Aircraft Accidents Are Avoidable
Aircraft accidents are a tragic thing however many people may argue that all are avoidable. When considering that all are avoidable, one must always remember there is an element that is in all accidents, Humans. Humans are involved in every aspect of flight such as, design, maintenance and as mentioned earlier piloted by. As long as humans are in 100% control of an airplane there will be accidents. However, a good side to an accident is it is thoroughly investigated and researched by the NTSB. (National Transportation and Safety Board). From a majority of the accidents, something can benefit aviation in general such as re-design, increased/specialized training, or as in this situation, re-evaluating current limitations to a safer level.
The focus of my project is going to be on a particular accident that took place on July 19, 1989 when a McDonnell Douglas DC-10-10 (Ref fig 1) aircraft crashed in Sioux City, Iowa. Not only the accident and why it happened but to delve into the root cause of the accident, the failure of the number two engine and why it happened. This particular type of aircraft, at the time of the accident, had been in production for about 18 years. The United Airlines jet was bound for Chicago when the tail-mounted engine a GE CF6-50C (ref figure 4) failed and managed to escape the engine containment case and severed the aircrafts main hydraulic lines. Without hydraulics, the aircraft wouldn’t be able to maneuver and landing would be another difficult task. This meant trouble for United Airlines flight 232 with 285 passengers and 11 crew- members on board.
On July 19,1989 a United Airlines DC-10-10 passenger plane took off Stapleton International Airport in Denver Colorado. The plane was scheduled to make a quick stop in Chicago, Illinois and then continue on to Philadelphia, PA. After an uneventful hour of flight a loud thump was heard, followed by a violent shaking of the aircraft. The flight crew observed the tail-mounted engine was not operating. While performing an engine shutdown procedure, the crew noticed that the fuel lever wouldn’t move to the closed position. They also observed that the hydraulic fluid quantities were all at zero. When the co-pilot tried to maneuver airplane it confirmed the loss of fluid when the airplane did not respond to his commands. The crew then radioed to the air traffic controller that they needed emergency assistance to the nearest airport. Sioux City Iowa was found to be the almost straight-ahead so it was chosen. After the captain briefed the passengers of the up-coming trials, a United Airlines DC-10 pilot instructor asked if he could be of any assistance. The captain asked him to see if there was any damage that he could see. The instructor found that the flight controls on the wings weren’t moving and that the horizontal stabilizer had damage to the left and right side. The horizontal stabilizer is the part of the aircraft that controls elevation and what is known as yaw. (Right and left turning) Without hydraulics the crew was limited on how to control the aircraft so the instructor pilot recommended the use of engine power to maneuver the airplane. The idea appeared to work when tried and being that all other options were few, the crew adopted this plan of action.
When flight 232 was about eight miles from the airport they lowered the landing gear. To their dismay they learned from the air traffic controller that they were lined up on a runway that was closed for repairs. The captain elected to stay on course because of the difficulty to turn the airplane. The air traffic controller said that the runway was 3000 feet shorter but they shouldn’t have any trouble landing due to the repairs. During the final approach the flight crew felt they were on course to make the landing. However they were a little faster than normal, due to not being able to lower the proper flight controls known as the flaps and slats. These two controls allow the airplane to sustain lift and fly at a lower speed. At about a hundred feet above the ground, the pilot reported that the nose dipped considerably and the right wing dropped also. Both pilots called for a reduction of power but the instructor pilot said that he needed to continue using the engine power to try and control the decent.
The airplanes right wing tip was the first to touch the runway, followed by the right main landing gear. Eyewitness accounts said the airplane ignited and cart wheeled upon impact. Figure 1 shows the final resting places of the different aircraft parts. In all, 111 people died. Amazingly, the remaining 155 people escaped with recoverable injuries.
The National Transportation and Safety Board (NTSB) now needed to understand why the accident occurred. Since the tail section was the potential cause of the crash, they transported the tail section parts to a nearby hangar and began to reconstruction of the tail. Once the tail section was completed this puzzle was starting to piece together. The left side of the horizontal stabilizer showed no damage that could have cause the crash. (See figure 2-3) However, the right side was a different story. The right side of the stabilizer indicated that approximately seventy pieces of metal had pierced through the skin. When aircraft engines are designed, the design engineers know they must be able to contain a sizable failure. The CF6-50 has what is called a containment case. (Ref figure 4) Like its name, is designed to contain an engine during failure. However, the engine involved in the Sioux City crash exceeded its design capabilities. When the tail mounted engine failed, metal debris violated the containment case and destroyed the number two hydraulic system lines. Metal debris also extended into the aircraft skin and severed the number 1 and 3 hydraulic system lines. Remember, this aircraft has three independent hydraulic systems. In this case, all three systems were damaged.
An uncontained failure was not something that was new to the aviation industry in 1989. In fact, there have been 14 uncontained failures since 1970. That doesn’t say too much when the jet engine was first introduced into commercial aviation in the 1960’s. One of the uncontained failures happened on November 3, 1973. A DC-10, powered by GE CF6-6 engines, had an uncontained failure on one of the wing engines. The fan disk struck a cabin window and a passenger was ejected from the aircraft and the aircraft subsequently decompressed. The fan disk had only 274 flight hours which is remarkable low, since the recommended life on a GE CF6-6 fan disk is 18,000 flight hours. It to say the least, uncontained failures aren’t something new to the aviation industry.
