A conclusion related with SVMI crosswind runways activation

[jalsina]

Since the last time we exchanged messages in the FSXPlanner forum,
http://www.fsdeveloper.com/forum/showthread.php?t=5729
I have made a lot of testing and arrived to a conclusion:

The first testing session was related with creating more than one fake runway between 09 and 10 and trying even runways beyond 10 and 9. I tried up to five additional fakes.
As I mentioned in that thread, I also tried activating runway 09 for both take off and landing in primary and secondary, as suggested by rfields.

The result was always the same. Most of the departing traffic was diverted to 09, and a particular ramp sent all their planes to 10. Only a few airplanes (10%-20%) in the gates were sent to 10.

The obtained effect is exactly the contrary of the real SVMI operation.

Later on, I displaced 09 closer to 10 (more to the north), by conclusions taken just watching Google Map and the airport diagram from the terminal procedures.

This last decision worsened the situation still more. About 20% of the ramp airplanes were diverted to 09 and all the other gates (100%) were sent to 09 as well.

Today I started another testing session, with several situations where I moved 09 around the airport, moving it south of its position, further from the gates, north of 10, crossed with 10, etc.

The definitive conclusion is that FSX ATC is sending the departing aircraft to the runway which AXIS (not rwy center or start point) is closest to the original ramp/gate of the particular aircraft. When the distance is similar, the airplanes has a chance to go to either 09 or 10.

I hope others can confirm what I am saying here, and this is not just an issue in SVMI. I will check the works by others to watch the ATC behaviour.

Jorge

 

[jvile]

Jorge

What you are seeing is not a new problem in FSX but has always been a issue with FS9.

Contrary to popular belief, AI Planes do not always taxi to the closest runway based on the center LAT/LON of that runway.

Draw an imaginary rectangle around the runway. If you have a hard time with the visualization then use a apron and place it all the way around a runway. Extend it out on both sides of the runway about 1500 ft and 4000 ft on each end. Another way to visualize the box is make runway 09 about 8000 ft longer (18092 long) and 3000 ft wide. Now look and see what parking spots are inside the rectangle of the runway. You probably only have about 20 percent parking spots left for RWY 10. When you moved RWY 09 North you put more parking spots into the bounding rectangle box and took departures away from RWY 10.

We will call this the runway bounding rectangle box. This box becomes one of the prerequisite for runway assignment if the parking spot sits inside the box. There is a pecking order of how ATC assigns runways for takeoff and all the varibles must be looked at and evaluated.

Other varibles that effect ATC runway selection for takeoff is the location of the Terminal Building in respect to the parallel runways. Does the Main Terminal set between the parallel runways are to one side of both runways. This has to do with the runway overlay technique that we used in FS2002 with the AFCAD program which MS now uses in FS9/FSX by default (when parallel) as a hard code.

When you parallel 2 runways with a 7.9 degree fake runway this also has an effect on runway selection at some airports. Most default airports use 2, 3, 4 or 5 parallels which are all within 7.9 degrees of each other. The Runway bounding box is not a concern except at airports like KDEN, EHAM, KIAH, etc, and of course any airport you force into a xwind technique.

When runways are parallel by default or forced into parallel (xwind), runway length vs Aircraft.cfg/Empty Weight value becomes more important to what runway a AI Plane is assigned for takeoff.

Reggie eluted to runway length vs Aircraft weight in the other thread.

If you have my KATL FSX airport from AVSIM you will notice that I control which runway out of all 5 parallel that ATC assigns the AI Planes to. This is done with the length of the runway and weight of the plane.

Even though we ask for the base runway end to be listed first (9/27, nn/nn, 10/28) with the xwind runway between the 2, at some airports we have to place the receipt end first (27/9, nn/nn, 28/10) in order to achieve a more realistic takeoff runway assignment.

There are other minor factors for runway assignment and all the above issues I am pointing out must be considered.

