Narrative:

These recollections are made after getting about 5 hours of sleep after arrival. The essence of the incident is that we ended up stabilizing on the glide slope late. Here is how it unfolded.it was the first officer's (first officer) leg; number 4; and the last leg in the trip pairing. We had initially flown an am leg; followed by day crew rest that was a poor one for both crew members; due to the 2.5 hour drive to the hotel. (There was a wreck on the [road]). When we met to begin our night duty period to fly; we both related that sleep did not come easily; having arrived at the hotel so late in the day. We proceeded uneventfully to [first destination]. At the hub; we each tried to get some rest in a sleep room-I dozed fitfully; and the first officer mentioned during our following flight planning that she had not been able to sleep.we proceeded to [next destination]; and after blocking in; while discussing the current and forecast weather for our final leg; discussed the possible need for me to make a cat ii auto land approach at ZZZZ; but would make that decision enroute. It was the first officer's leg; and she was pilot flying until we determined that although ZZZZ visibility was reported well above cat I minimums; the variable thin broken cloud report of varying between 200 and 2000 feet lead us to conclude that a cat ii auto land approach would be prudent. We completed the approach checklist and exchanged roles on radar downwind. During the remainder of the radar pattern; we had consecutive reductions in altitude such that we never leveled off until reaching the assigned 2000 feet altitude for glide slope intercept. As the airplane leveled off on autopilot; I noticed that for some reason; we were a little above 2000 feet; so I selected a vertical speed descent of a few hundred feet per minute to get us down. About this time; we were distracted by switching to tower; and continued a slow descent while cleared for the approach to what was in fact below 2000; although we did not realize it. Monitoring my radio altimeter; I could not understand why the glide slope indicator had not come alive; and was about to initiate a go-around and try to figure it out later. About this time the first officer said something like 'we are below glide slope;' so I leveled off; and we intercepted the glide slope from below; and continued the approach. During this phase; we were VMC; and could actually see the ground; but did not have slant range visibility to the runway-due to low broken clouds.after landing and debriefing; we questioned the correctness of my altimeter setting; and only after some sleep and further thought; have I pieced together what I think happened. Going back to the arrival sequence; since it had been the first officer's leg; the right autopilot was in command. Due to consecutively lower altitude assignments; we never actually leveled off until the 2000 feet assignment; which; as I mentioned above; we did not descend to without engaging vertical speed. The only explanation for this that I can come up with is that my altimeter was somehow still in standard versus set to qnh. I had completed the approach check; and was sure that at one time I saw the local altimeter setting; but until I selected the approach tile on the radar dogleg; which engaged all 3 autopilots; the right autopilot remained in command; reflecting the first officer's correct altimeter setting. That fact; combined with consecutive lower altitude assignments that precluded level off until final; masked the error. The net result was that we paralleled the glide slope just below it; intercepted the glide slope late; and pushed stabilized approach criteria. I estimate we were 4-5 miles out when all this took place; and about 1200 feet or so AGL. The reason it seemed so comfortable to continue is that we were actually VMC; and could see the ground below us. In retrospect; a better decision would have been to go around when I was initially inclined to do so; figure out exactly what was not right; and shoot another approach. It has taken some hours of recollection and thought to piece together what happened; time not available when it happened. Factors contributing were: third leg of an all-night duty period; challenging weather; and fatigue. Both crew members had not slept well; and I am shocked at the insidious onset of dull thought leading to poor decision making in my own case brought on by fatigue.for me; avoid hub turns. I do not sleep well during the day; and so must simply avoid the need to do so. In flight; go around when you determine things are not right.

Google
 

Original NASA ASRS Text

Title: B757-200 Captain reported fatigue was a factor in his approach that was flown below the glideslope to about 1200 feet AGL; 4 to 5 miles out.

Narrative: These recollections are made after getting about 5 hours of sleep after arrival. The essence of the incident is that we ended up stabilizing on the glide slope late. Here is how it unfolded.It was the First Officer's (FO) leg; number 4; and the last leg in the trip pairing. We had initially flown an am leg; followed by day crew rest that was a poor one for both crew members; due to the 2.5 hour drive to the hotel. (There was a wreck on the [road]). When we met to begin our night duty period to fly; we both related that sleep did not come easily; having arrived at the hotel so late in the day. We proceeded uneventfully to [first destination]. At the hub; we each tried to get some rest in a sleep room-I dozed fitfully; and the FO mentioned during our following flight planning that she had not been able to sleep.We proceeded to [next destination]; and after blocking in; while discussing the current and forecast weather for our final leg; discussed the possible need for me to make a Cat II auto land approach at ZZZZ; but would make that decision enroute. It was the FO's leg; and she was Pilot Flying until we determined that although ZZZZ visibility was reported well above Cat I minimums; the variable thin broken cloud report of varying between 200 and 2000 feet lead us to conclude that a Cat II Auto land approach would be prudent. We completed the Approach checklist and exchanged roles on radar downwind. During the remainder of the radar pattern; we had consecutive reductions in altitude such that we never leveled off until reaching the assigned 2000 feet altitude for glide slope intercept. As the airplane leveled off on autopilot; I noticed that for some reason; we were a little above 2000 feet; so I selected a Vertical speed descent of a few hundred feet per minute to get us down. About this time; we were distracted by switching to tower; and continued a slow descent while cleared for the approach to what was in fact below 2000; although we did not realize it. Monitoring my radio altimeter; I could not understand why the Glide Slope Indicator had not come alive; and was about to initiate a go-around and try to figure it out later. About this time the FO said something like 'we are below glide slope;' so I leveled off; and we intercepted the glide slope from below; and continued the approach. During this phase; we were VMC; and could actually see the ground; but did not have slant range visibility to the runway-due to low broken clouds.After landing and debriefing; we questioned the correctness of my altimeter setting; and only after some sleep and further thought; have I pieced together what I think happened. Going back to the arrival sequence; since it had been the FO's leg; the Right Autopilot was in command. Due to consecutively lower altitude assignments; we never actually leveled off until the 2000 feet assignment; which; as I mentioned above; we did not descend to without engaging Vertical Speed. The only explanation for this that I can come up with is that my altimeter was somehow still in Standard versus set to QNH. I had completed the Approach check; and was sure that at one time I saw the local altimeter setting; but until I selected the Approach tile on the radar dogleg; which engaged all 3 autopilots; the Right autopilot remained in command; reflecting the FO's correct altimeter setting. That fact; combined with consecutive lower altitude assignments that precluded level off until final; masked the error. The net result was that we paralleled the glide slope just below it; intercepted the glide slope late; and pushed stabilized approach criteria. I estimate we were 4-5 miles out when all this took place; and about 1200 feet or so AGL. The reason it seemed so comfortable to continue is that we were actually VMC; and could see the ground below us. In retrospect; a better decision would have been to go around when I was initially inclined to do so; figure out exactly what was not right; and shoot another approach. It has taken some hours of recollection and thought to piece together what happened; time not available when it happened. Factors contributing were: third leg of an all-night duty period; challenging weather; and fatigue. Both crew members had not slept well; and I am shocked at the insidious onset of dull thought leading to poor decision making in my own case brought on by fatigue.For me; avoid hub turns. I do not sleep well during the day; and so must simply avoid the need to do so. In flight; go around when you determine things are not right.

Data retrieved from NASA's ASRS site and automatically converted to unabbreviated mixed upper/lowercase text. This report is for informational purposes with no guarantee of accuracy. See NASA's ASRS site for official report.