control console, which may have altered them as well when the cables were torn. So where should we set our model's throttle controls?

We ran a couple of test runs, varied the engine settings, and found that if the engine #4 were throttled all the way up to takeoff settings the pull to the left increased quite a bit. Alternatively, throttling down to normal cruise settings diminished the left hand pull considerably but the Liberator begins losing altitude at a much faster rate. We therefore felt that it would have most likely made sense for Hatton and/or Toner to increase the RPM and manifold pressure up to or just above that of the maximum climb settings of 2550 RPM and 46hg . This would have not only been perhaps "instinctual" for a pilot to do, but would have made common sense too. For even though they had no reason to refrain from running the engine up to it's "War Emergency" range at that point (the ship was soon to be lost after all), it was still presumably going to require a few minutes to trim her level, prepare for bail out, and exit the plane cleanly. Knowing that the engine was tired and almost out of fuel after a dozen or so hours of continuous running, and that anything above 2550 RPM and 46hg manifold pressure is only recommended for brief periods of flight, it seems likely they wouldn't have wanted to take chances at that point of having the engine suddenly "running away" or falling precipitously by pushing it too hard. There is also the issue of the increased left hand torque exhibited by the engine at emergency power, which would have made it all the more difficult to hold on a straight course by trimming alone. We therefore decided to set the engine at 2600 RPM (just a hair above maximum climb power) and 46hg manifold pressure for all our test flights.

Flight Heading at Bail Out:

From the fix they got from Benina tower near Benghazi at 12:12 am. It's believed the LBG subsequently traveled on a course of about 150 degrees thereafter. From the position of Lt. Woravka as it aligns with the Benghazi area along that same line it appears they held this course for the next two hours until they were forced to bail out due to fuel exhaustion. But was the aircraft at that same heading during bail out? When the third engine quit, could the plane may have drifted off course a bit while the pilots struggled to trim her for bail out? It seems likely that since the aircraft, now on one engine, was still traveling at over two miles per minute, and that bail out procedures would take a few minutes to coordinate and execute, Hatton would have attempted to hold their "last known" course line because it was presumably going to be the first "line of travel" that would be flown by the search planes at dawn. Indeed, the currents in the Mediterranean might drift them off this line by dawn anyway, and compared to the immense benefit of being found in short order the next morning, with Toner in the cockpit to assist, it would have taken comparatively little effort to keep the plane on course for the last few minutes, therefore maximizing their chances of being found by search teams. Although there is no evidence to support of refute any particular heading for the aircraft at the time of abandonment, it's likely that they would have attempted to stay on their original line of flight for as long as possible. This could have been accomplished easily by one pilot manually holding the heading while the other cranked around the rudder trim to maximum right rudder, and began to find a reasonably balanced aileron trim that temporarily stabilized the compass drift. Considering these theoretical assumptions were a reasonably good pilots instnicts, and lacking any firm evidence to the contrary, we decided to set the heading at 150 degrees for the start of the simulation. Though this is the most important aspect of any simulated run, there are a several other factors to suggest this heading is probably correct; we'll discuss them later in the section Bail Out: A Question of Wind.

Payload:

This was pretty much a no brainer. We removed the weight of all nine crew members, all bombs (the crew had presumably dropped them in the sea after turning back from near Naples), and all but 50 pounds of fuel. This equates to just over over 6 gallons of fuel, which translates into roughly ten minutes of flight time for the remaining engine. This is a "guestimate". All the fuel that may have remained in the tank feeding engine #4 had completely evaporated by 1959).

Crash Site Requirements: Heading and Prop #4 Impact:

The crash site evidence requires that two parameters be met at the time of impact. The first requirement is that the aircraft must not only arrive at or near the location of the crash site, but must impact the ground at a heading approximating the alignment of the impact debris field. Neither Walker, McClendon, or Martinez however, mention the alignment of either the LBG or the debris field. We're unsure if it is even mentioned in the Fuller/Neep Report. It has been generally referred to having been found facing "almost due east" with the debris field stretching out across the desert directly in front of it as pictured at right, indicating it cart wheeled about 180 degrees before coming to a stop.

