‘Slackadaisical’ On-Demand Charter Operations Criticized

Safety Board also calls for a low airspeed alert before stall warning

A low airspeed alert is needed for all aircraft engaged in scheduled and on-demand commercial operations. Such a device might have helped prevent two substandard pilots from killing themselves and their six passengers, according to the findings of an investigation into a crash last year that killed a prominent politician.

The need for such a warning device emanates from the National Transportation Safety Board (NTSB) investigation into the Oct. 25, 2002, crash of a Beech King Air 100 that stalled and crashed into the woods just a few miles short of runway 27 at Eveleth, Minnesota. The Aviation Charter flight was carrying Sen. Paul Wellstone (D- Minn.), members of his family and campaign staff, on a flight from St. Paul. At the time of the accident, the company was operating a fleet of 24 charter aircraft of various types.

The airplane burst into a fireball on impact. Since the aircraft was not equipped with cockpit voice or flight data recorders, the 10:22 a.m. time of impact was determined from the radar plot of the aircraft’s final flight path. When soot was smeared away from the crystal of a watch worn by one of the victims, the stationary hands showed a time of about 10:30 a.m.

The flight had departed St. Paul at 9:37 a.m., with instrument flight rules (IFR) and icing conditions (light to moderate rime icing between 5,000 to 11,000 ft.) current over the flight’s planned route.

At the final Nov. 18 hearing into the tragedy, NTSB officials ruled out in-flight icing as a significant factor in the accident. Rather, investigators found that the pilots were engaged in an unstabilized approach in which speed steadily decayed to a point where the aircraft stalled.

The case played out against the backdrop of an unscheduled charter operation with few of the safety programs, managerial defenses-in-depth and regulatory oversight characteristic of scheduled airline operations. The marginal managerial milieu set the stage for two average, if not below-average, pilots to undertake a flight that should not have ended fatally.

The vertical component

As the aircraft descended through clouds to complete its flight, it spent only some two minutes in moderate icing conditions.

“We believe heavy icing did not occur in this scenario,” said the NTSB’s Charles Pereira. He served as aircraft performance group chairman for the investigation. If heavy icing had occurred, Pereira explained, the accident aircraft would have stalled at an airspeed well above the 77 knots at which it evidently did stall. It should be noted that there are events on record where two minutes of exposure to icing has been enough to cause a huge increase in drag.

NTSB meteorological group leader Kevin Petty said three additional factors rule out a heavy buildup of ice on the airplane. First, the aircraft spent “a minimal amount of time in icing conditions.” Second, the aircraft was well equipped to deal with icing (e.g., de-icing boots). Third, the aircraft spent more than four minutes flying in conditions where any ice buildup would have been shed from the airplane.

The horizontal component

The pilots did not intercept the runway 27 VOR (VHF omnidirectional range) approach path at the desired 10-mile point, but rather flew a mile north of it. Even after they intercepted the centerline at the 5-mile point, the aircraft continued to deviate from the approach course. Low clouds might have prevented the crew from seeing the airport, although this factor would not be germane to an instrument approach.

Dave Kirchgessner, NTSB operations group chairman for the investigation, said that the airplane never maintained a proper stabilized approach at any time in the approach.

“The stabilized approach principle is critical for safe landing,” he stressed.

Not on-track, not on-speed and not stabilized at that point, company policy prescribed a go-around. The NTSB said conditions should have prompted a go-around: “The inadequate airspeed or the full course deviation indicator needle deflection should have prompted the flight crew to execute a go-around; however, they failed to do so.”

The airspeed component

Investigators criticized the pilots for failing to monitor their airspeed as it decreased to a dangerously low level. They said it fell some 50 knots below the company’s 130- knot recommended approach speed and was diminishing for some 50 seconds before the wings stalled.

The landing gear should have been extended at an altitude of about 3,500 feet but apparently was not until the airplane was at a height of about 2,200 feet. “The airplane never should have been flown below 130 knots with the landing gear not extended,” Pereira declared. Late extension of the landing gear, inferred from the available information since the airplane was not equipped with a flight data recorder (FDR), was further evidence of an unstabilized and rushed approach.

The airspeed needle unwound as speed declined through the operator’s recommended approach speed, through manufacturer Raytheon‘s [NYSE: RTN] recommended approach speed, through the minimum approach speed (92 knots), and through the 82-85 knot speed where stall warning would have been triggered (based on the airplane’s weight, center of gravity, and configuration).

Pereira estimated the two pilots would have had about 5-8 seconds warning of incipient stall.

The aircrew component

The picture painted by the NTSB of the pilots is of two individuals who had previously demonstrated “serious performance deficiencies … consistent with below-average flight proficiency.”

John Clark, NTSB head of aviation accident investigations, said the pilots’ last course correction would have required a turn of 10� to 20�. “Not a big deal, but his gear wasn’t down and his speed was high,” he said.

