Accident Involving Air Tahoma, Inc., Flight 185 N586P August 13, 2004
This paper will examine the accident involving Air Tahoma Flight 185 at Covington, Kentucky. The area of examination will discuss the human factors involved in the pilot’s actions, the maintenance actions, the airliner’s training, and the design flaws highlighted in this accident. As with any aviation accident, there is a chain of events which led to the occurrence; each link in the chain will be highlighted.
Breaking the sequence down into constituent parts allows an understanding of all the factors involved. The paper will go beyond the simple statement of causes as determined by the National Transportation Safety Board (NTSB) and will cover topics of accident prevention that apply to the entire safety culture.
Keywords: aircraft, accident, fuel system mismanagement, design flaw, Convair 580
Accident Involving Air Tahoma, Inc., Flight 185 N586P August 13, 2004
Every aircraft accident is characterized by a chain of events, breaking any one of the links in the chain the accident would have been prevented. Reviewing the NTSB aircraft accident report (AAR 0603 – hereinafter referred to as “the Report”) shows that this accident matched the above generalization. In reviewing such accidents going beyond the simple statement of “pilot error”, the obvious question is why did the pilot do what he did. This is the essence of human factors theory and practice.
This accident was not the first fuel starvation forced landing accident. On 30 December 1964 a Convair 340, operated by United airlines, registration N73102, landed wheels up after the flameout of both engines. It has proved impossible to acquire the accident report from archives this old. It is disappointing that the lessons learned from this previous accident were not applied thereby preventing the subsequent accident near Cincinnati.
It will be useful to define what is meant by fuel cross feed and what is meant by the fuel transfer. Fuel cross feed is only intended to correct an imbalance between two tanks or two groups of tanks. In many installations this is achieved simply by opening a cross feed valve and switching off the fuel boost pump in the tank with the higher fuel contents. The remaining fuel boost pump feeds both engines and the pilot is required to observe the fuel quantity restoring normal operation when the fuel is balanced. Fuel transfer, on the other hand, is often achieved by a failsafe-protected system whereby a separate transfer pump allows fuel to be pumped from one tank to another. Because fuel transfer would be possible, in theory, to completely empty a tank there are limit switches that prevent over fueling of one group of tanks. Fuel transfer systems are rarely fitted to normal passenger or freight aircraft, and are usually only found on larger more complex aircraft such as military fuel tankers, or fitted to aircraft, such as the Concorde, which require center of gravity fuel transfer for supersonic flight.
On August 13, 2004, about 0049 Eastern daylight time, Tahoma, Inc., Flight 185, a Convair 580, N586P, crashed about 1 mile south of Cincinnati Northern Kentucky International airport (CVG), Covington, Kentucky, while on approach to runway 36R. The first officer was killed, and the captain received minor injuries. The airplane was destroyed by impact forces. The flight was operating under the provisions of 14 Code Of Federal Regulations part 121 as a cargo flight for DHL Express from Memphis International airport, Memphis, Tennessee, to CVG. The National Transportation Safety Board (NTSB) determined that the probable cause of this accident was fuel starvation resulting from the captain’s decision not to follow approved fuel cross-feed procedures.
Contributing to the accident where the captains inadequate pre-flight planning, his subsequent distraction during the flight, and his late initiation of the In-range checklist. Further contributing to the accident was the flight crew’s failure to monitor the fuel gauges and to recognize that the airplane’s changing handling characteristics were caused by a fuel imbalance.
This paper will be divided into the following sections:
- Bad Design
- Bad Certification
- Bad Modification
- Bad Maintenance
- Bad Regulation
- Bad CRM
To an outside observer the design of a fuel system that is capable of causing structural damage or of venting half the contents of the of the fuel system overboard cannot be safe. In addition, what is often referred to as tribal knowledge seems to have taken part in this accident.
The aircraft manufacturers approved flight manual says:
- “place boost pump switch for tank being used to the on the position and place the cross feed switch to the on position,
- place the boost pump switch for the tank not being used to the off position, and
- place the fuel shutoff valve switch for the tank not to being used to the closed position.”
Air Tahoma fuel cross feed procedures states:
To cross freed from either tank to opposite engine;
- both fuel boost pumps ON,
- both aircraft fuel boost pressure lights OFF,
- cross feed valve OPEN,
- fuel boost pump for tank not to be used OFF.
- Tank shutoff valve for tank not used OFF
Evidence was presented that pilots in general do not trust a fuel shutoff valve quoting valve failure as a reason for not using it. How this myth became so prevalent is perhaps based on the fact that the fuel shutoff valve and the fuel cross feed valve are both motor operated valves, which remain in their last position if power is lost. (Figure 1.) Therefore closing the fuel shutoff valve before an unexpected electrical power failure will mean the loss of half of the aircraft available fuel. Little wonder that pilots did not trust the fuel shutoff valve, and that one operator, the pilot’s previous employer, even stated in their checklist, “fuel shutoff valve close at pilots discretion.” Many fuel systems do not have a shut-off valve such as in this installation. A lack of a failsafe fuel valve must be considered a design flaw.
