December 15, 1997 Close-Up: ValuJet Flight 592

Email this article |Print this article

by	The Editors of AVweb
	Documents Provided by the NTSB This report originally appeared in the December 1997 issue of NTSB REPORTER.

About NTSB Reporter ... NTSB Reporter is a unique magazine for pilots, mechanics, or anyone who has a serious interest in aviation safety. Each month, this outstanding publication brings you the kind of detailed accident investigation data exemplified by the preceding report on the Jessica Dubroff accident. No opinions, no speculation, no advertisements...just 100% hard data from NTSB investigations of notable air carrier, military and general aviation accidents, plus late-breaking news briefs on aviation safety topics. AVweb's editor-in-chief Mike Busch has been a subcriber for more than a decade, and thinks NTSB Reporter is a must-read for every serious student of aviation safety. Subscribe on-line and help support continued continued free access to AVweb and AVflash.

In-Flight Fire And Impact With Terrain
ValuJet Airlines, Flight 592, DC-9-32
Everglades, Near Miami, Florida

HISTORY OF THE FLIGHT:

On May 11, 1996, at 1413:42 eastern daylight time, a Douglas DC-9 32 crashed into the Everglades about 10 minutes after takeoff from Miami International Airport (MIA), Miami, Florida. airplane, registration number N904VJ, was being operated by ValuJet Airlines, Inc., as flight 592. Both pilots, the three flight attendants, and all 105 passengers were killed. ValuJet passenger records indicated that 104 passengers boarded the airplane. A 4-year-old child also was aboard; however, the presence of this child was not shown on the passenger manifest or on the weight and balance and performance fawn. Visual meteorological conditions existed in the Miami area at the time of takeoff. Flight 592, operating under Part 121, was on an IFR flight man destined for the William B. Hartsfield International Airport (ATL), Atlanta, Georgia.

ValuJet flight 591, the flight preceding the accident flight for the same aircraft, was operated by the accident crew. Flight 591 was scheduled to depart ATL at 1050 and arrive in MIA at 1235; however, ValuJet's dispatch records indicated that it actually departed the gate at 1125 and arrived in MIA at 1310. The delay resulted from unexpected maintenance involving the right auxiliary hydraulic pump circuit breaker.

Flight 592 had been scheduled to depart MIA for ATL at 1300. The cruising altitude was to be flight level 350, with an estimated time en route of one hour 32 minutes. The ValuJet DC-9 weight and balance form completed by the flightcrew for the flight to ATL indicated that the airplane was loaded with 4,109 pounds of cargo (baggage, mail, and company-owned material [COMAT]). According to the shipping ticket for the COMAT, it consisted of two main tires and wheels, a nose tire and wheel, and five boxes that were described as "Oxy Cannisters [sic] - 'Empty'." According to the ValuJet lead ramp agent on duty at the time, he asked the first officer of flight 592 for approval to load the COMAT in the forward cargo compartment, and he showed the first officer the shipping ticket. According to the lead ramp agent, he and the first officer did not discuss the notation about the oxygen canisters on the shipping ticket. The ramp agent who loaded the COMAT into the cargo compartment stated that within five minutes of loading the COMAT, the forward cargo door was closed. He could not remember how much time elapsed between his closing the cargo compartment door and the airplane being pushed back from the gate.

Flight 592 was pushed back from the gate shortly before 1340. According to the transcript of Air Traffic Control (ATC) radio communications, flight 592 began its taxi to runway 9L about 1344. At 1403:24, ATC cleared the flight for takeoff and the flightcrew acknowledged the clearance. At 1404:24, the flightcrew was instructed by ATC to contact the north departure controller. At 1404:32, the first officer made initial radio contact with the departure controller, advising that the airplane was climbing to 5,000 feet. Four seconds later, the departure controller advised flight 592 to climb and maintain 7,000 feet. The first officer acknowledged the transmission.

At 1407:22, the departure controller instructed flight 592 to "turn left heading three zero zero join the WINCO transition, climb and maintain one six thousand." The first officer acknowledged the transmission. At 1410:03, an unidentified sound was recorded on the Cockpit Voice Recorder (CVR), after which the captain remarked, What was that?" According to the Flight Data Recorder (FDR), just before the sound, the airplane was at 10,634 feet mean sea level (MSL), 260 knots indicated airspeed (KIAS),and both engine pressure ratios (EPRs) were 1.84.

At 1410:15, the captain stated, "We got some electrical problem," followed five seconds later with, We're losing everything.. At 1410:21, the departure controller advised flight 592 to contact Miami Center on frequency 132.45 MHz. At 1410:22, the captain stated, We need, we need to go back to Miami,. followed three seconds later by shouts in the background of "fire, fire, fire, fire." At 1410:27, the CVR recorded a male voice saying, "We're on fire, we're on fire."

