An Interesting Situation.
In writing this article, I am reminded of the often used patter by young instructors in which the student is told to keep an eye on the temperatures and pressures on the take off. The reality is that oil and cylinder head temperatures are slow to indicate trouble, whilst oil pressure needles can flicker and vary by small amounts depending on engine power and RPM. A sudden drop of oil pressure near lift off could be an impending engine failure, or a gauge malfunction but there is simply no way of knowing for certain. A late abort for an engine gauge reading, when combined with a limiting runway length, has all the potential for an over run accident.
The correct time for a last minute assessment of engine health is at brake release with priority then given to directional control, lookout, and airspeed indications. Good airmanship dictates that a critical stop/go decision should not normally be based upon one gauge alone. Always check for corroborative evidence before committing oneself to an irrevocable course of action.
The following incident happened on a dark Pacific night, where a seemingly impossible combination of factors caused a B737 to come just a few seconds away from disaster.
The best job I ever had was flying a Boeing 737 for a small airline, which was based on a tiny island just 27 miles from the equator in the Central Pacific. The nearest daytime alternate was 375 miles away and a night diversion meant 500 miles to the Marshall Islands, or the burning of one's bridges of two hours island holding fuel, followed by landing or ditching! Our destinations included many of the Pacific islands between Hong Kong and Honolulu, the Solomons to Rarotonga, and New Zealand to Fiji. Our air hostesses all spoke English in a variety of charming regional accents and many were natural dancers who had been taught from childhood that music, laughter, and dancing was all that was needed to enjoy a full life.
We flew the popular Boeing 737 with most of the pilots expert at short runway operations and black hole approaches to remote islands. Some of us had been trained in the RAAF and our number included experienced ex fighter and transport pilots. Other pilots included Americans and Kiwis, while the remainder were Australian GA pilots who had been employed on charter and instructor flying before getting that lucky break into an airline.
The runway on our island base was 5600' long with a road cutting across only feet from each end. The over run area was just 100 feet from the ocean, with the prospect of fatal damage if the aircraft collided with huge phosphate rock boulders which formed the sea wall. In the wet season, strong southerlies meant cross winds up to 30 knots whilst huge waves whipped up by the winds dashed themselves against the rocks, sending mist and spray over the threshold.
Readers may remember a widely publicized accident to a Boeing 737 at Washington, USA, which tried to get airborne covered in snow and ice. It was unable to hold altitude and, after hitting a bridge, crashed nose first into the frozen Potomac river. The black box recorder was recovered and its evidence showed that the engines had not delivered full take off power during the take off. The engine power indicators had given false information to the crew, possibly due to ice blocking air inlet tubes which, in turn, sensed the power delivered. These tubes, which have an opening the size of a drinking straw, measure the pressure of air being drawn in by the engine compressors, and compare it to the pressure increase due to combustion which is pushed out at the back of the engine. Known as PT2 sensing tubes, they show the engine pressure ratio (EPR) on a cockpit gauge.
In simple terms, if the front tube is blocked, the sensor thinks no air is coming into the front of the engine. The rear sensor, operating normally, senses lots of high pressure hot air being ejected from the tail pipe and thus the EPR will overread. The natural tendency to remedy the apparent overpower indication on the EPR gauges is for the pilot to ease the throttles back in order to keep within perceived engine limits. The engine RPM gauge will, however, show the pilot the real power being produced. Obviously, if 100% RPM is indicated, the engine is really pushing out lots of power, regardless of a false reading on the EPR gauge caused by a blocked PT2 tube. The advantage of the EPR gauge is that accurate power settings can be measured, providing of course that the system works as advertised.
Following the lengthy investigation into the Potomac accident, notices were sent to all operators of Pratt & Whitney JT8D series engines, warning that crews should be on alert for erroneous EPR indications in icing conditions and to rely primarily on the engine RPM gauge for actual indications of power. Typically, the RPM gauge is called an N1 or Fan gauge and will usually show 35% N1 whilst idling, 83% in cruise, and 95% - 101% on take off. Blocking of PT2 tubes by substances other than ice was not discussed in the Alert Bulletin.