The report states that a one-way check valve was not fitted into this fuel system. “One-way check valves are not typically installed on airplanes, and the accident airplane did not have one installed. According to a Kelowna Flightcraft representative, one Convair 580 operator modified its fleet of about 30 Convair 580 airplanes in the 1960s under an engineering order by installing a one-way check valve, which prevented fuel from flowing back into the fuel tanks when the fuel tank shutoff valve was left open. However, the representative stated that this operator was no longer in business and that its airplanes had represented a small percentage of the entire Convair 580 fleet.”
It is clear that at least one operator considered this design flaw merited the fitting of a one-way check valve to prevent inadvertent transfer of fuel. In addition they retro-fitted warning lights to show the status of the cross feed valves and the shut off valves.
The position of the fuel gauges is also an example of bad design, bad ergonomics. With a seat in the aft position, often where non-flying pilots sit, meant that it was not easy at all for the captain to view the fuel gauges. (Figure 3.)
The design flaw and lack of failsafe having been certified by the FAA must, by definition, be a failure in that certification process. To simply accept an aircraft as airworthy by accepting procedures to cope with a non-safe installation is simply not good enough. Had the certification personnel checked with global regulatory bodies, they would have found widespread requirements to fit one way check valves to fuel systems.
PJCB 10-21, “Aircraft Fuel Boost Pump Output Pressure Limit-Reduce,” which was published in October 1969, provided details on an optional procedure that allowed Convair 580 operators to reduce the typical fuel boost pump output pressure setting of 21 psi to 15 psi to “improve the service life of the aircraft fuel boost pump.” This cost saving measure has now been shown to have introduced an unintended consequence, that differential fuel boost pump output pressure settings would guarantee that fuel would transfer from one tank to the other should the cross feed valves be left open. The service bulletin contained a recommendation that aircraft should preferably be operated with identical boost pump pressure; there was no requirement for this to be done.
Another incident of inadvertent fuel transfer had occurred to this operator on September 21, 2004. No evidence was given as to what modification state the incident aircraft had, other than post-incident bench testing showing a differential pressure of 15 and 21 psi between the two pumps.
Air Tahoma maintenance personnel reported that they were not aware of the service bulletin to lower the fuel boost pump output pressure setting to 15 PSI. In June 2004, maintenance and replaced the left fuel boost pump on the excellent airplane with a pump that have them output pressure setting of 21 PSI. However, they did not replace the right fuel boost pump and did not measure or alter the output pressure setting. As a result, they were unaware that it was operating the airplane with left and right differential fuel boost pump output pressure settings.
Maintenance procedures did not include a reference to the service bulletin, and only required setting of the fuel boost pump pressure to 21 PSI, as opposed to the lower service bulletin setting of 15 PSI.
Only 4 months before the accident, there had been the in-flight fuel imbalance incident to Nolinor Aviation. Maintenance personnel had changed a pump inadvertently introducing differential fuel pressure and had left the cross feed valve open. Other than switch position there is no cockpit indication that a cross feed valve is open.
The NTSB recommendations included the requirement for operators to set the left and right fuel boost pump pressure to the same setting. This recommendation clearly addressed the issues of poor maintenance.
It is the duty of the regulatory body, the Federal Aviation Administration (FAA) to regulate and thereby ensure safety of flight. In particular, Principle Operations Inspectors are required carry out specific functions including approval or acceptance of a certificate holder’s operating procedures. Experience has shown that there is a lack of standardization across an otherwise well motivated and experienced pool of such inspectors. Company checklists which include the words “at the captain’s discretion” having been approved or accepted by one inspector caused this particular pilot to ignore a critical step on his checklist. Whilst these approvals are well intended, the unintended consequences in this example show that improvements to the methods and procedures used by the FAA require critical examination.
In addition, the FAA is required to certify airmen. That is to say conduct check rides issuing qualifications for types or ratings. In the report evidence was given that this pilot had reduced training, presumably due to his previous qualification on type, and a check airman signed this pilot’s type qualification for his new employer. The check airman is either employed by or appointed by the FAA, therefore responsibility must be partly apportioned to poor regulation.