When you parallel a set of runways using the xwind technique do not assume that AI Planes are going to be assigned what might visually appear to be the closest runway to the parking spot. When you add trickery and break the rules of FS then you now must adhere to the new rules and test all the varibles that play a part in runway assignment.

hope some of this is useful.

 

[jalsina]

Jim,

There is not doubt in my mind that the runway centers theory is discarded. However I believe my "extended axis" theory prevails over your "boundary" theory, at least to what SVMI refers.

Take a look at SVMI.jpg (left) to understand what is going on. All aircraft from ramps A+B are being dispatched to 10. The remaining gates and the parking southeast are being sent to 09.

This diagram shows the latest situation of my SVMI. Previously 09 heading was nearly 4 degrees higher. By fixing the hdg I increased the influence of 09 over SVMI gates and ramps.

Before about 20% of the G gates (international terminal) went to 10 as well. Now after the 09 hdg fix all gates aircraft (100%) go to 09.

Ramps A +b remain going to 10.

The other uploaded image SVMI3.jpg (right) shows the suggested boundaries over 09 and 10. As you can see ramps A+B fall into both boundaries. However every aircraft there , as I said, goes to 10.

Also you can see that the most western part of gate G is not inside any of the two noundaries, but is closer to the 09 axis. All these airplanes keep going to 09.

Besides yesterday's testing session in which I displaced 09 all around the airport, in some situations the theory of extended axis seemed to prevail over boundary influences. I can't demosntrate this graphicaly since I deleted the different FSXPlanner files for the tests.

Last night after I posted this thread, I went to your KJAX, which has about the same traffic density of SVMI. While the traffic routing in Jacksonvilles is more straightforward, it totatly discards the rwy centers general believe. After watching the traffic for about an hour, it seemed to me that the runway axis influenced all departing flights to be routed to the closer aircraft.

I have your KATL installed, which by the way it's a great well done job. But I have a problem now with all large US airports because I installed Ultimate Terrain X with "all whistles and bells", so the performance in some of these airports can be dramatic. I have to tweak UTerrX before I can go further in all US. Besides the flights over US rural terrain needs much more time to load. It seems the landclasses textures are quite heavy.

Jorge

 

[jvile]

Jorge

At some airports the bounding runway box is not always identical and the Rectangle is a starting point only to help visualize what is happening.

Reggie has often reported that some runways use a triangle at the end of the runway for the bounding box effect.

We do know that the center line of a runway extends past the threshold at different lengths and every airport where we observe runway assignment behavior seems to use different box values.

KDEN differs from KIAH when monitoring how ATC assigns runways for takeoff based on the bounding box but what they all have in common is the center line of a runway takes precedence.

Make my rectangle more narrow and not has long and draw a triangle at the end of runway 09 centering the rop of the triangle on the center line. See if that gives you a visual of why some Parking spots use RWY 9 over RWY 10.

I am not clear yet on the axis theory of the runway but would like to pusue that a little further to see if it fits better at other airports.

 

[jalsina]

I'll keep trying with this. I can see by your explanations this is much more complex than I thought (and that should be).

In the meanwhile I am busy fighting now to increase the general aviation traffic at SVMI (and other country airports). I can't see more than two aircraft simultaneously with ruler set at 100%.

I wonder if there is something I could do to increase it from FSX.

Last night I tried the GA Traffic Generator which conflicts with my Ultimate Traffic, so I can't see the added traffic unless I remove the general traffic.bgl.

Jorge

 

[rfields]

Jorge,

I think you are seeing impact of both your concept of centerline and Jim's bounding box.

First one clarification which Jim discovered a while back - it's not the location of the parking spot that is used to determine takeoff runway selection - but the location of the aircraft when it is pushed back and ask for taxi instructions. In your A+B area that makes a difference.

I think we all understand Rwy 9/27 will be for takeoff of aircraft in the Cargo + Ramp area.

