Another View of CANPA (Or, How I Learned to Appreciate the Stabilized Descent)
GUEST COMMENTARY. In his Pelican's Perch columns #24 and #25, AVweb's John Deakin took the position that constant-angle non-precision approaches (CANPAs) may not be appropriate for use by other than glass-cockpit aircraft with sophisticated VNAV automation. Captain Erik Reed-Mohn of Scandanavian Airlines System (SAS) differs with that view, based on both his airline's experience flying such approaches in a wide range of transport jets from older round-dial DC-9s to the latest-generation glass-cockpit Boeings, as well as his participation in studies and working groups of the Flight Safety Foundation. The author illustrates his commentary with a number of actual approach procedures flown by SAS.
This article is written in response to Mr. John Deakin's Pelican's Perch columns #24 and #25 concerning Constant-Angle Non-Precision Approaches (CANPA), in which Mr. Deakin takes the position that such stabilized non-precision approaches may not be appropriate for use by other than glass-cockpit aircraft with VNAV automation. I differ with that view.
The concept of a stabilized constant-angle non-precision approach may be new to most of us, but has in reality been used by some airlines for decades. My first encounter with the concept came during my work on the Flight Safety Foundation's (FSF) Controlled Flight Into Terrain (CFIT) Task Force, and its successor, the Approach and Landing Accident Reduction (ALAR) Task Force. Both the Flight Crew Training and Procedures Working Group of the CFIT Task Force and the Operations and Training Working Group of the ALAR Task Force came up with unequivocal support of the CANPA concept.
Having flown non-precision approaches the step-down way since my Air Force days in the 70s, I must admit to an initial reluctance. But seeing the way the statistics work out, with a concentration of wrecks along the extended centerlines of runways, I did not need very much convincing that traditional dive-and-drive (DaD) approaches were less than optimum. (See also Mr. Dick Slatter's article "Chart Design Revision Could Enhance Safety of Non-Precision Approach and Landing Operations", ICAO Journal May 1994.)
The clincher, as if that was needed, came with two reports published by the FSF:
- Airport Safety: A Study of Accidents and Available
Approach-and-Landing Aids. (Flight Safety Digest, March 1996.)
Two of this report's conclusions were:
- "On a worldwide basis, there appears to have been a five-fold increase
in accident risk among commercial aircraft flying nonprecision approaches
compared with those flying precision approaches."
- "When stratified by ICAO region, the risk increase associated with
flying nonprecision approaches compared with those flying precision
approaches ranged from three-fold to almost eight-fold, depending on the
- "On a worldwide basis, there appears to have been a five-fold increase in accident risk among commercial aircraft flying nonprecision approaches compared with those flying precision approaches."
- An Analysis of Controlled-Flight-into-Terrain Accidents of Commercial Operators, 1988 through 1994. (Flight Safety Digest, April-May 1996.) This report showed that approximately 60% of CFIT accidents occurred within five nautical miles (NM) from the threshold.
Mr. Don Bateman's statistics show that it is primarily navigation error in the vertical plane which is responsible for CFIT accidents, since CFIT accidents form a cluster along the runway centerline.
The point of trying to do something about these sad numbers was thus born out of a desire to find a safer way of managing the vertical path during a non-precision approach. The horizontal path seems to present fewer problems if the numbers are to be believed, and I think they should be.
Not Just for Glass Cockpits
Mr. John Deakin seems to equate CANPA with VNAV or automated aircraft. While they can certainly fly CANPAs too, this is a common misunderstanding that needs correction.
As mentioned, some carriers in Europe, notably BA and KLM, have been doing it this way for decades. No glass cockpits were available decades ago, and those carriers (plus others who have seen the light) still operate steam-driven, round-dial equipment and CANPA works fine there too. The fact that the first U.S. operators to migrate to CANPA will probably have aircraft with VNAV capability through the auto flight system does not change that fact.
To say that CANPA "is easier, but often it doesn't get the job done, it's not the way the procedure was designed, and above all, it can be dangerous" to my mind shows that Mr. Deakin may not be fully informed about developments in this field.
Maybe CANPA is easier. So what? Task overload had been shown to be a significant factor in approach and landing accidents. Procedures that make our lives easier should be welcomed. The FSF's Operations and Training Working Group said it this way: "The data clearly show that task saturation and overload for the pilot flying are significant contributors to Approach and Landing Accidents."
Why wouldn't CANPA get the job done? As far as I can see, Mr. Deakin's arguments against CANPA are just as applicable to the standard DaD concept. He may be right that the classic manner "will quietly, legally, and safely make the approach and landing while the glass cockpits go merrily to the alternate," but I doubt it. Especially the safely bit.
Not the way the procedure was designed? Says who?
