Flying in FSX with AI gliders based on replayed IGC log files


Soaring over the Victorian Alps – Aerosoft Discus X in FSX

I was contacted recently by another flight simmer that is interested in virtual gliding in FSX. He goes by the on-line name of harryosh and he described a very interesting way to fly in FSX against real-world flight logs (IGC files) while using an approximation of real-world weather injected by Active Sky Next (ASN). As you can see in some of my YouTube videos, I have some doubts about achieving an effective soaring environment using ASN with CumulusX!. However, harryosh has provided a good description of how this combination can work in the right circumstances and has also provided a sample task on his blog at:

So I will not attempt to describe the process in detail because it is well covered on harryosh’s blog at the link above. The sample task is a real-world cross-country flight flown by Mark Bland on the 18th December 2014 at Mount Beauty in North Eastern Victoria, Australia. I downloaded the task using the link provided by harryosh and installed it and set it up with ASN and CumulusX! as described.  harryosh included his own IGC log in the task package giving me the opportunity fly with both recorded IGC logs.  My first effort with flying against both IGC log files in FSX with weather provide by ASN was successful in one sense but, as is often that case with any first attempt of anything, I made a bunch of stuff ups. Nevertheless, the good news is that the task package provided by harryosh worked as per the instructions on his blog.

For this first trial flight I decided to fly the Discus B (Aerosoft Discus X) against Mark’s ASK21 because Mark Bland is the CFI at Mount Beauty and knows a heck of lot more about flying there and about cross-country soaring in general. Actually my first couple of real-world flights were at Mount Beauty a few years back and I have relatives and friends living in that area, so I do at least have some knowledge of the geography associated with this cross-country task. So this area of Australia is my favourite for cross-country gliding in FSX, particularly with the Orbx FTX Australian scenery and the OzX Mount Beauty scenery.

I set up to launch 5 minutes ahead of Marks’ launch as suggested by harryosh. Unfortunately I got worried that I was going to be left behind when I saw harryosh’s glider manoeuvring to winch launch, or so it appeared. So I immediately launched on the winch and released at about 19oo feet AGL in the hope that I would be able to stay in touch with harryosh’s glider and then I could follow both gliders after Mark had launched.

Initially I mucked about looking for lift and tried to climb a lumpy blue thermal near the airfield but I could not see any CuX clouds within reach. Moreover I could not see any other gliders nearby but I didn’t realise at the time that my labels were cream coloured instead of red so I probably would not have noticed them unless they were close enough to see the airframe clearly. At one stage I caught a glimpse of a glider still on the ground, which I assumed was Mark preparing to launch. Also in my rush to launch I did not think to startup Plan-G, which would have helped me locate the other aircraft in the air.

Because the thermalling conditions were so poor near the airfield, I decided to press on with the task with the I intention to wait for the two other gliders to catch up once I got up near cloud base – but that wasn’t to be. I struggled up into the mountains towards the first turn point without finding any thermals and eventually ended up a long way off the course track and inside the Mount Hotham airspace. I struggled to get onto the ridge tops and all I could find was patchy ridge lift.  My path through the mountains was eventually blocked by a high ridge and I was forced to turn back towards Mount Beauty.  Feeling very despondent at this point, I reluctantly decided that I should abandon this task altogether and just be satisfied making a safe return to the airfield. However, in the desperate struggle to get back to the airfield I blundered into a gusty blue thermal and managed by repeated dolphining manoeuvrings into the wind to claw my way back up to a decent height. So I headed back towards the first turn point again with a little bit more confidence, although I still doubted that I could complete the whole task given that were no visible thermal clouds. Having flown so far off course and wasted so much time trying to get out of those mountains, I assumed that I must be miles behind the other two gliders. So I pressed on as hard as I dared without stopping to climb very often. I tried to make full use of the ridge lift since there didn’t seem to be any CuX clouds anywhere on the course. For me this par for the course when I combine ASN with CumulusX!. I only slowed down to dolphin through patches of lift and went out of my way to stay on the windward side of any ridges or any decent sized hills even though that meant adding a lot of distance to the task. Moreover, after the bad experience of getting trapped in those mountains on the first leg, I decided to maintain a much higher working set height than usual, which necessitated maintaining a fairly conservative speed on the cruise segments. Nevertheless, I was still trying darned hard to catch up to the other two gliders for the remainder of this task.

I miss-judged the final glide badly and arrived home at the finish line wastefully high. Also I had my aircraft realism settings enabled in FSX so I was restricted by the turbulent and gusty conditions on how fast I could fly on that final glide. I could not risk ruining this long task at this late stage by over stressing my glider. Needless to say I was very despondent on that final glide, believing that I must be falling even further behind the other two gliders, which I had not sighted in the air at any stage after launching. i even began to doubt that I had started the recorded logfiles in sim_logger correctly.


