Crossing Restriction Altitude Deviations on SIDs and STARs (ASRS Directline #10)

A History of Ups and Downs

Throughout 21 years of operation by the Aviation Safety Reporting System (ASRS), approximately 35 percent of all incidents reported to the ASRS have been altitude deviations. Previous ASRS reviews of altitude errors have identified multiple contributing factors for these events. A 1982 ASRS study, Probability Distributions of Altitude Deviations, found that altitude deviations reported to ASRS were exponentially distributed with a mean of 1,080 feet, and that deviations from ATC-assigned altitudes were equally likely to occur above or below the assigned altitude.¹ Another ASRS review of altitude deviation problems, One Zero Ways to Bust an Altitude,² looked at the percentage of altitude deviations by altitude pairing, (i.e., confusing one altitude for another) and found that 35% of all paired deviations occur at 10,000 and 11,000.

More recently, ASRS analysts have noted that approximately 15 to 20 percent of the altitude deviations reported to ASRS involve crossing restriction errors on Standard Instrument Departures (SIDs) and Standard Terminal Arrival Routes (STARs). SIDs and STARs are published instrument routings whose primary purpose is to simplify ATC's clearance delivery procedures.

Altitude crossing restrictions associated with SIDs and STARs may be published on navigation charts or assigned by ATC. Crossing restrictions exist for two primary purposes: 1) to provide vertical separation from traffic on different routings that cross the same fix, and 2) to contain traffic vertically within a given ATC controller's sector in cases where other sectors within the same facility, or sectors in another facility, are layered above and below. ATC-assigned crossing restrictions (as opposed to published crossing altitudes) may be temporary requirements imposed to meet changing operational conditions, including facilitating traffic hand-offs to another sector. Pilot compliance with SID and STAR altitude assignments is important, for if a controller permits traffic penetration of another sector either laterally or vertically without prior coordination and approval from the controller in that sector, an operational deviation results.

No previous ASRS review of SID and STAR-related altitude deviations has been conducted. Thus we undertook this review to determine the causes and contributors to altitude deviations that occur during SID and STAR procedures, and to compare the results of this analysis with selected findings of the 1982 ASRS study.

Looking at Reports

The objective of this review was to categorize the types (i.e., undershoot or overshoot) and frequency of crossing restriction altitude deviations, and to determine the types of human performance errors that contribute to crossing restriction altitude deviations. Additionally, we looked at how and by whom these deviations are detected and corrected, and compared the number of deviations for traditional versus glass cockpit technology aircraft.

Reports selected for review in this study involved Part 121 or 135 aircraft in scheduled or non-scheduled air carrier operations conducting Standard Instrument Departure (SID) or Standard Terminal Arrival Route (STAR) procedures under Instrument Rules, where the flight failed to level at or cross a specified crossing restriction altitude as instructed by ATC or as required by a published procedure. Two hundred full-form records, from December 1988 through February 1996, were extracted from the ASRS Database and reviewed. Of these, 172 met the criteria for inclusion in this study. A five-page coding form was developed to extract pertinent information from the data set.

What Doesn't Matter

Of the 172 air carrier reports in the study, 159 involved turbojet aircraft and 13 involved turboprop aircraft. We found no evidence that the day of the week, time of day, aircraft type or configuration, or weather factors played a role in these altitude deviations. Similarly, it did not intuitively appear that crossing restriction altitude deviations were more likely to occur at any given ATC facility.

Finally, altitude crossing restriction errors were detected by ATC and the flight crew in approximately equal proportions: 53 percent were detected by flight crews, and 41 percent by ATC controllers.

ATC-Assigned vs. Charted Requirement

Where the required crossing restriction altitude was assigned by ATC, the flight failed to meet a crossing restriction on a SID or a STAR in 66 percent of events, while in 34 percent of events the crossing restriction was a charted requirement. The preponderance of incidents in which ATC assigned the crossing restriction altitude may be attributable to diminished time for climb or descent planning and to breakdowns of communications.

The following report excerpt demonstrates a communication problem:

"We had assumed that while in radar contact...we could safely descend to the cleared level of Flight Level 70 -- apparently Santiago Approach intended for us to observe the arrival procedure altitude restrictions, even though they had us in radar contact and had cleared us to descend to Flight Level 70. We both feel that this incident was in part due to communications misunderstanding." (# 294836)

And now for one that illustrates the problems of reduced time for descent planning:

"ATC deviated from the expected CIVET 1 Arrival. [We received an] unusual crossing restriction not normally used or expected during an approach into LAX. Too many short-notice clearances issued, with very little time between each of them." (# 304840)

Deviations Up -- Going Down

Only 23 percent of altitude deviation events in the data set occurred on occurred on SIDs (climb), while a full 77 percent occurred on STARs (in descent). One possible explanation for this variation may be workload: in the descent (STAR) phase of flight, flight crews have a large number of tasks and issues to contend with, including obtaining ATIS, adjusting or planning for changing weather conditions, conducting company communications, confirming gate assignments, planning for terminal procedures and runway configurations, traffic watch, configuring the aircraft, or alerting and communicating with cabin crew.