If you look at the parking spots, the nodes and how the aircraft push back - the parking spots for the G and the H areas are in both runway bounding boxes - but closer to the Rwy 9/27 centerline than the Rwy 10/28 centerline.

However the aircraft in the B and the A+B area will be closer to the Rwy 10/28 centerline.

Now - let's get your performance goals complete:

1. All jet takeoffs on Rwy 10/28

2. All turboprop takeoffs on Rwy 10/28

3. All cargo aircraft takeoffs on Rwy 10/28

4. GA and small piston takeoffs on Rwy 9/27

The way to accomplish this is to make Rwy 9/27 4992 feet long (1521.6M) and set the runway to be between Taxiway E and Taxiway M. Move the start locations to those positons.

Change the runway designator to 9A - turn off all markings and lights.

Change the link lines which are no longer on the runway to Taxiways with no lights, markings or designators.

There needs to be a break in the replacement taxiway between Taxiway J and Taxiway E so aircraft won't try to taxi up that 'runway'

Create a new runway 09 with the exact dimensions of the original runway set this runway to shows the designators, markings and lighting you want - place it over the old shortened runway in the original location. This new runway must be closed for takeoff and landing on both ends.

That's the approach I would use.

(BTW - are the default FSX Taxiway signs for that runway wrong - ADE reports them as saying 8/26)

 

[jalsina]

You two guys have an extraordinary knowledge of all this traffic/afcad subjects. Thanks for helping me so much. I'm learning a lot.

I will answer the final part first. I have modified the airport since I posted the XML/BGL. I am attaching current files to this message, so you can take a look to it. The fake runway in the XML has already the latitude conversion.

The major difference is a rotation of rwy 09 by 4 degrees (according to availble data). But I have re-made many details of the airport as taxiway widths, new paths, added parking spots and also fixed all signs including the wrong ones you mention. I could get the 09 hdg back to the FSX original heading to dispatch more airplanes to 10.

The ramp/cargo area south of rwy9 is not parallel to the runway as it should be. I'll change that in the near future. I have still to work a lot in that ramp. For the time being I have added 11 remote parking positions for GA (but GA aircraft are not available... It seems it's a problem related with FSX flightplans in which I will have to work...for the full country).

You are right, most of the real traffic parking in that area take-off in rwy09. However the cargo heavies are always sent to rwy10. You can see frequently such planes as DC-10 and 747-200. However I don't see a problem cargo take-off in FSX by using 09.

What you mention about the boundaries and the axis solutions intersecting could be right. My puzzle is the remote area in gates G. After all the final pushback point is closer to 10 and further from 9.

Jorge,

........

Now - let's get your performance goals complete:

1. All jet takeoffs on Rwy 10/28

2. All turboprop takeoffs on Rwy 10/28

3. All cargo aircraft takeoffs on Rwy 10/28

4. GA and small piston takeoffs on Rwy 9/27

Actually these are not my goals. I think I mentioned that 09 is a runway to alleviate the take-off traffic in 10.

Most of the real parked traffic at A+B is sent to to 09. And some of the local airliners at Gate H (southeastern part of main ramp), are also sent there. You seldom see an international airliner depart via 09.

However yours is a solution I have considered since you mentioned it in another thread. This solution would be interesting if there were a middle position. I mean the turboprops could be sento 09 and maybe some small jets.

The other solution would be displacing the runways somewhat to south and get the 09 hdg back to the FSX default. Or closing the international gates a little bit to north. That would increase the traffic from the international terminal

The way to accomplish this is to make Rwy 9/27 4992 feet long (1521.6M) and set the runway to be between Taxiway E and Taxiway M. Move the start locations to those positons.

Change the runway designator to 9A - turn off all markings and lights.

Change the link lines which are no longer on the runway to Taxiways with no lights, markings or designators.