The 55-year old pilot in command, Richard Conry, worked as a part-time dialysis technician. After the accident, some fellow pilots interviewed by investigators expressed concerns about his flying skills. He tended to delegate the actual flying to the copilot. One pilot told investigators he’d had to take control from Corny, when he’d turned around to answer a passenger’s question and inadvertently allowed the aircraft to fall into a 45� bank with the airplane descending at 1,000 feet per minute (during cruise). Another pilot said he’d had to take control from Conry when he was unable to hold altitude on an instrument approach.

Yet Conry had passed a proficiency check just three days before the accident.

The copilot, 29-year old Michael Guess, was a graduate student building time in hopes of flying someday with the airlines, although he had flunked out of an airline aircrew ground training instructor program. The book on Guess was not any better than on Conry. He had to be reminded to keep one hand on throttles during an approach, tended to become fixated (stopping his instrument scan), and had a tendency to allow airspeed and torque to get too low.

Both were described as “non-assertive.”

Two unassertive types flying together. The clear impression is of an accident going somewhere to happen.

The company component

NTSB officials determined that the company was culpable in the accident. The lack of rigor at the management level was not cited as a contributing factor in the probable cause of the accident, which placed blame on the pilots. Nonetheless, NTSB investigators expressed dismay over inadequate training, inadequate and under-emphasized standard operating procedures, and inadequate recurrent training .

The Federal Aviation Administration (FAA) was also criticized for not looking closer at on-demand charter operations such as this. The FAA hadn’t been conducting en route inspections of flight operations, because it had not required them. For two-pilot operations, cockpit resource management (CRM) training still is not required for Part 135 operations, as it is for Part 121 carriers. Aviation Charter had been flying with two pilots in its aircraft ever since a 1997 accident, investigators noted, even though planes like the King Air 100 are certified for single-pilot operation. Inadequate CRM may have played a part in the fatal March 29, 2001, crash of an on-demand charter Gulfstream III jet at Aspen, Colorado. In the wake of that accident, CRM advocate Robert Baron said CRM training for on-demand charter operations should be a matter of course. “There is a higher likelihood of putting crews together with drastically different backgrounds and levels of experience, [and] the quality of aircraft-specific pilot training can vary,” he cautioned in a paper he produced on the subject earlier this year (see ASW, March 10).

In a telephone interview, Baron said, “CRM has not been mandated by the FAA for Part 135 operations, although it was recommended by the NTSB following the crash at Aspen.”

NTSB Chairman Ellen Engleman said the management and oversight shortcomings “sadly and starkly” revealed by the accident postmortem point out the need “for more aggressive action to improve safety in the on-demand charter industry.” Indeed, the findings in this case recall a 1998 study by Transport Canada (TC), that country’s counterpart to the FAA, on the safety of air taxi operations (SATOPS). The findings of the SATOPS report read very much like the situation found during the Wellstone crash inquiry.

A director of safety for a major U.S. air carrier said the NTSB might have understated the seriousness of the management deficiencies in the Wellstone crash case, and for Part 135 on-demand charter operations generally. Unlike for the Part 121 scheduled carriers, he said, “There was no requirement for the company to have a director of safety. There was no requirement for a safety reporting system.”

This accident, he said, “was preventable by techniques we know work, and which we have in place at the major carriers.” To this captain, “the same elements of management and oversight and safety should be in place, irrespective of the size of the airplane.”

The National Air Transportation Association (NATA), the public policy group representing the Part 135 industry, took exception to Engleman’s call for improving safety. NATA President James Coyne acknowledged the tragic loss of life in the Wellstone accident but said it must be considered in context. “It is misleading and wrong for the NTSB to characterize the entire on-demand air charter industry as unsafe because of one isolated incident, regardless of who was on board.”

“The outstanding safety record of the Part 135 industry speaks for itself,” Coyne declared. “Part 135 charter operators fly passengers millions of miles every year without a single incident or accident.”

The NTSB’s accident statistics database shows a significant difference in the accident rate for Part 121 scheduled carriers and Part 135 on-demand air taxis. For the latter, the accident rate for the five-year period 1998-2002, the most recent data available, is some eight times higher.

The picture that emerged from the NTSB hearing was one of “slackadaisical” management and oversight (a combination of inadequate regulatory standards and inadequate management and enforcement).

The crew alerting component

NTSB investigators said they found 18 cases – involving 12 killed and four planes destroyed – since 1982 where pilots failed to maintain proper airspeed during approach and landing. These cases excluded icing events, some short-field Alaska operations and floatplanes, and cases with aggravating factors, such as sudden gusts and terrain/obstacle avoidance. Of these cases, 11 (60 percent) involved Part 135 operations.

“Staff believes it is time to take a hard look at low airspeed alerts,” Pereira said. “Stall warning and pilot vigilance are not enough.”