Cross-wind landing at Mount Beauty – Aerosoft Discus X in FSX

After landing and saving my log file, I noticed that the other gliders were not on the airfield. At first I thought that their replaying log files must have terminated already. However, I started up Plan-G and was shocked to see that the other two gliders were still airborne and only now approaching the second last TP – so apparently they must had been behind me all the way or else at some point I must have passed without sighting them. Clearly I should have waited for them above the airfield regardless of the poor thermalling conditions and then stuck close by as originally intended. Also in hindsight I probably should have been flying a slower glider than the Discus B against an ASK21 even though I am a poor virtual cross-country pilot. Moreover, the ASN weather was suspect in that it may have been providing over-generous ridge lift for my flight. I seemed to be getting sufficient lift to maintain height several thousand feet above the ridge tops most of the way around the course, which is unusual although the upper level wind was blowing around 18 knots. I am not sure that that is realistic compared to the real-world weather on that day and I was expecting to see more thermals. When I compared my log file with Mark’s I could see that he was needing to take time out to climb thermals a lot more than I needed to, and even the difference in performance between an ASK21 and my Discus B does not completely account for that.

Another minor issue with flying against a real-world IGC log file is that the height recorded barometrically may be different from the height captured by the IGC in FSX.  In the case of this specific task, the height of Mark’s aircraft was about 200 feet above my aircraft when parked on the ground. I would assume that this discrepancy  would be less noticeable once both aircraft are airborne although in this case it means that I would have to fly 200 feet higher to see Mark’s aircraft at the same level.  Also both of the AI aircraft produced by replaying the log files using sim_logger were rendered as DG808 gliders. I am not sure if that can be changed or not.

In spite of  few issues, it was an interesting task and this method for racing against real-world IGC logs is definitely worth some more testing.

Here is the Dropbox link to my IGC log file for this task and the Cambridge Aero Explorer analysis for this task based on a 10k start line, a 5k finish line and 500 metre diameter TP cylinders (with no height limits) is as follows:

File Name    d-6945_olc – mount beauty 18-12-2014_2014-12-26_1954.igc
Date    December 18, 2014
Pilot    James Allen
Glider ID    D-6945
Glider Type    DiscusB TT
Comp ID    D-6945
Number of Fixes    2588
Total Flight Time    02:50:36
Task Status    Completed Task
Distance    297.08 Km
Time On Course    02:36:57
Speed    113.57 Km/Hr
Roll Time    14 : 40 : 45
Start Time    14 : 46 : 58
Finish Time    17 : 23 : 55
Landing Time    17 : 31 : 21
Acheived Turnpoint ID’s    2,3,4,5,6

2 thoughts on “Flying in FSX with AI gliders based on replayed IGC log files

  1. Harry O

    Really enjoyed the read James thanks for the effort! I’ve loaded up your trace of the flight. Your initial efforts off the winch would have been a challenge! Yes the lift around the airport does seem sparse doesn’t it! I’ll have to tell you the truth. I already knew that there was a thermal near Mount Beauty off the launch. It is in the valley that leads to Bogong. I had already flown the course unsuccessfully in an ASK-21 the day before and the thermals are in the same place :-) That made it a lot easier for me! I think that off the launch, considering you didn’t have that benefit of hindsight, if you would have picked the ridge line just one further west (which is what the real world pilot did) you would have nailed it. It all comes down to a split second decision and all the ridges look so similar in the simulator!

    The weather you got looks to be very similar to the weather I got. Did you see the CumulusX thermals at the north of the course near Lake Hume? They were the only thermals that were not blue on the whole course. I actually had to use them to keep up with the real world pilot.

    The main thing is that it was enjoyable!

    I have the tail numbers labels in red and I can switch them off or on with a hat switch. That makes it easier. Amazing that Plan-G can trace the Simlogger planes. Didn’t know that!

    Thanks again.

    1. James Allen Post author

      I did not notice any CumulusX! clouds up at the northern end of the course, although I do recall seeing a line of distant FSX cumulus clouds in the far distance towards Albury/Wadonga. There were also a few small CumulusX! clouds out near Mount Hotham at the other end of the course but I couldn’t reach them at the time that I needed them – such is the nature of mountain soaring.

      Mark seemed to find thermals in the real-world at numerous locations around the course – he even managed to thermal up to 10,000 feet at one stage or perhaps he found some wave lift. On the whole I don’t think ASN in conjunction with CumulusX! was producing representative real-world weather on that day. Of course we don’t know if Mark was finding cloud based thermals or blue thermals but I suspect that CumulusX!’s pedantic insistence on removing all thermals in the presence of even a scattered stratus layer may have been the culprit in this case. Either that or I somehow missed most of the blue thermals – maybe some of that patchy ridge lift I was experiencing was actually blue thermals attached to the windward slopes of the ridges.



Leave a Reply

Your email address will not be published. Required fields are marked *