Improvement of Automatic Route Generation

[Eefoe]

Hi,

I’m writing in regards to the automatic route generation feature that Simbrief has. Using the latest AIRAC from a Navigraph subscription, all the available routing options should be visible within Simbrief. Unfortunately, however, I still feel like this feature falls short in some ways:

---

NORTH AMERICA
Although the routing feature does generate routing well in general, the following shortfalls are commonly seen:

• Planning incorrect SIDs/STARs for the type of aircraft (turboprop vs turbojet)
• Planning incorrect SIDs/STARs for capable equipment (e.g., when flying a CRJ-700 into KDCA, the autoroute generator gave me the NUMMY3 arrival (conventional nav) instead of the FRDMM5 (RNAV), even though they transitioned from the same enroute waypoints and terminated to the same runways and approaches).
• Directs are too long: the TC AIM section 3.15.6.2 and FAA AIP Section 5.3 have limits for direct routes in their airspaces, including time limits and distance limits. As far as I can tell, these directs have been exceeded at times: One example is KDCA to KICT, where Simbrief planned a direct of 367 nautical miles from TOY to ICT:
  
  
• Directs can sometimes violate the minimum 1-waypoint-per FIR/UIR: The FAA has an Enroute Operation handbook where it says (on page 2-25) that Random RNAV routing should have at least 1 waypoint for each FIR/UIR it crosses, and it must be within 200NM of the preceding FIR/UIR’s border. Once again, I’ve seen that violated a few times.
• Direct routings in Mexico: As far as I understand, direct routings in Mexico are prohibited: all aircraft must remain on published airways. However, a generated route from KSAN to MMUN generated random RNAV routing within Mazatlan and Monterrey airspace. Ideally, this routing should have gone from LA FIR to Albequerque to Houston to Houston Oceanic to Merida, staying in US Airspace until the very end rather than cutting through Mexican airspace:
  
  

---

EUROPE
There seems to be much better handling here, despite all the weird funky rules in EU airspace. Nonetheless, I see room for improvement:

• I generated a random IFR route from LIRF to UUEE in a CJ4. Although this was well handled, the IFPS validation still produced an error for staying on upper airway routing above FL330 (which I believe is the vertical boundary for FRA).
• I also believe that it doesn’t take aircraft navigation capabilities into account, or aircraft type.
• I’m not sure if this is already a factor in flight planning, but the FRA restrictions can be time and weekday dependent (such as CDR1, CDR2, CDR3). I don’t see evidence of it being taken advantage of, though I could be wrong since I have no definitive demonstration.

---

COMMON
I regularly see airspace clipping. Although it may not be 100% forbidden in some regions (such as US airspace; flightplans won’t get rejected they’ll just get rerouted last second), it’s generally bad form to clip airspace (e.g. a 5 mile leg in the corner of a UIR). If it could be avoided, that would be awesome as well.

[SimBrief]

Hi, thanks for your feedback! We will keep it in mind whenever we work on the route generator in the future.

That being said, I couldn’t really reproduce some of your routes when I tried on my end. Can you confirm if you were using SimBrief’s “Route Finder” tool to generate these routes? Or were you perhaps simply looking through the “Suggested Routes” list?

Or maybe you were generating these routes through the Navigraph Charts software instead (which is where your screenshots seem to come from)? Navigraph Charts uses its own routing logic, independent from SimBrief, so that might explain the different results I’m seeing.

Also note that by default, SimBrief doesn’t actually “generate” routes when you enter departure/arrival airports. At first, the routes it suggests are pulled from a variety of databases (either real world North American routes from FlightAware, known good European routes, or previous routes used by other SimBrief users). Then a series of prioritization steps and filtering takes place to try and recommend the best routes out of those options.

Unfortunately, sometimes even the best route out of these options isn’t very good, which is where it can be beneficial to try generating a fresh route using SimBrief’s “Route Finder” tool instead. It can be found just below the route box on SimBrief’s options page (apologies if you already know all of this, just trying to be thorough):

.

Trying SimBrief’s route finder just now, the route it gave me for KDCA-KICT was:

REBLL5 OTTTO Q68 HVQ J78 IIU J112 FAM J98 SGF ICT

And for KSAN to MMUN, I got:

BRDR7 BROWS AXASA UT41 CUU UJ2 CAMUI UJ13 REX UJ10 MAM UT22 VESKO UT11 SASEN UT31 XUDUN XUDU1A

This isn’t to say that your feedback isn’t relevant. SimBrief’s route finder is far from perfect and can be improved in many areas, including the ones you brought up. I’m just not seeing the same results as you and I am trying to figure out why.

Thanks,

[Eefoe]

I’m not sure either. I can absolutely say for certain however that that’s the tool I was using, correct.

The KDCA - KICT route was actually a KIAD to KICT route, my bad. Here are the parameters:

And here’s the output (and yes that’s after selecting “Find Route”):

[Eefoe]

And further on, the KSAN to MMUN route generation…:

Avoiding the BRDR7 SID since it’s a conventional nav SID…

And the end result, once again forcing it to “Find Route” just to be sure:

[SimBrief]

Ah ok, so that explains the differences then.

I can’t say that there are any immediate plans to work on SimBrief’s route finder tool (if I’m honest it needs to be completely rebuilt at this point). But when we do work on it we will reference the feedback you’ve brought up here.

Since you also have access to Navigraph Charts, perhaps that route finder will have better results for you? I just checked KSAN-MMUN and KIAD-KICT and the routes it gave looked better to me. Once your route is created in Charts, you should be able to copy the route string by clicking the “Type route” button (for example, if you want to transfer it into SimBrief).

SimBrief’s focus/strength is more on route databases, suggesting real world or IFPS compliant routes for some of the more popular airport pairs, etc. But for pure route generating, I think the Navigraph Charts one is better at the moment.

Best regards,

[Eefoe]

I gave the Charts software a try, and though it does stick to upper airway routes, the KSAN-MMUN routing does start very bizarrely…enough to get any ATC (VATSIM or IRL) to loathe you:

That being said, the rest of the route does perform better overall in, though it still seems to have weird behavior around Monterrey, taking a more northern path instead of trying to maintain some affinity to flying directly overhead Monterrey airport to avoid congestion from climbing and descending traffic:

Meanwhile, the KIAD-KICT route I had shown earlier does do much better, though it sticks to upper airways rather than taking advantage of direct routings in the far less congested mid-continental United States to save fuel and time, rather stringently following the upper routes…:

Oh and of course, it absolutely fell short on IFPS compliant routing: I tried the LIRF-UUEE route and put it up for IFPS validation, and that route spat out too many errors to be considered acceptable; I think Charts just doesn’t recognize that Upper Italian airspace is FRA. My first post here was actually to the Charts forum, but I was told it wasn’t within scope and asked to post here instead for my Simbrief-specific questions.

I realize this isn’t within scope, though if you ever consider redoing the route generation system, I can tell you this would be a VERY welcome overhaul, especially to us simmers who constantly bounce back and forth between flying in Europe and North America and have trouble keeping up with all the bizarre flightplanning rules each country has.

[SimBrief]

Indeed! In time, hopefully the route finding can be improved further, but it’s also one of the more complicated things to program so it might be a while.

IFPS is a bit of a different animal, complicated by the fact that Eurocontrol does not permit access to their real world validation/routing services for flight sim use. And even Eurocontrol’s own systems can fail to suggest a valid route between airports, requiring human intervention at times.

Long story short, it will probably never be perfect, only progressively improved over time. But the plan is certainly to improve it when able.

Cheers,