"Number 1 Flight Attendant came into the cockpit asking for gate connections and giving a cabin write-up. Managed to get distracted and forgot to reset altimeters to the proper setting below 18,000 feet." (# 306840)

It is also possible that on STARs there is greater ambiguity about ATC expectations, that is, when or where ATC expects the flight to initiate descent.

Undershoots and Overshoots

A majority of altitude deviations -- 75 percent -- were altitude undershoots (failure to reach the assigned altitude -- usually on descent). This indicates that flight crews may have been late in planning or execution of the procedure.

"The Captain began programming the FMC when we should have started down to Flight Level 190. Afterwards, the Captain commented that he always tells new copilots to begin the descent before programming the FMC if there is any doubt as to meeting the crossing [restriction], and he was upset that he had tried to program the FMC first." (# 112925)

Point of Detection

In over half of all events in the data set (51 percent), the error was detected before reaching the required or specified altitude. In 28 percent of events, the error was discovered at the required or specified crossing restriction altitude. In 17 percent of events the error was discovered after passing the required altitude.

In those events where the error was discovered at or before the required crossing altitude, climb or descent rates may have been sufficiently high to preclude recovery before the deviation occurred.

How Much Did We Miss By?

1. Point of Detection: The magnitude of the altitude deviation at the point of detection averaged 2,400 feet, with a median of 1,500 feet.

2. Point of Maximum Excursion: The altitude deviation magnitudes at the point of maximum excursion were examined using methods employed by the 1982 ASRS study, and were found to be exponentially distributed, with a mean deviation of approximately 2,500 feet. The mean for crossing restriction deviations at point of maximum excursion was substantially larger (approximately 1,400 feet greater) than the mean for undifferentiated altitude deviations (1,080 feet) reported in the 1982 ASRS study on altitude deviations. The median for the point of maximum excursion was 1,500 feet.

Controller Actions

ATC did not intervene, or was not required to intervene in order to avoid airborne conflict in 43 percent of incidents in the data set. (This supports the research team's subjective assessments of incident severity.) In 60 percent of incidents (100 of 168), the flight continued the climb or descent, with ATC concurrence.

"We were given descent clearance from Flight Level 230 to 13,000 feet by ATC, on the MINEE 1 Arrival (MCA). We read back 'Descending out of Flight Level 230 for 13,000 feet' and dialed 13,000 feet in the altitude select and began the descent. ATC then told us to contact Approach. We checked in with Approach and stated we were descending to 13,000 feet. As we passed through 14,700 feet, Approach asked us if we were level at 15,000 feet, we replied 'Negative, we are descending through 14,700 feet for 13,000 feet.' We also said we would stop the descent and return to 15,000 feet, if necessary. Approach replied, 'No, descend and maintain, 13,000 feet.' We then advised them we were given 13,000 feet by ATC and had checked in with him stating we were descending to 13,000 feet. Approach then said 13,000 feet was O.K." (# 297750)

Advanced and Traditional Cockpits

There were slightly more (61 percent) advanced cockpit (EFIS and/or NAV control) than traditional cockpit aircraft in the data set. This compares to 51 percent advanced cockpit versus 49 percent traditional cockpit air carrier aircraft in the entire ASRS database for the same time period.

It was expected that advanced cockpit aircraft would be more likely to be involved in crossing restriction altitude deviations due to the greater complexity in programming descents and descent crossing fixes. While we did see this pattern, the difference in numbers between advanced and traditional cockpit aircraft was not large.

Human Performance Errors

Reporters of incidents in this data set referenced human errors as shown in Table 1.

Table 1 -- Human Errors Based on 233 Citations from 171 of 172 Reports
Human Errors	Citations	Percent
Exercised poor judgment	43	25.1
Neglected to cross-check data	42	24.6
Delayed implementing procedure	41	24.0
Misunderstood clearance	35	20.5
Other (unspecified)	32	18.7
Forgot clearance	15	8.8
Did not read, or misread chart	14	8.2
Not stated or ambiguous	9	5.3
Did not hear clearance	1	0.6
Looked at wrong chart	1	0.6
TOTALS	233	136.4
Note: Multiple citations are possible in the category, thus the total number of citations exceds the number of reports.

