Well I haven't dropped off the face of the earth, but I was swamped with client work in the last quarter of the year, thankfully now subsided. I have continued researching the Hump airlift, bought a couple of excellent books on the subject, reread General Tunner's account, and did a "Big Day" play test.
Thanks to Mako's suggestions, I learned how to make a serviceable hex-grid map in Word, using different fill colors to indicate different elevations, and also did some counter sheets in square-grid for the various aircraft types and game markers.
I recently had a chance to set up my game "kit" and play out a simulation of "Big Day," 1 August 1945, and I have to tell you it is pretty magnificent to see 500 aircraft in flight across the Himalayas hauling the cargo over to China and coming back to India for another round trip! It would look even better with miniatures on a big table, but that is a challenge for another day!
The Big Day test pulled together all of the various aspects of aircraft characteristics, turn-around sequences, command point allocations, accident rates (pretty low at this period of the war), low hump and high hump, Assam bases and Bengal bases; it was all there, except at this point there were no Japanese interceptions, so the dice roll for interceptions was excluded.
There is some mystery to me about how to set up the scenario for Big Day. I used the number of aircraft as provided in the order of battle for July 1945 as 230 C-46, 167 C-47, 132 C-54, and 67 C-87/109, and then applied a maximum effort "one-day" Efficiency rating of 0.9, one tenth more than Tunner's normal maximum (actual daily availability around this time period was 0.75), to reflect scheduling 50 and 100 hour inspections to occur before or after Big Day, while also keeping up the normal daily delivery quotas in the days before and after Big Day. Rather than dicing it for availability, I just used 90% of each type for a flying order of battle for the 1 August 1945 scenario of 27 C-46, 150 C-47, 118 C-54, and 60 C-87/109.
In my research, I haven't yet found the "scoring rules" for how Big Day performance was determined, so I set it up to count the number of tons landed in China beginning at 0001 and finishing at 2359 hours on 1 August 1945, rather than flights taking off during the 24 hour period and counting the tonnage until the last of these flights had landed. As the airlift was in 24-hour operation during this period that meant that planes that landed in China shortly after midnight had already been in flight for over three hours departing their Assam and Bengal bases in the eight o'clock hour the preceding evening. So, I more or less went from a standing start at about 2030, and surged the airlift from there. It turned out this may have been somewhat inaccurate as I delivered more tonnage than historically, and what I perhaps should have done was more or less randomly placed aircraft from "about to land" in China, backtracking additional aircraft back to their bases, and then lining up the planes still on the ground for take-off after midnight, so that throughput might be reduced to more historically accurate results.
It also isn't so clear to me yet where all the planes were based. For my play test, I put all the C-54s in Bengal bases to fly out over the low hump because C-54s are unable to fly loaded at the elevations required for the high hump, and all the rest in Assam to fly out over the high hump. This makes for some congestion in Assam, although this becomes somewhat regulated as time goes on by the congestion in Kunming in turning around the aircraft for the return flights.
Cargo loads were assumed as follows: C-47, 2.5 tons; C-46, 4.0 tons; C-87, 4.0 tons; C-54, 7.0 tons (Tunner gives C-54 as 6 tons, Craven and Cate gives 7 tons, and I used 7 tons for this play test). During this time period the loss rate was about 1 loss per thousand flights east, and I used the loss rate as 1/1000, or three snake eyes (0) on three ten-sided dice. The loss check was as the aircraft first crossed into the Himalayan mountain range.
I played this out using counters, map, and dice, unlike the earlier play tests that were done on spread sheets, with either dice or random number generators in Excel as described in early posts in August and September 2011. I liked playing it out on the game table much better than on spread sheets, the only disadvantage was that it is harder to keep track of the number of flights without meticulous manual log keeping, which I didn't want to do … With the spread sheet the game is a meticulous log, but playing it on a table top doesn't fully capture the numbers of flights unless you deliberately do that. The tonnage delivered is accurately recorded on a track printed right on the map, but flight by flight data is not, although it could be done with log sheets, about 10 pages with 50 aircraft per page in two columns.