The next phase of the investigation was now focused on why the engine failed. It was known that number two engine had failed but there was a problem. The investigation team couldn’t find the front section of the engine, particular the fan disk. (Ref fig 5) It wasn’t until three months later in a cornfield in Alta, Iowa, the fan disk was found. Reference figure 6 which shows the fan disk and the reconstructed blades attached. The NTSB had now narrowed there root cause to the Stage 1 fan disk. They new that this had cause the uncontained failure, buy why? After evaluation of the pieces of the fan disk, it appeared that a flaw was present in the titanium which was probably had been there since it was originally forged. First, it must be understood that there are three primary steps in manufacturing of titanium alloy fan disks, such as the one on the GE engine. The processes are material processing (making the material) forging (forming the material to a particular shape) and final machining (taking the part to specified measurements.) (Ref figure 7) Titanium alloys are known for their strength and light weight which makes them ideal for use in aircraft engines. However, during the forging process three major types of melt related abnormities occur: Hard Alpha inclusions, high density inclusions and segregation. After extension evaluation of the stage 1 fan disk it was determined that the disk had a “hard alpha inclusion” which in lamens terms is an area of the disk had an enriched are of nitrogen or oxygen. Hard Alpha inclusions have melting point significantly higher than the normal remaining structure does. To help prevent these kinds of inclusions, a process known as triple melt is done. The material is heated to a temperature much higher than required, usually three times to ensure all inclusions are gone. Obviously something went wrong in the forging process of the material of the fan disk. This area, if not corrected, will become very brittle. When a brittle area is continually placed in high pressure situations, will eventually crack or break apart, as did the fan disk in this crash. Typically, records of the manufacturing process are not detailed enough to tell investigators enough about how it happened. The records only record the serial number. When a records search was done at General Electric Aircraft Engines on the fan disk serial number, there was a problem.
The fan disk involved in the accident was listed as S/N MPO 000385. GE records showed two different stories for MPO000385. One of the serial numbers was shown as to have been rejected for abnormalities in the disk. A paper trail shows that the part was set aside and eventually sent to a third party for evaluation and was eventually discarded by GE in November of 1972. GE maintains that the disk was discarded however no record of a credit to GE from the original material vendor existed. The second serial number shows a completely serviceable disk and that is was installed on an engine in January 1972.
Regardless of how the faulty material managed to get passed GE, the Fan disk was subsequently tested at several shop visits by what is known as NDT or Non Destructive Testing. (Ref figure 8) When the engine comes in for heavy maintenance, all life limited parts are tested using a variant of NDT. NDT testing is very important because oftentimes defects are not visible with the naked eye and to make the defects cracks) visible, NDT is required. To better understand how cracks can propagate, you must first understand how the forces are exhibited on them. When disks are rotated at high speed, a centrifugal force produces a stress that results in a strain or growth of the disk. If stressed within in its elastic region, the disk material returns to its natural form when the stress is removed. If the stress was of a sufficient magnitude, a small portion of the tensile strength is lost and the material will eventually fatigue and a crack initiates, propagates, and ultimately burst.
There are many variants of NDT but a process that is known as Florescent Penatrant Inspection is commonly used on LLP’s. (Life limited Parts) The part is dispensed in a florescent fluid and then placed under a ultra-violet light. If a surface or sub surface crack exists, it would be visible. Much of the effort in NDT has gone towards identifying when and where cracks have developed in rotating hardware. In a jet engine, rotating parts are inspected at regular intervals. Another form of NDT crack detection is a process known as eddy current testing. Eddy current testing is a better (safer) process than FPI due to the fact that eddy current can protrude through the entire part unlike FPI which can only do the surface and sub-surface testing. The major drawback to eddy current testing is the testing equipment can only be for a specific or localized area unlike FPI, which doesn’t require special testing equipment.
Finally, something had to be done to assure this type of situation would not happen again. The NTSB determined that United Airlines “had failed in giving adequate consideration to human factors which resulted in the failure to detect the crack.” McDonnell Douglas was asked to come up with a hydraulic system safeguard, given the same circumstances that total hydraulic system depletion wouldn’t occur. McDonnell Douglas designed two new safety guards: One, an electrically operated shutoff valve that would close if fluid levels dropped below a preset limit in the main hydraulic system, also known as a hydraulic fuse. Two, a sensor to detect the fluid loss that would trigger a light in the flight deck alerting the crew of the actuation of the valve. Looking at this accident from a distance it shows many remarkable things. The way the crew pulled together to come up with a plan of how they were going to land is absolutely remarkable. However, there is more to this story if one looks closer. To examine the flight crews’ actions would be unfair. In fact, the exact scenario was given to pilots in a DC-10 simulator and not one of them could come close to landing the airplane. As a result of the research and analysis conducted, the following conclusions were developed. First, if General Electric would have caught the flawed material, none of this would have ever happened. GE should have had safe guards to not allow this type of”quality escape” to occur. Next, is the United Airlines maintenance facility According to the analysis done on the disk, at the time of the last inspection, a crack would have been visible approximately one half of an inch in length. The inspector should have found the crack. Had he done so, the stage 1 fan disk would have not been deemed serviceable. To make sure that the disk had truly be tested using FPI, a chemical analysis was done on the disk. The disk showed residue of the penatrant, which leads one to believe it truly was tested. Finally, McDonnell Douglas should have installed the hydraulic fuses when the aircraft was initially made. An after thought is sometimes not good enough. The possibility of this type of failure was somewhere around a million to one but it happened. However, design engineers cannot predict every mechanical failure possible. If they put extra parts in for possible failures, then the aircraft would have a lot more parts, which means more maintenance and ultimately more cost to the customer.
NTSB. [No date] United Airlines Flt 232 McDonnell Douglas D-10-10 Sioux Gateway Airport Sioux City Iowa. Available: [online] http://amelia.db.erau.edu/reports/ntsb/aar/AAR90-06.pdf [08/15/03]
Religoustolerance.org [no date] Apa Format (American Psychological Association) Available: [online] www.religioustolerance.org/int_cita.htm [8/15/03]