At 1410:28, the controller again instructed flight 592 to contact Miami Center. At 1410:31, the first officer radioed that the flight needed an immediate return to Miami. The controller replied, "Critter five ninety two uh roger turn left heading two seven zero descend and maintain seven thousand." The first officer acknowledged the heading and altitude. According to a pre-existing agreement between the FAA and ValuJet, air traffic controllers used the term "Critter" as a callsign when addressing ValuJet aircraft. "Critter. referred to the logo of a cartoon airplane painted on the ValuJet fleet. The peak altitude value of 10,879 feet MSL was recorded on the FDR at 1410:31 and, about 10 seconds later, values consistent with the start of a wings-level descent were recorded.

According to the CVR, at 1410:36, the sounds of shouting subsided. About four seconds later, the controller asked flight 592 about the nature of the problem. The CVR recorded the captain stating "fire" while the first officer radioed, "uh smoke in the cockp...smoke in the cabin.. The controller responded, "Rogers and instructed flight 592 to turn left when able to a heading of two five zero and to descend and maintain 5,000 feet. At 1411:12, the CVR recorded a flight attendant shouting, "Completely on fire."

The FDR and radar data indicated that flight 592 began to change heading to a southerly direction about 1411:20. At 1411:26, the north departure controller advised the controller at Miami Center that flight 592 was returning to Miami with an emergency. At 1411:37, the first officer transmitted that they needed the closest available airport. At 1411:41, the controller replied, Critter five ninety two they're gonna be standing (unintelligible) standing by for you, you can plan runway one two when able direct to Dolphin [VOR] now. At 1411:46, the first officer responded that the flight needed radar vectors. At 1411:49, the controller instructed flight 592 to turn left heading one four zero. The first officer acknowledged the transmission.

At 1412:45, the controller transmitted, "Critter five ninety two keep the turn around heading uh one two zero." There was no response from the flightcrew. The last recorded FDR data showed the airplane at 7,200 feet MSL, at a speed of 260 KIAS, and on a heading of 218 degrees. At 1412:48, the FDR stopped recording data. The airplane's radar transponder continued to function; thus, airplane position and altitude data were recorded by ATC after the FDR stopped.

At 1413:18, the departure controller instructed, "Critter five ninety two you can uh turn left heading one zero zero and join the runway one two localized at Miami." Again there was no response. At 1413:27, the controller instructed flight 592 to descend and maintain 3,000 feet. At 1413:37, an unintelligible transmission was intermingled with a transmission from another airplane. No further radio transmissions were received from flight 592. At 1413:43, the departure controller advised flight 592, "Opa Locka Airport's about 12 o'clock at 15 miles."

The accident occurred at 1413:42. Ground scars and wreckage scatter indicated that the airplane crashed into the Everglades in a right wing down, nose down attitude. The location of the primary impact crater was approximately 17 miles northwest of MIA.

STATEMENTS OF WITNESSES: Two witnesses fishing from a boat in the Everglades when flight 592 crashed stated that they saw a low-flying airplane in a steep right bank. According to these witnesses, as the right bank angle increased, the nose of the airplane dropped and continued downward. The airplane struck the ground in a nearly vertical attitude. The witnesses described a great explosion, vibration, and a huge cloud of water and smoke. One of them observed, "...the landing gear was up, all the airplane's parts appeared to be intact, and that aside from the engine smoke, no signs of fire were visible."

Two other witnesses who were sightseeing in a private airplane in the area at the time of the accident provided similar accounts. These two witnesses and the witnesses in the boat, who approached the accident site, described seeing only part of an engine, paper, and other debris scattered around the impact area. One of the witnesses remarked that the airplane seemed to have disappeared upon crashing into the Everglades.

CHEMICAL OXYGEN GENERATORS CARRIED AS CARGO: Events Preceding the Accident: On January 31, 1996, ValuJet agreed to purchase two McDonnell Douglas MD-82s (registration numbers N802W and N803W) from McDonnell Douglas Finance Corporation (MDFC), and on February 1, 1996, agreed to purchase a Model MD-83 (N830VV) from MDFC. All three airplanes were ferried to the Miami maintenance and overhaul facility of the SabreTech Corporation for various modifications and maintenance functions. SabreTech was a maintenance facility with which ValuJet had an ongoing contractual relationship for line and heavy maintenance.

One of the maintenance tasks requested by ValuJet was the inspection of the oxygen generators on all three airplanes to determine if they had exceeded the allowable service life of 12 years from the date of manufacture.

SabreTech determined that all of the generators on N830W had expiration dates of 1998 or later, but that the majority of oxygen generators on N802W and N803W were past their expiration dates. Because the few oxygen generators on N802W and N803W that had not reached their expiration date were approaching it in the near future, ValuJet directed SabreTech to replace all of the oxygen generators on these two airplanes.