The FAA defines Crew Resource Management (CRM) as, “The effective use of all resources to include human and other aviation system resources”. The history of accident prevention includes a time when training was primarily limited to the skills of flying and the technical knowledge in aircraft systems that were required to operate an aircraft. Realization that human factors were the main cause of accidents led to the development of CRM over several decades. This was because research looked beyond “what the pilot did” to understand “why the pilot did it”. (Salas p. 565)
This accident is typical and includes many of the elements that CRM attempts to address. The “insidious enemies” of erroneous “facts”, incorrect assumptions, miscommunications, misunderstandings and the individually generated disconnect between “situational awareness” and “truth” all play a part here. (Salas p.564) Examining each of these elements in relation to the accident will, no doubt, give a clue as to the real causes.
The pilot believed that the fuel shut off valve was unreliable. It has already been speculated that this may have been simply because the valve was not fail-safe. It is also speculated that as a fraternity, the pilot population would in general be uncomfortable shutting off fuel supplies on an aircraft in flight. Whilst understandable it is, on the face of a lack of evidence of fuel shut off valve failures, one of those erroneous facts that pervades aviation. Any number of commonly held beliefs based on no real scientific evidence has led to many an error. Perhaps aircraft designers need to consider these “urban myths” when designing aircraft systems.
The pilot assumed that opening the cross feed valve and doing nothing else would have no consequence. As any pilot about a cross feed system and they would probably agree with this assumption. If both fuel pumps are running it is assumed that opening the cross feed valve will have no effect. Having no indication of fuel supply pressure downstream of the fuel pumps other than a pressure switch set to operate at a pressure determined by either the high or low setting, neither this pilot nor any pilot on this type of aircraft would have any reason not to make this assumption. It is clear that the setting of differential pressures on the two pumps made this assumption wrong.
In addition, there was evidence that the pilot believed there was a one way check valve in the system preventing the undesirable fuel transfer. One operator had fitted these check valves. Other aircraft have them as standard equipment. Therefore through incomplete or inadequate training on the systems this pilot was led to failure through his hazy notion that transfer was not possible, despite the warnings in the flight manual and on the placard in the aircraft.
This pilot was distracted throughout the flight. His concerns with an error on his weight and balance calculations led him to forgetting that he had started a fuel balance procedure. On six occasions the first officer complained about the handling of the aircraft. Post accident actions included the reminder that under the principles of CRM a crew member has a duty to be more assertive in expressing concerns. To be fair to the captain, the first officer’s words were ambiguous at best if not misleading at worst. The only reply from the captain was to have the controls checked on landing. There were clues which are obvious in hindsight given by the first officer, but not expressed in a way that would have reminded the captain of what was likely to be the cause of the control difficulties.
Disconnect between Situational Awareness and Reality
Given the erroneous facts above and the resultant incorrect assumptions it is clear that the captain was not aware of the realities of the situation. It is true that once a person has formulated a certain way of thinking about a problem, it appears difficult to get out of that way of thinking and try a different interpretation of the data. (Green P.61) This “confirmation bias” was clear in this instance. For all the reasons above the pilot could not make a new determination of the cause of the control difficulties reported by the first officer.
The biggest single failure on the part of this captain was not to have been more pro-active in responding to his first officer’s concerns. He did not, for example, ask any questions to clarify the information, nor did he take control of the aircraft to see for himself. This has to be the most surprising aspect of this accident. Given the timing of events, it is uncertain whether he would have concluded correctly that he had a fuel imbalance, but he should have done more to identify the problem.
This was a preventable accident. There are CRM lessons for all operators here, not only for pilots, but for maintenance, management, regulation and supervision. To misquote an old adage, “To err is human, to make the same mistake twice is unforgivable, to make the same mistake three times is unbelieveable”.
Air Safety Net Convair 580 hull write off database. Retrieved 29 February, 2012 from: http://av. iation-safety.net/database/dblist.php?field=typecode&var=164%&cat=%1&sorteer=datekey&page=1
Convair 580 Aircraft Data Retrieved 29 February, 2012 from: http://www.airliners.net/aircraft-data/stats.main?id=169
Green R.G, Muir H., James M., Gradwell, D., Green R.L, Human Factors for Pilots (2nd ed.) (1996) Aldershot: Avebury Aviation
NTSB Aircraft Accident Report NTSB/AAR-06/03 PB2006-910403 Notation 7778 Retrieved 29 February, 2012, from: http://www.ntsb.gov/doclib/reports/2006/aar0603.pdf
NTSB Brief Incident Report, CONVAIR CV-340, registration: N73102 Retrieved 29 February, 2012 from: http://www.ntsb.gov/aviationquery/brief.aspx?ev_id=77667&key=0
Salas E., Maurino D., Curtis M., Human Factors in Aviation (2nd ed.) (2010) Burlington: Elsevier
Seamster T. L., Boehm-Davis D. A., Holt R. W., Schultz K. Developing Advanced Crew Resource Management (ACRM) Training: A Training Manual 1998 Federal Aviation Administration Retrieved 29 February, 2012, from: http://www.hf.faa.gov/docs/dacrmt.pdf