Boeing recommended that the crew calculate the expected EPR and N1 gauge readings for each take off. These readings, which are placed on a take off data card will vary, depending on the take off weight of the aircraft, length of runway, ambient air temperature, and aerodrome pressure altitude. Also on the card will be the V1 decision speed, rotation speed, and other information pertaining to the take off.
The following episode began when I was rostered to fly as a passenger on a non-stop night flight to Guam in the Western Pacific. Flight time was four hours and, on arrival at Guam in the early morning, I had planned to catch some sleep at The Hilton Hotel before crewing another flight to Manila.
There were 60 passengers including some deadhead crew on the flight and, after boarding, I settled into a first class seat, adjusted my reading glasses, and watched the senior hostess brief her cabin crew as the engines were started. A few minutes later, at 0130 local time, the aircraft moved onto the runway, back tracking for take off into the north west. From my window seat one could see a line of cars on the nearby road only 50 yards from the runway. The flashing blue light on a traffic policeman's motor bike indicated that he had stopped all traffic to prevent anyone getting blown off the road by the jet blast on take off. A few seconds later the senior hostess came to me and said that the captain had invited me up front for take off. Like most pilots, I welcomed the chance of observing the action from the cockpit and, leaving my reading glasses on the seat next to me, I entered the darkened flight deck, quickly sat on the jump seat and thanked the captain, whom I had trained for his command some months earlier.
The first officer was to carry out the take off and I caught the last part of the emergency briefing as we slowly turned to line up. The take off data card indicated 10 degrees of flap for take off, V1 130 knots, VR (rotate) speed 135 knots, and initial climb speed 145 knots. Even without reading glasses I could plainly see the EPR gauge digital cursors set for 2.18 EPR, which meant full power was needed. This was understandable, considering the short runway, the hot night, and the extra fuel needed for a long flight. The data card also showed that the crew had calculated 100% N1 was needed for take off and this tied in with the 2.18 EPR limit. The N1 gauges were dimly lit and I could not see the needles clearly without my glasses, which I had left in the cabin.
The traffic officer's blue strobe light was still flickering on the road ahead and, from our position on the runway threshold, I could just make out the surf of the Pacific a few feet behind us and the dark shape of the control tower some two thirds down the runway and just off the parallel main road. From previous experience, I knew that the indicated air speed should be around 120 knots as the aircraft passed abeam the tower, with lift off speed usually 10 seconds later.
The captain opened the throttles to 1.6 EPR on the brakes, checked that both engines spooled up evenly, then quickly advanced the throttles to the planned take off power of 2.18 EPR. The brakes were released and the first officer began to steer the aircraft down the centre line. Acceleration appeared normal, and I could clearly see both EPR gauges steady at 2.18. The airspeed indicator needle began to accelerate past 60 knots and I checked all engine gauges in a swift eye scan. Fuel flow, N1, Exhaust Gas Temperature (EGT), were all pointing in the right area, although somewhat blurred to my vision without my glasses. Seconds passed and the captain called "80 knots", as the dual airspeed indicator check. A sixth sense warned me that the acceleration was not the solid kick in the back that I would have expected from 2.18 EPR, and at the same instant I noticed the captain begin to glance rapidly from the instruments to the remaining runway ahead. There was no readily discernible problem but I had an uneasy feeling that something was not quite right.
The company procedure was that, apart from the 80 knot airspeed check, no calls were to be made by either pilot unless something was seriously amiss. On this occasion, the take off seemed to proceeding normally and, apart from my vague unease at the perceived lack of marked acceleration, I was unable to pinpoint any impending problem.