DESCRIPTION OF CHEMICAL OXYGEN GENERATORS: The MD-80 passenger emergency oxygen system uses chemical oxygen generators together with oxygen masks mounted behind panels above or adjacent to passengers. If a decompression occurs, the panels are opened either by an automatic pressure switch or by a manual switch, and the mask assemblies are released.

A plastic tube through which the oxygen will flow is connected from the mask assembly reservoir bag to an outlet fitting on one end of the oxygen generator. Additionally, a lanyard, or slim white cord, connects each mask to a pin that restrains the spring-loaded initiation mechanism (retaining pin). The lanyard and retaining pin are designed such that a one- to four pound pull on the lanyard will remove the pin, which is held in place by a spring-loaded initiation mechanism.

When the retaining pin is removed, the spring loaded initiation mechanism strikes a percussion cap containing a small explosive charge mounted in the end of the oxygen generator. The percussion cap provides the energy necessary to start a chemical reaction in the generator oxidizer core, which liberates oxygen gas. A protective shipping cap that prevents mechanical activation of the percussion cap is installed on new generators. The shipping cap is removed when the oxygen generator has been installed in the airplane and the final mask drop check has been completed.

The oxidizer core is sodium chlorate which is mixed with less than five percent barium peroxide and less than one percent potassium perchlorate. The explosives in the percussion cap are a lead styphnate and tetracene mixture.

The chemical reaction is exothermic, which means that it liberates heat as a byproduct of the reaction. This causes the exterior surface of the oxygen generator to become very hot. The maximum temperature of the exterior surface of the oxygen generator during operation is limited by McDonnell Douglas specification to 547 degrees F., when the generator is operated at an ambient temperature of 70 to 80 degrees F. Manufacturing test data indicate that when operated during tests, maximum shell temperatures typically reach 450 to 500 degrees F.

GUIDELINES FOR REMOVAL OF GENERATORS: Chemical oxygen generator removal and installation practices and procedures are contained in the Douglas MD-80 maintenance manual and on the ValuJet MD-80 work card 0069. The Douglas MD-80 maintenance manual specifies that non-expended oxygen generators are to be removed from service 12 years after the date of manufacture to maintain reliability in the operation of the generators. According to the generator manufacturer (Scott Aviation), the primary concern that led to establishing the 12-yearservice life was the continued mechanical integrity of the core and its support structure, not changes to the chemical composition of the core. The 12-year limit was established based on tests conducted by Scott Aviation. ValuJet provided these documents to SabreTech.

The Douglas MD-80 maintenance manual provides a six-step procedure for removing the oxygen insert units from the passenger overhead environmental panels. Step 2 of that removal procedure states, "If generator has not been expended, install safety cap over primer." ValuJet work card 0069 refers to this maintenance manual chapter. Work card 0069 also delineates a seven-step process for removal of a generator. Step 2 states, "If generator has not been expended, install shipping cap [same as a safety cap] on firing pin.n

Work card 0069 and both relevant chapters of the Douglas MD-80 maintenance manual (chapters 35-22-01 and 35-22-03) contained warnings that generators, when activated, generate case temperatures up to 500 degrees F. The warnings also advised individuals to use extreme caution while handling the generators. Additional warnings in chapter 35-22-01 of the Douglas MD-80 maintenance manual call for individuals to "obey the precautions" and to refer to the applicable material safety data sheet (MSDS) for more precautionary data and approved safety equipment. According to SabreTech, the MSDS for the Scott Aviation oxygen generator was not on file at SabreTech's Florida facility at the time the generators were removed.

Neither the work card nor maintenance manual chapter 35-22-03 (the only maintenance manual chapter referenced by ValuJet work card 0069) gave instructions on how to store unexpended generators or dispose of expended canisters.

MAINTENANCE TASKS: About the middle of March 1996, SabreTech crews began replacing the expired and near-expired generators with new generators. According to the SabreTech mechanics, almost all of the expired or near-expired oxygen generators removed from the two airplanes were placed in cardboard boxes, which were then placed on a rack in the hangar. However, some of these generators (approximately a dozen) were not put in boxes, but rather were left lying loose on the rack.

According to the mechanics, when an oxygen generator was removed from an insert, a green SabreTech "Repairable" tag (Form MO21) was attached to the body of the generator (although one mechanic stated that he ran out of green tags and put white "Removed/Installed" tags on four to six generators). In the "reason for removal. section, near the bottom of the green "Repairable. tag, the mechanics made various entries such as "outdated," "out of date," and "expired," all indicating that the generators had been removed because of a time limit or date being exceeded.