To narrow down the heading at impact we studied several photos of the LBG taken at low sun angles, one on the morning February 28th, 1959 when Sheridan's party was there, and another taken May 24th or 25th, 1959 when the Army investigators first arrived. By placing the flight sim model at the crash site and then setting the date and sun angle to approximate that which is seen in the photos, it appears that the LBG's nose faces approximately 107 degrees, meaning that the heading at impact would be the reciprocal of about 287 degrees. It's a bit hard to tell however if the debris field lines up perfectly with the fuselage itself but it doesn't seem to be any more than a few degrees off and perhaps a hair to the north. We will therefore be looking for an impact heading of somewhere between 275 and 295 degrees ***.

The second requirement our virtual Liberator must fulfill is what part of the aircraft should touch the ground first. Evidence at the scene strongly suggests that it was propeller #4, still running on it's last few drops of fuel, that sliced into the ground first. Why? The engine was violently torn from it's housing and ended up scattered in pieces across the debris field. This also may explain the clockwise frisbee-like cartwheel that the aircraft apparently made before slamming into the ground, breaking it's tail section just before sliding to a stop. Neither wing tip had any sign of impact to it's underside, strongly suggesting that this cartwheel was quite "clean" and even perhaps somewhat "airborne" until nearly the end. We will thus be looking for an attitude at impact where propeller #4 is clearly the first thing to contact the ground.

*** If anyone has any hard data as to the exact alignment of the LBG fuselage and debris field as originally observed when first discovered please contact us and let us know.

Wind Considerations:

We suspect there was a bit of wind on the night of the incident, but for the sake of the original set of controlled tests we did not load any. Our theory about this is based on the positions of Lt. Woravka as compared to the place where the "Rally Point" items were discovered, and other associated factors. We will explore this in more detail in the discussion below entitled Bail Out: A Question of Wind

So here we go. The charts below chronicle the results from a total of 15 simulated runs at 5 different altitudes, 3 flights for each altitude.

With the aircraft configured with the above flight parameters, we ran each flight from a point originating at Lt. Woravka's position, and repeated this at each of these altitudes: 5000 ft., 4500 ft, 4000 ft, 3500 ft, and 3000 ft. We repeated the run at each particular altitude three times, with different aileron trim settings: 2.3 points right trim (blue lines), 2.4 points right trim (green lines), and 2.5 points right trim (red lines). The results are mapped below in screenshots overlays taken from FS9's in-sim flight tracking map. The three thumbnails below each map are screenshots taken of the aircraft at the time of first impact for each respective run.

Moving your mouse over any thumbnail will bring up a larger view of the impact screenshot with the data recorded by the simulator at that moment. This data contains location, speed, and heading information. We've color coded the thumbnail borders to match the corresponding flight line for the respective aileron trim settings used in each run. The black airplane icons are the position at which the Liberator first reaches the ground for each flight track. The position of the actual LBG crash site is marked on each map for reference.

Test Runs Results

After running and charting these tests we observed the following:

1. The first thing that becomes apparent in studying the results is how drastically different the flight paths become by altering the aileron trim by only a tenth of a point. In all 15 flights the aircraft appears to fly relatively straight for the first minute or so before it begins "settling" into it's particular descent path. Either of the three of these aileron trim settings, or altitudes would have temporarily appeared to Hatton and/or Toner to have stopped the compass drift long enough to proceed with and execute the bail out routine. In this sense, the final fate of the LBG having been trimed in a way that preserved it's structural integrity at impact appears to have been completely random.

2. The second thing that becomes apparent is that the scenario that begins at 3,500 ft. appears to best approximate the distance required to reach the ground at or about the approximate location of the crash site. The 3,500 ft. 2.5 trim scenario in particular (red line) puts the aircraft nearest to the crash site of all the tests we ran.

3. The third thing you will notice when comparing the maps is that the lower the starting altitude, the less pronounced the tendency to arc to the right becomes for a given aileron trim. In other words the flight track straightens or tends to open more to the left if started at successively lower altitudes. Why? We presume this to be a factor of the slightly increased air pressure at the lower altitudes allowing for greater thrust or "grab" from the remaining engine's propeller.

4. Fourth, by studying the thumbnails it becomes clear that almost all the blue tracks have the aircraft fuselage hitting the ground simultaneous to, or before prop # 4 does, and that the red tracks from the higher altitudes show the right wing nearly touching first. The red track from 3,500 feet however not only ends up closest to the actual LBG position, but the bank angle of the aircraft clearly illustrates that prop #4 strikes the ground first, and the fuselage and right right wing are equally well off the ground, just the requirement needed for a clockwise cartwheel to occur.