Both TERPS and PANS-OPS (the ICAO equivalent) have as a goal keeping the pilots and their aircraft safely away from unwanted and premature contact with the ground. They do not, however, in all details specify how a pilot should operate his aircraft to stay within the design parameters of the approach procedure design. To stay within the lateral boundaries, maximum speeds are in some cases specified, but as far as vertical navigation goes, there are few guidelines except: do not bust specified minimum altitudes.
What TERPS and PANS-OPS do not do is to specify that the pilot should maximize their exposure to minimum terrain clearance the way Mr. Deakin proposes.
I would much rather do the wimpy gradual descent, with mental surplus capacity to monitor the altitude restrictions which may apply, than wrench a 50- or 250-ton airplane into a 1000 ft/min — or worse, the 1500 ft/min descent towards the MDA Mr. Deakin advocates. In my book, any descent rate in excess of 1000 ft/min close to the ground should be treated as dangerous. That is not to say that it will not happen or that it cannot be handled safely, but it should be treated with caution and carefully briefed and prepared for.
If this leads to more go-arounds and diversions, maybe that's just what the doctor ordered.
The airline I work for, Scandinavian Airlines System (SAS), has its way of doing this. We operate equipment ranging from round-dial DC-9s through the latest generation B-737-600/700s and MD-80/90 and B-767 in between. All our pilots use the same approach plates built on the same principles. They are based on TERPS in the U.S. and PANS-OPS in most of the rest of the world. We do not as yet fly GPS-based approaches.
We do not allow our pilots to design and fly auto flight guided VNAV approaches. Putting the waypoints and altitude constraints into a flight management system (FMS) is a very time-consuming task fraught with possibilities for mistakes. In addition, the preset waypoints and altitude constraints put in the FMS by the database producer do not permit easy corrections of minimum altitudes for low temperatures. It is therefore a hard job to ensure the vertical (obstacle clearance) integrity of a computer-based vertical path displayed to the pilots.
FMS-equipped aircraft are, however, excellent at keeping the pilot honest in the lateral dimension, provided the raw data are closely monitored.
We fly basically two types of non-precision approaches: those with DME guidance and those without. NDB approaches are almost always without, and VOR or LOC approaches may or may not have a DME associated with them.
In order to generate the reviled pseudo glide slope on an approach without DME guidance, timing must be used. This is reflected in our guidance material to our pilots. (See Figure 1.)
This is of course not enough. An operator must have clear and specific guidance in place for its pilots to be able to interpret and fly complicated procedures which all instrument approaches are. Accordingly our instructions contain descriptions of how, for instance, the minima are presented. (See Figure 2.)
This shows an example of a time-vs.-altitude box on a non precision-approach without DME guidance — in this case, for the NDB approach with an MDA of 530 feet. OHT is the locator where the timing and final descent starts. Rate Of Descent (ROD) vs. Ground Speed (GS) is given as guidance.
Here's a real life example from the U.S. (See Figure 3.)
This is our approach plate for Chicago KORD ILS –04R. What I want to highlight here is the LLZ (LOC) approach without DME guidance. It is a TERPS design (upper left corner) and has the altitude checks vs. timing. Timing and final descent is performed from the Outer Marker (OM). MDA is 1290 feet.
In Figure 4, the principle applied to an approach with a DME vs. ALT table is explained. Underlined altitudes are minimum altitudes and must not be penetrated before the specified DME.
Figure 5: This is what it looks like in a real-life approach plate for Charleston VOR-15. It is immediately obvious that the DME vs. ALT procedure is simpler than the one with timing, but they are both eminently flyable.
It is furthermore not enough for pilots just to have a reasonable understanding of how to interpret the approach plates when in flight. They must have a thorough understanding of the principles applied in their design and the inherent limitations in the presentations, which will invariably be present either due to lack of space or simple neglect.
Figure 6 is used to illustrate such a point. There can be no doubts in pilots' minds how to read this.
But even this is not enough. Somewhere in the operator's Standard Operating Procedures (SOPs) must be a guide showing how to handle unintentional deviations from for instance the ideal vertical path. Minimum altitudes must not be penetrated until they no longer apply, and SOPs must handle unplanned deviation events. Likewise it seems prudent to have a guide for how much deviation is tolerated on advisory altitudes, with associated call-outs for the non-flying pilot. "Slightly low, low, very low, slightly high, high, very high" and so on.
A Safer Way to Fly
Easy? Not exactly. Perfect? Not at all. But it is an attempt to get to grips with the minimum altitude busts, unstabilized approaches and resulting premature ground contact we all too often see in CFIT and approach and landing accidents.
I cannot guarantee that what I have presented here will be counted as great progress in the fullness of time, but I can guarantee that we will have to strive towards the objective of flying safer approaches. Technology will only get us so far, although equipment like GPS with WAAS/LAAS and EGPWS hold great promise.
Those of us who do not have this equipment today and will not have it until earliest the day after tomorrow, must do with flying smarter in the meantime.
The CANPA is here for the taking.
[Editor's Note: Click here to read John Deakin's response to this article.]