Of the approximately 144 oxygen generators removed from N803VV and N802VV, approximately six were reported by mechanics to have been expended. There is no record indicating that any of the remaining approximately 138 oxygen generators removed from these airplanes were expended.

According to the corporate director for quality control and assurance at SabreTech, 72 individuals logged about 910 hours against the work tasks described on work card 0069. SabreTech followed no consistent procedure for briefing incoming employees at the beginning of a new shift, and had no system for tracking which specific tasks were performed during each shift.

The mechanic who signed work card 0069 for N802W further stated that he was aware of the need for safety caps and had overheard another mechanic who was working with him on the same task talking to a supervisor about the need for caps. This other mechanic stated in a post accident interview that the supervisor told him that the company did not have any safety caps available. The supervisor stated in a post accident interview that his primary responsibility had been issuing and tracking the jobs on N802W and that he did not work directly with the generators. He stated that no one, including the mechanics who had worked on the airplanes, had ever mentioned to him the need for safety caps.

The mechanic who signed work card 0069 for N802W said that some mechanics had discussed using the safety caps that came with the new generators, but the idea was rejected because those caps had to stay on the new generators until the final mask drop check was completed at the end of the process. He also said that he had witnessed both the intentional and accidental activation of a number oxygen generators and was aware that they generated considerable heat. When asked if he had followed up to see if safety caps had been put on the generators before the time he signed off the card, he said that he had not.

According to this mechanic, there was a great deal of pressure to complete the work on the airplanes on time, and the mechanics had been working 12-hour shifts seven days per week.

The mechanic who signed work card 0069 for N803W stated that he and another mechanic cut the lanyards from the 10 generators that he removed to prevent any accidental discharge, and then attached one of the green "Repairable. tags. He stated that he didn't put caps on the generators, but placed the generators into the same cardboard tubes from which the new ones had been taken. He then placed the cardboard tubes containing the old generators into the box in which the new generators had arrived. He said that he placed them in the box in the same upright position in which he had found the new generators. He said that although he did not see any of the generators discharge, he had worked with them at a previous employer and was aware that they were dangerous. This mechanic stated that his lead mechanic instructed him to "go out there and sell this job," which the mechanic interpreted as meaning he was to sign the routine and non-routine work cards and get an inspector to sign the non-routine work card. He said he looked at the work that had been done on N803W, focusing only on the airworthiness of that airplane.

Of the four individuals who signed the "All Items Signed" block on the subject ValuJet 0069 routine work cards and the "Accepted By Supervisor. block on the SabreTech non-routine work cards for N802W and N803W, three stated that at the time the generators were removed and at the time they signed off on the cards, they were unaware that the need for safety caps was an issue. However, the SabreTech inspector who signed off the "Final Inspection. block of the non-routine work card for N802W, said that at the time he was aware that the generators needed safety caps. He further stated that he brought this to the attention of the lead mechanic on the floor at the time (but could not recall who that was), and was told that both the SabreTech supervisor and the ValuJet technical representative were aware of the problem and that it would be taken care of "in stores," the air carrier's parts department. According to hire, after being given this reassurance, he signed the card.

SHIPPING: By the first week in May, 1996, most of the expired and near-expired oxygen generators had been collected in five cardboard boxes. Three of the five boxes were taken to the ValuJet section of SabreTech's shipping and receiving hold area by the mechanic who said that he had discussed the issue of the lack of safety caps with his supervisor. According to the mechanic, he took the boxes to the hold area at the request of either his lead mechanic or supervisor. He said that he placed the boxes on the floor, near one or two other boxes, in front of shelves that held other parts from ValuJet airplanes. He stated that he did not inform anyone in the hold area about the contents of the boxes. It could not be positively determined who took the other two boxes to the hold area.

According to a SabreTech stock clerk, on May 8, he asked the director of logistics, "How about if I close up these boxes and prepare them for shipment to Atlanta. He stated that the director responded, "Okay, that sounds good to me.. The stock clerk then reorganized the contents of the five boxes by redistributing the number of generators in each box, placing them on their sides end-to-end along the length of the box, and placing about two to three inches of plastic bubble wrap in the top of each box. He then closed the boxes and to each applied a blank SabreTech address label and a ValuJet COMAT label with the notation "aircraft parts." According to the clerk, the boxes remained next to the shipping table from May 8 until the morning of May 11.