The control tower and passenger terminal building flashed past the right wing tip, as I strained forward against my shoulder straps in an attempt to focus more clearly on the vital N1 gauges. The EPR needles were clear - exactly 2.18, but again I could not get an accurate look at the N1 without glasses. The airspeed reading went through 110 knots, we should have been perhaps 10 knots faster from my experience, and my unease grew stronger. One thing was happening for sure, and that was we were rapidly using up the remaining runway. Six runway lights to go, and we were still at least 10 knots below V1, the go/stop decision speed. It was, to say the least, an interesting situation and I hoped that the captain would not make a split second decision to abort the take off, because we could now never pull up in time, even with maximum reverse thrust and braking. Our V1 speed was useless now, and the invisible sea swept rocks were only seconds ahead.
My unease had just changed into the cold realization that we were never going to attain lift off speed before reaching the end of the runway, when suddenly the captain urgently called " ROTATE NOW!", and hauling back on the control column he pushed both throttles hard against their forward stops. Boeing term it "firewalling the thrust levers",to be used as a last resort to climb out of trouble.
The last runway light disappeared under us, as did a fleeting close up sight of the brilliant blue flashing strobe light of a shocked traffic cop's motor bike. I felt the reassuring surge of thrust propel the 737 upwards at a body angle of 20 degrees, and silently thanked the Lord that the captain had made an instant correct decision to firewall the Pratt & Whitneys. I knew that the Potomac accident might have been averted if the crew had only hit the throttles wide open to the stops, to prevent their ice laden Boeing from stalling.
We were later told that the flight data recorder showed that the aircraft had lifted off at 15 knots BELOW the calculated VR or rotation speed, and that the aircraft had flown just 19 feet above the sea for several hundred yards before gradually climbing away. We never did see the towering metal structure of the phosphate cantilevers that passed above our altitude, and situated 200 yards to the right of the extended runway centre line...
Ahead was sheer blackness and the captain locked on to instruments as the ASI needle crept towards safe flap retraction speed. The VSI was held at 1000 fpm, and the first officer set the climb thrust at 1.93 EPR as the flaps were slowly retracted in sequence. It seemed an abnormally long time before the aircraft reached 250 knots, which was the scheduled climb speed that night, and the rate of climb was well below normal. Finally we passed 5,000 ft, engaged the autopilot, and called for coffee whilst we held a round table conference on the recent events. The maintenance engineer who had been seated in the cabin came up front and said that a couple of deadheading pilots down the back sent their respects to the captain but they hoped he had finished playing silly buggers with the aircraft as they were hoping to get some shut eye! They had obviously felt the thrust change through the seat of their pants.
We turned our attention to a detailed scan of the engine instruments and the engineer remarked that the N1 indications seemed low when compared with the 1.93 EPR climb setting. From the back of my mind came the recollections of previous problems that I had experienced several months ago with an over reading EPR gauge. On the first occasion, we had just attained take off thrust early in the roll, when one EPR needle moved to an apparent overboost figure of 2.35 EPR, whilst the second needle stayed steady at the planned 2.10 EPR. The other engine parameters were normal for take off and, in particular, both engines were turning up nicely at 100% N1. Clearly the problem was a faulty EPR indication and, as our speed was only 50 knots, I decided to stop the take off run and return to the terminal for a chat with the engineers. A check of the PT2 tube, plus an engine run up indicated the problem had cleared itself and we departed an hour later.
More recently, at 100 knots on take off, a similar fault occurred and this time, the F/O urgently called that the engine was over speeding. He attempted to pull back the throttle on that engine to limit the peak EPR, but I quickly stopped his hand and told him to ignore the faulty reading. He was convinced however, that the engine was over boosting because of the high EPR reading, although I felt no asymmetric yaw on the flight controls. I again prevented him from dragging the offending EPR back and we continued the take off using the N1 RPM (which was steady at normal take off thrust). Once at a safe altitude, I turned on the hot air bleed system to the engine anti ice, and almost immediately the offending EPR needle did a few cartwheels and returned to normal. We were not in icing conditions but the hot air used for de-icing had obviously cleared some obstruction in the PT2 tube. The flight was continued without further incident.