An example of poor judgment is flight crew failure or reluctance to use speed brakes to meet descent profile requirements:

"Flight plan called for a (SIE CAMRN 2) STAR to JFK. ATC instructed us to cross CAMRN at 11,000 feet, 250 knots. At 19,000 feet, I told Captain we would not make restriction unless he used speed brake to increase rate of descent. He responded there would be no problem. I informed him I would tell ATC we were unable to comply with restriction. ATC responded, 'give us the altitude first and then the airspeed.' We crossed CAMRN at 13,000 feet and 290 knots. We were handed off to Approach for a normal continuation of flight to JFK with no comments made to us by ZNY or Approach Control reference the CAMRN crossing. The Captain's comment to me was that he did not like using the speed brake and thought he would be able to make the restriction." (# 315639)

Flight crews failing to cross-check data typically resulted in use of the wrong waypoint:

"Inbound to SLC at Flight Level 310. We were cleared the OGDEN 5 Arrival with a descent clearance to cross BEARR (25 nm NW Ogden VOR) at 17,000 feet. STAR path tracks outbound Burley VOR 117 degree radial to Ogden VOR 302 degree radial, then radar vectors. The STAR depicts the Ogden VOR very close to the Salt Lake City VOR. I (Captain) was flying the aircraft outbound on the Burley VOR radial, First Officer had switched his VOR to Salt Lake City for distance to the field. However, he did not verbally announce that he had switched to SLC -- I thought he was on Ogden. Because of unfamiliarity with arrival (only second time into SLC), I switched over to SLC VOR inbound (should have been Ogden). We discussed the fact the outbound and inbound radials did not match up but neither of us discovered my mistake. I therefore tracked off course and, because [I was] looking at the wrong DME, started the descent too late to make the crossing restriction." (# 300912)

Cockpit Workload

Reporters cited cockpit workload on SIDs and STARs as a factor in 44 percent of reports. The most commonly noted workload issues are shown in Table 2.

Table 2 -- Cockpit Workload Issues Based on 97 Citations from 171 of 172 Reports
Workload Issues	Citations	Percent
FMS Programming (automation issues)	18	24.0
High quantity radio communications with ATC	17	22.7
Lack of planning on the part of flight crew that led to time-compressions (such as cabin attendant in cockpit)	17	22.7
Other (misread altimeter, company com, etc.)	15	20.0
Flight Attendant call or cockpit-cabin interphone communication	12	16.0
A change in clearance	10	13.3
Weather factors	8	10.7
TOTALS	97	129.4
Note: Multiple citations are possible in the category, thus the total number of citations exceds the number of reports.

"I tried unsuccessfully to enter the restriction in the FMS. After three attempts, the Captain tried unsuccessfully and tried to explain why it wouldn't take. Meanwhile, no descent was started...we are flying an airplane, not a computer. My focus on the FMS got in the way of doing a very simple descent profile. I will be focusing on flying first, programming second." (# 259889)

SID and STAR Charts

In 88 percent of reports, there were no complaints about chart graphic depiction or procedures. There were, however, some complaints regarding chart text narratives, specifically that the font size was small, and that text blocks were sometimes not placed sufficiently close to the appropriate area of the graphic depiction. In one event, the flight crew of a turbojet transport followed instructions specific to turboprop aircraft, thus deviating from an altitude requirement.

Event Resolution

Table 3 provides event resolution information:

Table 3 -- Incident Resolution Based on 172 Citations from 171 of 172 Reports (categories are Mutually Exclusive)
Event Resolution Categories	Citations	Citations	Percent
Controller Actions		68	39.6
Controller intervened	52
Controller issued new clearance	16
Flight Crew Action		84	48.8
No action taken / anomaly accepted	26
No action taken / detected after the fact	23
Flight crew returned to original clearance / course	14
No action taken / insufficient time	13
Flight crew overcame equipment problem	4
Flight crew became reoriented	3
Avoidance maneuvers / evasive action	1
Unspecified		20	11.6
Not resolved / unable / other	17
Other	3
TOTALS	172	172	100.0

Incident Severity

In more than 95 percent of incidents in the data set, the analysts' subjective assessment was that there was minimal impact on flight safety or efficiency. While there was no direct evidence of loss of separation in the majority of these events, there may have been implications for ATC, such as sector penetration, of which the pilot reporters in this study were unaware.

Summing Up

Crossing restriction altitude deviations occur more often on STARs than SIDs, but traffic separation was known to be compromised in only a small portion of these events. Aircraft configuration or type did not appear to play a role in these incidents. Most deviations were altitude undershoots. An altitude undershoot on a STAR may indicate a flight crew's failure to adequately plan for the STAR, or their distraction from effectively monitoring the descent.

In instances of altitude overshoots, the flight crew or ATC often detected the error before the altitude deviation occurred; however, climb or descent rates may have been sufficiently high to preclude recovery before a deviation occurred. Crossing restriction altitude deviations occurred more often when the crossing altitude was assigned by ATC.

It is good practice to advise ATC of any altitude change, specifically the altitude being vacated and the destination altitude, and to confirm with ATC the point of anticipated or expected initiation of descent.

Flight crews anticipating or experiencing difficulty adhering to crossing restriction requirements should advise ATC as soon as practical.

Cockpit workload was commonly cited as a contributing factor in altitude deviations on STARs. Therefore, flight crews may wish to complete checklists early (mid-cruise or before descent), and review STAR charts before descent initiation.