According to the stock clerk, on the morning of May 9 he asked a SabreTech receiving clerk to prepare a shipping ticket for the five boxes of oxygen generators and three DC-9 tires (a nosegear tire and two main gear tires). According to the receiving clerk, the stock clerk gave him a piece of paper indicating that he should write "Oxygen Canisters - Empty on the shipping ticket. The receiving clerk said that when he filled out the ticket, he shortened the word "Oxygen" to "Oxy" and then put quotation marks around the word "Empty." He then completed the ticket and put the date (5/10/96) on the date line at the top of the form. He also said that- after finishing the ticket, he was asked to put ValuJet's Atlanta address on eight pieces of paper and to attach one to each of the boxes and tires. The receiving clerk stated that when the stock clerk asked for his assistance, the boxes were already packaged and sealed, and he did not see the contents.

According to the stock clerk, he identified the generators as "empty canisters. because none of the mechanics had talked with him about what they were or what state they were in, and that he had just found the boxes sitting on the floor of the hold area one morning. He said he did not know what the items were, and when he saw that they had green tags on them, he assumed that meant they were empty. The stock clerk stated in post accident interviews that he believed green tags indicated that an item was "unserviceable," and that red tags indicated an item was Beyond economical repair" or "scrap.. When asked if he had read the entries in the "Reason for Removals block on these tags, he said that he had not.

According to the stock clerk, he weighed the boxes and determined that each one was 45 to 50 pounds. He stated he asked a SabreTech driver, once on May 10, and again on the morning of May 11, to take the items listed on the ticket over to the ValuJet ramp area. He said that the driver was busy on May 10, and was not able to load and deliver the items until May 11.

According to the SabreTech driver, on May 11, the stock clerk told him to take the three tires and five boxes over to the ValuJet ramp area. He said that he then loaded the items in his truck, proceeded to the ValuJet ramp area, where he was directed by a ValuJet employee (ramp agent) to unload the material onto a baggage cart. He put the items on the cart, had the ValuJet employee sign the shipping ticket, and returned to the SabreTech facility.

According to the ValuJet ramp agents who loaded cargo bins #1 and #2 of the forward cargo compartment on flight 592, bin #2 was loaded with passenger baggage until full. Bin #1 was loaded with passenger baggage and U.S. mail (62 pounds), which included a mailing tube, a film box, and one priority mail bag. These items were followed by the three tires and the five cardboard boxes of oxygen generators. According to the lead ramp agent, who remained outside the airplane when the tires and boxes were loaded, "[the boxes] were placed on the side of the tires, facing the cargo door." According to the ramp agent inside the cargo compartment when the boxes were being loaded, "I was stacking the boxes on the top of the tires." The ramp agent testified at the Safety Board's public hearing that he remembered hearing a "clink. sound when he loaded one of the boxes and that he could feel objects moving inside the box. The ramp agent said that the cargo was not secured, and that the cargo compartment had no means for securing the cargo. It could not be determined whether any other items, such as gate checked baggage, were subsequently loaded into bin #1 before flight 592 departed.

PERSONNEL INFORMATION: The Captain: The captain, age 35, held an airline transport pilot (ATP) certificate with an airplane multi-engine land rating and type ratings in the DC-9, B-737, SA-227, and BE-1900. She also held flight instructor, ground instructor, and ATC tower operator certificates. The captain's first class medical certificate was current with no limitations.

According to company records, the captain had accumulated 8,928 total flight hours before the accident flight, of which 2,116 hours were in the DC-9 and 1,784 hours were as DC-9 PIC.

ValuJet records indicated that on September 23, 1995, while serving as PIC of a ValuJet flight that departed DFVV, the captain experienced an emergency that was later determined to have involved an overheated air conditioning pack. According to the incident report filed by the captain, flight attendants notified the flightcrew of smoke in the cabin shortly after takeoff. The captain stated in her report that the flightcrew could smell smoke in the cockpit. She stated, "the crew suspected a bleed air problem, but had no time to troubleshoot, since smoke was reported and the threat of a fire existed. It was felt Believed] that the safest course of action was to get on the ground as soon as possible." According to the first officer of that flight, he and the captain discussed whether to don their oxygen masks and smoke goggles as they maneuvered to descend and return to the airport. They decided that the situation did not warrant donning the masks or goggles. According to the first officer, no visible smoke was in the cockpit, although they could smell smoke. The airplane returned safely to DFW.

First Officer: The first officer, age 52, held an ATP certificate with ratings for airplane single-engine and multi-engine land, and a type rating in the DC-9. He also held flight engineer and airframe/powerplant (A&P) mechanic certificates issued by the FAA.

The first officer held a restricted FAA first class medical certificate. FAA records indicated that the FAA Aeromedical Certification Division was monitoring the first officer for a self-reported history of diabetes (a disqualifying condition for an unrestricted medical certificate). These records also indicated that he was taking the medication Diabeta, to lower his blood sugar levels.

According to company records, the first officer had accumulated 6,448 total flight hours as a pilot before the accident flight. (His ValuJet employment application also cited 5,400 hours as a military and civilian flight engineer.) He had 2,148 hours of DC-9 experience, including 400 hours as MD-80 international relief captain.