There needs to be a break in the replacement taxiway between Taxiway J and Taxiway E so aircraft won't try to taxi up that 'runway'

Create a new runway 09 with the exact dimensions of the original runway set this runway to shows the designators, markings and lighting you want - place it over the old shortened runway in the original location. This new runway must be closed for takeoff and landing on both ends.

That's the approach I would use.

(BTW - are the default FSX Taxiway signs for that runway wrong - ADE reports them as saying 8/26)

I like much the way you put this. I really will take a look at this solution, which should be better if I have some GA traffic enabled (only two aircraft are acheduled now).

There is something though. I believe you mentioned 7000 ft in the other thread, instead of 4992ft. Is there a middle length for other traffic size?

Jorge

 

[jalsina]

I just tried to shorten 09 length, but I started at 7000 ft.
All heavies but one, (a 757 probably not defined as a heavy), departed via rwy 10.

So it seems there are middle positions after all.

Combining your solution with some changes could make this airport work a little bit better.
I will try shorter lengths to see what happens.

Jorge

 

[jvile]

Jorge


Some AI Planes do not have a correct Empty Weight in their Aircraft.cfg file.

Look at the EW of the B757 and it is probably somewhere around 74,000 lbs.

 

[rfields]

You may be seeing one of FS AI aircraft modeler dirty little secrets - aircraft empty weight.

B752 - 74,170 lbs - vs 128,500
B738 - 80,170 lbs - vs 93,500
A320 - 80,200 lbs - vs 93,000
B747 - 102,170 lbs - vs 399,500
DC-10 - 200,000 lbs - vs 268,000

These weights make it easier to play with the flight dynamics - but it makes the way FS needs aircraft empty weight to assign runways not work in many cases.

There is a formula for determining aircraft takeoff runway length vs empty weight - but when the PAI forums were moved - this very important work got lost.

Play with the runway length to get where you need to be.

Don't change an AI aircraft empty weight - it will cause major problems.

I'm using Jan Martin's FD sets rather than the 'ultra-realistic' FD which the 'best designers' package with their aircraft.

 

[jalsina]

Finally, this seems to be the right solution, given the behaviour of the sim. Keep reading.

It won't be necessary to tweak aircraft weights, but I was wondering what I had to look for in the aircraft.cfg or even in the .air flight dynamics.
Actually, Jim, that particular 757-200 (by AIA) is set for an empty_weight = 74170.
Another 767 behaved like that after I posted (curiously the same airline but different designer) when the rwy was set at 7000 ft.

(BTW, will not be this weight the reason for shorter tahn real landing runs in some small jets like MD80s or DC9?)

However, I tried different lengths between 5000 and 7000ft, and after watching the behaviour at a departures peak hour I am leaving it at 6.300 ft. The rwy length is no doubt proportional for an airliner to be routed there (still influenced for what we have been discussing before).

With this runway 09 length, all heavies go to 10 including those rebel 757 and 767. Most of the small aircraft from the international gates (west), like the A320 and B737 are being sent 80-90% to 10 and very few to 09.

Local flights from gates mostly go to 09 but a few of them go to 10. Some A+B ramps planes even go to 09. Most keep going to 10 (so there is still the influence of boundaries).

I expect big cargoes to take the 10 even coming from south ramp (I have not tested yet).

What I need to do is going to the master file in FSXPlanner and build the intersected 09 runways as rfields told me to do.

And test the full thing with GA if I get some planes to fly from/to SVMI.

So this is it (or mostly). I have seen a few message here and there with people that seem to have about the same help I had from you both. I guess they are reading these messages.

What we need is Russell to implement the fake runways in FSXPlanner, so it saves without further memory errors when the flight is started.

A big thank you, to both of you.

Jorge

 

[Phantoms]

Jorge, there a good older read on FS at http://www.flightsimdownloads.com/pub/FlightDynamics.pdf

Note this was for a few versions ago, but includes some good info.

 

[jalsina]

Got it James,

Thank you.