5. In addition to the "first impact" requirement, the red track from 3,500 ft. also appears to come closest to the heading requirement as well. This track impacted the ground at 308 degrees, just a bit north of our estimated 275 to 295 expectation. Looking at the chart for the 3,500 ft tests, it's apparent that if the aileron trim were moved slightly left, from the 2.5 setting to say 2.48, that the arc of the flight would open to the left a bit and the position of impact would correspondingly move to the south. This would then naturally move the impact heading into the required range as well. Unfortunately we could not test this trim configuration because the aileron trim in the simulator model can only be changed in whole tenths of a point. The actual trim tabs on a B-24 however, are linked by a cable that is controlled by linear turns of the trim knob and could have easily ended up at this incremental setting.

So, based on the 3,500 ft red track flight data then, some of the other questions we posed can now be answered:

1. The bail out began at somewhere near 3,800 feet ASL, by the time crew had evacuated the aicraft it had fallen to around 3,600 or so. This is actually less than 3,000 feet off the ground, as the Canascio Plain is about 800-900 feet above sea level.

2. The speed of the aircraft at bail out was near 111 knots IAS (128 mph IAS)

3. The descent rate "averaged" about 300 feet per minute throughout the flight. This number was slightly higher at atitude and seemed to decrease as the plane descended. (This varied from an average of 450 fpm for the 5k red track, to 200 fpm for the 5k blue track.)

4. The elapsed time (as clocked in the sim) from Lt. Woravka's position to impact was 7 minutes, 36 seconds.

5. The speed at impact was approximately 114 knots IAS (131 mph IAS).

The VSI Photo...

An interesting footnote to the descent rate we observed may be found in a close-up view of the vertical speed indicator (VSI) as photographed by the Army Investigators in 1959 in the cockpit of the Lady Be Good after several of her gauges had been removed for study. VSI needles indicate an aircraft is flying level when the needle is at the 9 o'clock position, they move upward toward the "10 o'clock" position when ascending, and downward the "8 o'clock" position when descending. VSI's operate by measuring incremental changes in atmospheric pressure inside a semi-sealed chamber with a sensitive diaphragm at its center. There is a small "calibrated leak" allowing air into the chamber from the outside which expands or compresses the diaphragm momentarily allowing it to transfer this difference in air pressure to the indicator needle by way of delicately calibrated linkages.

Because it takes a moment for the changing air pressure to enter the chamber, effect the diaphragm, and transfer the data via the linkages, there is usually a slight delay in sending the data to the needle. In addition, because the internal parts of a VSI are quite sensitive, it's not uncommon for the diaphragm or linkages to become damaged at impact. This can sometimes "freeze" the needle at or near the position it was indicating when the damage occurred. Because this is a sealed gauge that cannot be tampered with without opening it, it's very unlikely it was changed by anyone prior to the photograph below. The photo clearly shows the VSI needle is "frozen" at a descent rate consistent with the 300 feet per minute rate we observed, and compellingly corroborates the 3,500 ft. red track descent rate.

Bail Out: A Question of Wind

One of the nagging questions surrounding the bail out of the crew has been why the apparent "rally point" of the crew was found 4/10th of a mile southwest of Lt. Woravka's remains. If the LBG passed over Woravka's position (as it must have because his chute did not fully open), and the crew had landed aligned along a line of roughly 150/330 degrees from his location, how did they end up rallying 4/10ths of a mile southwest? Bail out protocol, as chronicled in the B-24 pilots manual, calls for the crew to prepare for bail out by opening the bomb bay doors, the nose wheel hatch, and the rear escape door, located on the underside near the tail of the plane. The first to exit are to be the bombardier (in this case Woravka), and then the navigator (Hays) through the nose wheel hatch. Then the tail gunner (Adams) and the left waist gunner (Shelley) are to leave through the rear escape hatch. Then the right waist gunner (Moore), followed by the flight engineer and the radio operator (Ripslinger and LaMotte) are to jump through the open bomb bay doors. Finally, the co-pilot and pilot (Toner and Hatton) through the bomb bay doors as well. Had they followed this protocol, Woravka would have naturally been the the first crew member to leave the plane. Had they all followed shortly thereafter, even at an interval as tight as 3 seconds a piece, the bail out would have taken 24 seconds to achieve, during which time the LBG, flying at 111 knots would have covered nearly a mile of ground.