WRECKAGE AND IMPACT INFORMATION: The primary impact area was identified by a crater in the mud and sawgrass. The crater was about 130 feet long and 40 feet wide. Most of the wreckage debris was located south of the crater in a fan shaped pattern, with some pieces of wreckage found more than 750 feet south of the crater.

The majority of the wreckage was recovered by hand and placed on airboats that transported the pieces to a nearby levee for decontamination. The pieces were then transported by enclosed truck to a hangar for examination

The airplane structure was severely fragmented. In general, fewer pieces of right side forward fuselage skins were identified, and pieces from the right side were generally more fragmented. The majority of identified pieces were from the wing and fuselage aft of the wing box.

Examination of the engines revealed no signs of inflight or preimpact failure.

The tires and wheel assemblies from the landing gear system of the accident were recovered. The tires exhibited numerous rips and tears. Main landing gear actuators were found in positions corresponding to retracted landing gear.

The majority of both the left and right wings were recovered.

Most of the right and left horizontal stabilizers were recovered in fragments, including center sections, spars, skin panels, and both hinge fittings. No marks were found to identify pitch trim or elevator orientation at the time of impact with the swamp.

Several pieces of the rudder were recovered. The largest piece measured 57 inches by 43 inches. The preimpact position of the rudder was not determined.

Passenger service units from the cabin were found with the oxygen masks in the stowed positions.

Three hand-operated fire extinguishers were found, all with severe impact damage. Because of the impact damage, laboratory analysis could not positively determine if the extinguishers had been used.

FORWARD CARGO COMPARTMENT: All recovered wreckage identified as being from the area of the forward cargo compartment was assembled into a full-scale, three-dimensional mockup. These pieces included the cargo floor, cargo liners, and fuselage structure. They exhibited soot and heat damage.

About 50 percent of the forward bulkhead and about 25 percent of the aft bulkhead of the forward cargo compartment were recovered.

Recovered airplane wiring was examined for heat and fire damage and evidence of arcing. Heat and fire damage was observed on many of the wire bundles and cables that ran adjacent to the forward cargo compartment. The heat-damaged wires and cables showed no evidence of electrical arcing, and the burn patterns on those wires and cables were consistent with those resulting from an external heat source.

ANALYSIS: CARGO COMPARTMENT: Although class D cargo compartment are designed accident and events before this accident illustrate that some cargo, specifically oxidizers, can generate sufficient oxygen to support combustion in the reduced ventilation environment of a class D cargo compartment. The in-flight fire on American Airlines flight 132, a DC-9-83, on February 3, 1988, clearly illustrated the need for systems that would provide flightcrews with the means to detect and suppress fires in the cargo compartments of airplanes. As a result of its investigation of that accident, the Safety Board recommended that the FAA require fire/smoke detection and fire extinguishment systems for all class D cargo compartments. The FAA responded, stating that fire/smoke detection and fire extinguishment systems were not cost beneficial, that it did not believe that these systems would provide a significant degree of protection to occupants of airplanes, and that it had terminated its rulemaking action to require such systems. The Safety Board concluded that had the FAA required fire/smoke detection and fire extinguishment systems in class D cargo compartments, as the Safety Board recommended in 1988, ValuJet flight 592 would likely not have crashed. Therefore, the failure of the FAA to require such systems was causal to this accident.

The crash of ValuJet flight 592 prompted the FAA to state in November, 1996, that it would issue an NPRM (Notice of Proposed Rulemaking) by the end of the summer of 1997 to require, on about 2,800 older aircraft, the modification of all class D cargo compartments to class C compartments, which are required to have both smoke detection and fire extinguishment systems. The accident also prompted the airline industry group ATA to announce in December, 1996, that its members would voluntarily retrofit existing class D cargo compartments with smoke detectors. As of mid-1997, the Safety Board was unaware of any airplanes that have been modified and are in service.

On June 13, 1997, the FAA issued an NPRM that would require the installation of smoke detection and fire suppression systems in class D cargo compartments. According to the NPRM, the airline industry would have 3 years from the time the rule became final to meet the new standards. The FAA indicated that it anticipated issuing a final rule by the end of 1997. The Safety Board is disappointed that more than one year after the ValuJet crash and nine years after the American Airlines accident at Nashville, the class D cargo compartments of most passenger airplanes still do not have fire/smoke detection or suppression equipment and there is no requirement for such equipment. Recent incidents of continued shipment of undeclared hazardous materials, including oxygen generators, highlight the importance of getting the fire safety equipment installed as rapidly as possible. Therefore, the Safety Board believes that the FAA should expedite final rulemaking to require smoke detection and fire suppression systems for all class D cargo compartments.