So to avoid keeping you in suspense any longer, what were the results? Well, first the actual airlift in 1945 included 1,118 round trips and delivered 5,327 tons, or 4.76 tons per aircraft, without losing a single aircraft. In my play test, the average static lift tonnage per aircraft across the whole fleet was 4.2 tons, and at the end of the 24 hours more than 5,800 tons were delivered, and without losing a single aircraft! The historical account indicated that "scores" of aircraft made three round trips, and in my play test the C-87 fleet and the C-54 fleet all made three round trips, too many compared to historical performance. C-87s with higher speed and the same lift tonnage as C-46s were given take-off priority over C-46s and C-47s, and C-46s with more tonnage than C-47s were given take off priority over C-47s. Once in Kunming, where there was a lot of turn-around congestion, C-54s with the largest tonnage and a much longer flight to get back to Bengal for another trip, were given priority over all other types. The average per aircraft delivered would have been higher than the fleet static load average of 4.2 tons, because the smaller C-47s made fewer deliveries than the larger aircraft. If I had maintained an accurate flight by flight log, I could provide the actual delivered average, but alas I didn't keep a log and even my turn by turn notes are not amenable toward determining the number of flights and deliveries by type … maybe next time …
Given the loading throughput and take-off mechanics and constraints, as well as take-off priorities, there were a lot of C-47s sitting around waiting for a take-off spot, and the last C-47 made its first take-off from Assam at 0300, almost seven hours after the surge began, and just before the first C-87s were about to land back in Assam for their second round trip. I haven't seen any anecdotes that this long lead time for C-47s occurred, but there must have been some congestion in the system, otherwise historically every plane would have made three round trips, and there would have been more than 1,500 round trips and more than 7,000 tons lifted. The last of the C-47s out from Assam landed in China at 0640, and many remained there most of the day until they could get a few take off spots in between C-54s, C-87s, and C-46s. The first of the second arrivals came in soon after that at 0700 as the fast flying C-87s came in right behind the tail end C-47s (fast being a relative term in WWII airlift, with average speed over the ground at about 180 mph for C-87s and C-54s compared to about 150 mph for C-46s and C-47s).
So, it seems that I'm pretty close to the actual parameters although a little too high, and may tweak these a little more before another Big Day play test. Among the adjustments, I think I'll reduce the Tunner Big Day Efficiency factor to 0.8, and make the normal Tunner Efficiency factor as 0.7 (compared to approximately 0.75 aircraft availability during this period). Also, I think I'll add in a third load stage on the India side, so that it takes at least a half hour to fully load an eastbound flight, plus as many efficiency delays as provided by the 0.8 final load chance, usually not many in the play test with the 0.9 factor. It seems to me that loading up a full load, taking care to balance the weight and secure the cargo against shifting in flight, should take longer on the India side than rapid unloading and draining down the fuel to the allowed return fuel load limit on the China side.
I'm also toying with a Take-Off stage throughput factor. Currently each of the two mandatory load stages takes 10 minutes with a maximum of 10 aircraft going through each stage, and upon a successful Efficiency die roll, the aircraft is cleared for take-off and can take off immediately up to a limit of 10 take-offs per 10 minute turn. With two mandatory load stages followed by a successful die roll for the risk stage a plane can take off 20 minutes after the turn that it landed (the third turn after landing, as in Land, Load 1, Load 2, Take-Off if successful roll; for an on the ground elapsed time of 20 minutes). The fastest turnaround mentioned by Tunner in his autobiographical account is 12 minutes which I have rounded up to 20 … Tunner didn't mention if this was on the loading or unloading end and I have assumed, pending further research, that it was unloading rapidly in China and then getting back in the air. Adding a third mandatory load stage will slow down the India end, and a Take-Off factor might further regulate throughput down toward a more historical result.
The Take-Off factor might be thought of as a local weather factor, where the fields might be "socked in" occasionally to greater and lesser degrees over a 10 minute take-off window. During the Big Day period, it was pretty much all weather, around the clock flying, but there could be local weather effects hindering 10 take-offs in 10 minutes. My first attempt at this Take-Off factor will be to roll two 10-sided dice and take the highest of the two as the take-off limit for that turn. During Monsoon season subtract one from the highest die and use that as the factor. Big Day is right in the middle of the Monsoon season, but that didn't seem to be much of a hindrance at the time from the research that I have done to date.
If anyone has any insight on aircraft availability, turn-around times, or take-off constraints, I would be happy to hear about it, and also any comments or reactions you may have to the evolution of this simulation as reported back in August and September 2011 and today.
So there we are for now, a few more play tests to run, maps to smooth up a little, fine tuning some representative scenarios over the course of the 3-1/2 year airlift operation, and a few adjustments in the factors, and I should soon be ready to produce a smooth version in Word, make it into a PDF file, and then post it for others to sample.
This is not a particularly exciting air simulation, and in fact it is a lot like playing the old Avalon Hill Dispatcher railroad game of the early 1960s, never a big hitter on the popularity parade, but it should provide some insights into the trials and tribulations of those intrepid aviators who flew some of the most dangerous missions in World War II.
I have been having a ton of fun playing around with all of this, and in the fullness of time, I hope that some of you might also enjoy flying a few missions Over the Hump!
Enjoy your games!
Jan
paints up some WWII Soviet aircraft.