Back now to the present situation, where early indications of long distance storm activity began to show up on the weather radar screen. The storm tops were around 35,000 ft, and at our dismal rate of climb we would be in the thick of things in the next 20 minutes.
With the throttles set at the computed climb EPR, it was readily apparent that both N1 readings of 88% and commensurate low fuel flows meant that some common denominator was affecting both engines at the same time. We
discussed fuel contamination but decided that it was unlikely, given that the engines had delivered maximum available overboost when the throttles had been firewalled earlier. I gave fleeting thought to the possibility of EPR gauge malfunction, especially after my previous experience with this problem, and a knowledge of the Potomac accident. With an aerodrome temperature of 30 degrees centigrade, icing of the PT2 tubes could be discounted and, in any case, it would be highly unlikely that an identical malfunction would affect both PT2 tubes simultaneously.
On my suggestion, the captain momentarily switched on the engine anti-ice to both engines. This would normally cause a small loss of about 5% N1 and an EPR drop of .08, which reflected the stealing of some hot compressor air for piping to the engine inlet cowls and PT2 tubes.
The N1 dropped obediently but both EPR gauges went crazy, increasing by an unheard of amount, and in the opposite direction to that expected. My mind went back to a paragraph in the Potomac accident report which mentioned that with the engine anti-ice switched on and PT2 tube blocked, the EPR needle would indicate a reverse reading to that expected. Thus, tonight, the impossible had occurred, an identical erroneous reading on both EPR gauges at the same time. The PT2 tubes were obviously still blocked but we now knew for sure that both engines were operating normally. The decision was made to return to land and, at 500 ft, the landing lights were switched on, illuminating drifting mists of phosphate dust from the nearby mine. The touch down was perfect, right on the 1000 ft runway marker.
Well coordinated reverse thrust and braking gave nice cool brakes on arrival back at the terminal.
As the passengers disembarked to wait out the delay in the airport terminal, engineers had already removed the engine nose cones in order to check out the PT2 system. With the aid of a torch the cause of our troubles was soon discovered. The PT2 tube of each engine, the sensor that gave the vital Engine Pressure Ratio readings for take off, were blocked, not with ice but with congealed phosphate dust and some other glutinous substance. It was impossible to determine the precise time that the tubes became blocked, or how the substance found its way into the system.
At dawn a few hours later, early workers driving past the departure end of the runway were the first to see debris from the coast road and nearby cliff face rocks, blasted back over the threshold by the jet efflux of the 737. Black skid marks on the road showed where the traffic policeman had burned up rubber in a spectacular scramble for safety.
Later calculations showed that the actual power on take off was around 2.05 EPR, even though the EPR needles were steady at 2.18. That power would have been ample for a long runway, and in fact was a setting frequently used for the right combination of runway length and gross take off weight. The N1 gauge scale between 91% and 100% is less than 3mm and very difficult to read in dim light, especially at a quick glance. This might explain why the crew were unable to pick the apparent lower than normal N1 readings on the take off run. At night especially, it is also impossible to make any meaningful correlation between rate of acceleration and runway remaining. Until it is almost too late, that is.
Several months afterwards, I read a report which described an incident on a Boeing 727 which departed at night from an airport in USA. The aircraft used 9000 ft of runway and during rotation, it cleaned up the ILS localiser aerials situated more than 1000 ft beyond the overrun area. The aircraft was damaged but continued to fly. Investigation revealed that icing conditions had prevailed and the crew had failed to actuate the engine anti ice switches for take off. All the PT2 tubes had iced up during the take off roll, giving significant EPR gauge error. The crew did not detect any acceleration problem until almost too late and also did not firewall the throttles.
Later versions of the Boeing 737 have CFM56 engines which rely on N1 gauges as the primary power indication. EPR gauges still remain on many older jet transports.