FLIGHTCREW DECISIONS AND ACTIONS: Beginning at 1410:12, the flightcrew noted and verbalized concerns about electrical problems.

Based on the shouts from the passenger cabin recorded by the CVR cockpit area microphone at 1410:25 end the comment two seconds later, "we're on fire, we're on fire," it should have been clear to both flightcrew members that a very serious emergency situation existed in the cabin. Although the captain decided immediately to return to Miami and initiated a descent, for the next 80 seconds the airplane continued on a northwesterly heading (away from the Miami airport) while the flightcrew accepted ATC vectors for a wide circle to the left and a gradual descent back toward Miami.

The Safety Board evaluated the electrical system, engine, and flight control malfunctions that occurred in the 80 seconds during which the airplane continued northwestward, away from MIA. The electrical problems that first made the flightcrew aware of the emergency (at 1410:12) likely were the result of insulation burning on wires in the area of the cargo compartment. Electrical system wiring is routed outside of the cargo compartment of the DC-9, in accordance with federal regulations which require the wiring not be located against the cargo compartment liner and to incorporate a high temperature insulation. Therefore, the flightcrew's comments about the electrical problems indicate that the fire had probably already escaped the cargo compartment by 1410:12. (However, it probably had not yet burned through the cabin floorboards.) The flightcrew comments recorded by the CVR from 1410:12 through 1410:22 reflect the pilots' concerns about and attention to these electrical problems. It is possible that these concerns continued to occupy some of the pilots' attention during the initial period of their attempt to return to the ground.

Another malfunction began at 1410:26, just as the shouts from the cabin would have alerted the flightcrew to the seriousness of the fire there. According to FDR data, while the left engine remained at its previous EPR setting, the right engine's EPR decreased to the flight idle value. The reduction in thrust would likely have been an intentional act by the flightcrew to reduce power for the descent to return to the ground. The activation of the landing gear warning horn at 1410:28 suggests that the flightcrew had reduced power to idle (the warning horn is activated by one or both throttle levers being positioned at approximately the flight idle position). Because the flightcrew would not have intentionally reduced thrust on one engine only, they must have been unable to reduce the thrust on the left engine because of fire damage to the engine control located above the compartment. The inability to reduce left engine thrust could have distracted the flightcrew.

Further, the thrust asymmetry continued throughout the period and resulted in a sideslip and lateral accelerations that were not corrected with rudder application. Therefore, left-wingdown (LWD) aileron deflections would have been necessary to keep the airplane from rolling to the right. Because there were no right roll indications in the FDR heading data, the flightcrew must have been applying the LWD control inputs.

The FDR indicates that at 1411:20, vertical acceleration increased to about 1.4 G. although the control column had not moved. Subsequently, the control column position was moved forward about 5 degrees to reduce the vertical acceleration back to 1 G. At this time, the airplane leveled temporarily at about 9,500 feet. These events indicate that the flightcrew was confronted with a disruption in pitch control (m the elevator or trim systems), and was active in maintaining at least partial control of the airplane. The pilots could have found the disruption in control to be distracting, and the level off is consistent with their attempts to handle the pitch controls carefully. The development of malfunctions from the electrical system to engine thrust controls and flight controls indicates that the flight experienced a progressive degradation in the airplane's structural integrity and flight controls.

At 1412:00, FDR-recorded altitude suddenly decreased and no longer agreed with the altitudes recorded from radar transponder returns (these altitudes are derived from different static sources). The disagreement between altitude values indicates that the fire damage continued to increase.

Radar data show that at 1412:58, when the airplane was at 7,400 feet, it began a steep lefiturn toward Miami and a rapid descent. For the next 32 seconds, the descent rate averaged about 12,000 feet per minute, and the airplane turned from a southwesterly heading toward the east. If asymmetric thrust were providing right yaw/rolling moments during this turn, the flightcrew would have had to counter this tendency with continuing left roll control inputs throughout the turn. The radar data indicated that the left turn then stopped on a heading of about 110 degrees at 1413:25, which was toward MIA. Further, the rapid descent rate was being reduced, with the last transponder-reported altitude at 900 feet. The control inputs required to balance asymmetric thrust during the steep left turn, followed by the level-off, indicates that the flightcrew initiated a turn and descent, and that the captain and/or the first officer were conscious and applying control inputs to stop the steep left turn and descent (until near 1413:34). Thus, the airplane remained under at least partial control by the flightcrew for about 3 minutes and 9 seconds after 1410:25.

Ground scars show that the airplane was in a large right roll angle and steep nose-down attitude at impact. To achieve that attitude and fly through the position indicated by the primary radar return at 1413:39, the airplane would have had to start rolling to the right at 1413:34, at least 8 seconds before the crash.

