VNAV Descent on the CRJ / DO-283 Compliance

[amahran 608]

Vertical Navigation is deficient and doesn't seem to recalculate the descent path when changes are made to the flightplan. It's possible to end up with descent profiles that have climb commands in them, and Tops of Descent that are after the first descent restriction. Generally, the CRJ as developed by Digital Aviation does not seem to comply with the standards of DO-283 Sec. 2.2.2.2

 

• There is no indication of Descent Path Construction based on geometric path boundaries during descent; the advisory VNAV will command one of the two limits of the next VNAV constraint in the flightplan.
  • General rules for Descent Path Construction:
    • The constructed descent path shall pass through AT altitude constraints
    • The constructed descent path is not required to pass though "AT or ABOVE" or "AT or BELOW" altitude constraints at the "AT" altitude, but the path shall satisfy the restriction
    • The constructed descent path shall pass between the "AT or ABOVE" and the "AT or BELOW" portions of the "WINDOW"
    • The constructed descent path shall also stay between the space described by connecting geometrically (constant barometric gradient) the "AT or ABOVE" portions of "WINDOW" and "AT" constraints, followed by connecting geometrically the "AT or BELOW" portions of "WINDOW" and "AT" constraints
  • These rules are taken verbatim from DO-283 Sec. 2.2.2.2.6.1
• I will also note that DO-283 also does reference equipment being able to parse a flight into four phases, which the CRJ has demonstrated that it is incapable of accurately parsing what is considered a descent. A good indication of this is climb commands being given during arrival procedures:
  • Climb
  • Cruise
  • Descent
  • Final Approach Segment 

 

I believe I've isolated it to an approach that contains a Discontinuity or Vector Leg: once VNAV has to calculate a path with that in the way, it seems that VNAV permanently breaks for the rest of the flight. The aircraft tends to handle Geometric-Point-to-Point VNAV very well (e.g., flight plans with no discontinuities and very strict altitude restrictions all the way to the runway). Outside of that, the performance is deficient.

 

I suggest the following test case:

• Set aircraft on the ground at KTEX
• Set up a Simbrief-imported flightplan at FL400 with the route: ETL DVC KRINA TTRUE LUCKI1 KSAN, using the RNAV Z 27 via LYNDI
• Depart and continue on flight
• When next leg is DVC -> KRINA and still not at cruising altitude, change the arrival from LUCKI1 to TPGUN2, using the ILS Z 09 via vectors.
• The aircraft will demonstrate failure to compute an accurate descent profile, and will require manual intervention.

 

I realize that this is only advisory VNAV, considering that the system does not provide an actual computed path and vertical deviation from the path (i.e., the snowflake only tells you a target vertical speed to maintain), but at the very least there's some kind of expectation that the snowflake should be reasonably reliable enough to depend on in terminal operations.

 

Any CRJ operators that wanna chime in about their experience with the snowflake when it comes to "WINDOW" or "AT or ABOVE" or "AT or BELOW" geometric descent paths?