At first glance one might postulate that the LBG was traveling at a heading towards the southwest in it's final moments, and that this could explain why the rally point items were found where they were. The two problems with this scenario however are that, first, had the crew bailed at this heading they would have presumably assumed that Woravka could be found northeast of them, and therefore would have logically walked in that direction a bit farther before deciding to head northwest towards the coast. The second problem is that if the LBG was headed in a southwesterly direction at the time of bail out, the bank angle required for the aircraft to reach the crash site and impact propeller #4 first could not have occurred as it would have taken on more of a level winged attitude, impacting the fuselage either before or simultaneous to prop #4, and thus not experiencing the clockwise torque required to support a 180 degree cartwheel that leaves no marks on the underside of either wing.

These anomalies can be readily explained however if a light wind from a generally NE direction were present at the time. But can this be verified? We searched for meteorological records of seasonal winds in central Libya and found that the NOAA database extends back to 1949. Using the wind data from this year (and nearly identical trends from later ones) a seasonal pattern is clearly evident. This data shows thaw the predominant surface winds at the bail out location for March thru April are generally "light winds" from the NE to ENE. The average speed of these winds can even be estimated from the java based applet at the Columbia University web site that hosts the data. We've charted and interpreted the months in question here: NOAA Libyan Winds Database - 1949. Granted, though no one can be certain of what the exact weather conditions actually were on the Calanscio Plain on the night of April 4th, 1943, the historical data clearly suggests that light NE winds were certainly not uncommon.

With this in mind, adding an 11 knot wind from the NE into the scenario suddenly explains a great deal of the bail out anomalies, as illustrated in the graphic below:

Test Run With Wind

If the the above wind factors were actually in play, how would it effect the unmanned descent of the LBG in the simulation? We ran a few more more test runs to see. We entered an 11 knot wind from 50 degrees into the simulation and ran four test runs varying in starting altitude from 3700 down to 3400 feet. The 3600 foot test run results are charted below:

This run impacted the ground only 0.53 Miles NE of the actual position of the LBG, the closest impact observed in all our testing.

Final Thoughts

Considering the plethora of variables that obviously existed on the night of the crash, (fuel load, weight load, throttle, trim, engine performance, atmospheric pressure, wind, etc.) it would be extremely presumptuous to assume that we had resolved the truth of the Lady's last minutes. The assumptions of accuracy that we attempted to input into the simulation were based on informed conjecture, nothing more. It's tempting to think that some of the data gleaned can be understood outside the envelope of reasonable doubt, but it can of course never be verified. That said, the simulation results tend to nonetheless support, rather than refute, at least some of the evidence of the crew's bail out positions, and evidence observed at the crash site nearly 50 years ago. Ultimately, it can never be known exactly WHAT happened and exactly HOW things went down on the night of the crash, but perhaps a few tidbits of reasoned conjecture may be in evidence from our exercise.

One More Run...

In all our testing...(gee wiz folks, sorry)...we seemed to have neglected to film a video of the last few seconds of flight. Sheesh. So here it is. Unfortunately, the simulator model is not imbued with any "crash physics". Too bad too. Because it would help diagnose many questions about the geometry of the impact itself (a complex topic we may devote a future discussion to). Notice however how the #4 engine propeller touches the ground first. The belly of the fuselage has not impacted the grount yet. Simple physics dictate this would have thrown the aircraft into a clockwise rotation resulting in the "inverted' heading in which the wreck was found, and is consistent with thefuselage "nose bending", the tail breakage, and other damage evidence.

The video "pauses" briefly at the exact moment of impact to scan around the aircraft, so as to examine it's attitude in relation to the ground at this moment. It then continues into the ground, although without the clockwise cartwheel that would have been present in life. Most of the focus here is on on the flight gauges, sense of aircraft speed, and angle of impact, etc.....

LBG Impact Video

Qustions or comments about any of the information contained in the above report may be addressed to: edtruthan@gmail.com