Because of the lack of evidence from the CVR, FDR, and the wreckage, the Safety Board was unable to determine with certainty the reason for the loss of control that occurred at that time. However, examination of the wreckage showed that before the impact the left side floor beams melted and collapsed, which would likely have affected the control cables on the captain's side. It is possible that the first officer might have taken over flying from the captain, but the remaining control cables also were possibly affected by distorted floor beams. Based on the continuing degradation of flight controls and the damage to cabin floorboards in the area of the flight controls, the Safety Board concluded that the loss of control was most likely the result of flight control failure from the extreme heat and structural collapse; however, the Safety Board could not rule out flightcrew incapacitation during the last seven seconds of the flight.

ValuJet emergency procedures for handling smoke and fire uniformly instructed the pilots to put on their oxygen masks and smoke goggles, as the first item to be performed on the emergency checklist. However, the flightcrew comments recorded on the CVR sounded unmuffled. Further, these comments were recorded on the cockpit area microphone channel of the CVR; this microphone would not have picked up verbalizations made under an oxygen mask. This indicates that neither the captain nor the first officer donned their oxygen masks during the period of the emergency in which the CVR was operative and the pilots were speaking. The last recorded verbalization by the captain was at 1410:49; the last by the first officer was at 1411:38. Because smoke goggles of the type provided to the flightcrew must be put on after the oxygen mask to have any effect, the pilots probably did not put on their smoke goggles from the onset of the emergency, at 1410:07, through at least 1411:38. There is no evidence to indicate whether they donned their masks end goggles after 1411:38.

The donning of oxygen masks and smoke goggles at the first indication of smoke anywhere in the airplane can provide flightcrews with a sustained ability to breath and see in the event of a subsequent influx of smoke into the cockpit. Although in this accident the donning of oxygen masks and smoke goggles would not have assisted the crew in the initial stages of the emergency (because of the absence of heavy smoke in the cockpit), early donning of the smoke protection equipment might have helped later in the descent, if heavy smoke had entered the cockpit.

In an informal survey conducted by the Safety Board, pilots from several air carriers indicated that they would not don their oxygen masks and smoke goggles for situations such as reports of a galley fire, smoke in the cabin, or a slight smell of smoke in the cockpit. Based on the circumstances of this accident and the results of its survey, the Safety Board concludes that there is inadequate guidance for air carrier pilots about the need to don oxygen masks and smoke goggles immediately in the event of a smoke emergency.

Based on the Safety Board's simulator evaluation of the equipment furnished to the flightcrew of ValuJet flight 592 and its informal survey of air carrier pilots, the Board concludes that smoke goggle equipment currently provided on most air carrier transport aircraft requires excessive time, effort, attention, and coordination by the flightcrew to put on. The Safety Board believes that the FAA should establish a performance standard for the rapid donning of smoke goggles and ensure that all air carriers meet this standard through improved smoke goggle equipment, improved flightcrew training, or both.

During its investigation, the Safety Board learned that many current installations of smoke goggles at a variety of U.S. air carriers place their goggles within sealed plastic wrapping, and this wrapping is sufficiently thick such that it cannot be easily opened (without using one's teeth to tear the plastic material or requiring the pilot to obtain and manipulate a sharp object and devote both hands to opening the bag). The Safety Board is concerned that flightcrews attempting to put on these smoke goggles in an emergency might be unable to open the wrapping material quickly because the configuration of the equipment requires that the oxygen mask be secured over the pilot's face before attempting to don the smoke goggles. The Safety Board concludes that the sealed, plastic wrapping used to store smoke goggles in much of the air carrier industry poses a potential hazard to flight safety.

PROBABLE CAUSE: The National Transportation Safety Board determined that the probable causes of the accident, which resulted from a fire in the airplane's class D cargo compartment that was initiated by the actuation of one or more oxygen generators being improperly carried as cargo, were: (1) the failure of SabreTech to properly prepare, package, and identify unexpended chemical oxygen generators before presenting them to ValuJet for carriage; (2) the failure of ValuJet to properly oversee its contract maintenance program to ensure compliance with maintenance, maintenance training, and hazardous materials requirements and practices; and (3) the failure of the Federal Aviation Administration (FAA) to require smoke detection and fire suppression systems in class D cargo compartments.

Contributing to the accident was the failure of the FAA to adequately monitor ValuJet's heavy maintenance programs and responsibilities, including ValuJet's oversight of its contractors, and SabreTech's repair station certificate; the failure of the FAA to adequately respond to prior chemical oxygen generator fires with programs to address the potential hazards; and ValuJet's failure to ensure that both ValuJet and contract maintenance facility employees were aware of the carriers "no-carry" hazardous materials policy and had received appropriate hazardous materials training.