In the case at hand, Pereira said some 40 seconds elapsed between the time pilots Conry and Guess slowed below the company approved approach speed of 130 knots and the likely time the stall warning alert sounded, somewhere between an indicated airspeed of 82 to 85 knots. This being the case, they lost nearly 50 knots of airspeed without any apparent action to arrest the counterclockwise motion of the needle on their airspeed indicator.

Pereira said a low airspeed alert is required by airworthiness directive (AD) on the Embraer EMB-120 twin turboprop. In icing conditions with flaps retracted, the alert activates if airspeed decays to 160 knots.

Pereira said a low airspeed alert could be likened to a low altitude alert, for which ground proximity warning system (GPWS) technology has been installed to warn pilots.

Clark acknowledged there are a number of issues involved with a low airspeed alert but hastened to point out that “it’s already being done on the EMB-120.” He reminded chairman Engleman of her desire to see simple solutions for safety problems, and the low airspeed alert concept seems to accord with that perspective.

The board agreed, and recommended that the FAA:

“Convene a panel of aircraft design, aviation operations, and aviation human factors specialists … to determine whether a requirement for the installation of low-airspeed alert systems in airplanes engaged in commercial operations under … Parts 121 and 135 would be feasible, and submit a report of the panel’s findings.”

Indeed, groundwork already has been laid for a low airspeed alert in recent industry research and proposed rulemaking efforts spanning the 1996-2003 timeframe. For example, the FAA’s 1996 Human Factors Team Report intimated the need for better airspeed awareness before the stall warning system alarms:

“The HF [human factors] team is concerned that flightcrews may not be provided adequate awareness of airplane energy state, particularly when approaching or trending toward a low energy state.

“Transport category airplanes are required to have adequate warnings of an impending stall, but at this point the airplane may already be in a potentially hazardous low energy state. Better awareness is needed of energy state trends such that flightcrews are alerted prior to reaching a potentially hazardous low energy state.”

Two conceptual components

One industry observer suggested that the apparent incompetence of the pilots in the Wellstone crash case makes the case for establishing an experience and currency grading system for air transport and commercial rated pilots – otherwise customers can never be confident that they won’t also suffer Wellstone’s fate. This individual suggested a grading system where grade 1 would be a pilot with more than 3,000 hours on type and more than 300 hours on type in the last six months, with five instrument rating renewals. Grade 2 might be something less and grade 3 would be an ab initio pilot with less than 500 hours on that type of aircraft. Retention of grade would be the subject of FAA review after any incident.

In the present situation, standards are lacking, as evidenced by the thrust of the NTSB discussion. Simulators typically are not available for on-demand charter pilots and checkouts too often smack of a buddy system. Airplane knowledge may be scant, as was apparent in the 2001 crash of a King Air 200 (also with the ride-along copilot) that killed members of the Oklahoma State University basketball team. In that tragedy, too, another Part 135 operation, management deficiencies lay behind the aircrew shortcomings (see ASW, June 23).

Add the turboprop icing factor, pressure from management and/or passengers to complete a trip, the lack of sophisticated workload-reducing tools on the big jets and, to mix metaphors, the portrait too often is of a one-armed paperhanger who’s operating behind the eight ball as a matter of course. A grading system for experience and currency might materially redress the situation.

There is another potential factor that may be worth considering. Why would two pilots allow airspeed to remain slow and then decay to the point of stall without taking action to arrest and reverse the situation? It seems inconceivable that two pilots would mutually abide dangerously low airspeeds, approaching a 50-knot deficit. Maybe that was not what they were seeing. Could instrument error lie at the heart of such pilot passivity? If the aircraft was cold-soaked in the hour’s cruise at 13,000 feet, and any water trapped in the static line froze (but later melted on descent), the indicated airspeed would begin to overread during descent and approach (and so the speed, as measured by radar, would be inordinately low). The problem would not show up on the airspeed indicator (ASI) or altimeter during cruise. A tendency for the two cockpit ASIs to read a deceptively high airspeed could result even if the line-trapped ice melted during descent, sources say. Melted-ice water trapped at a low point in static lines seemingly would allow the airspeed to recover to normal (accurate) values during descent and the other pressure instruments (altimeter and vertical speed display) would appear to behave normally. But in a static line just a few millimeters in diameter, capillary action and meniscus formation of water trapped (vertically) in the lines might prevent the pressure sensed at the static ports from being reflected in full at the airspeed indicator.

To illustrate, think about how one can suck water up a straw and then keep it suspended there. There is a lower air-pressure on the tongue’s side of the straw than there is at the open end. Per the straw analogy, the air between the trapped line-water and the instruments can never totally catch up to the pressure increase resulting from the descent from the cruise level of 13,000 feet. Within that pressure discrepancy may lie an overreading error (of whatever residual magnitude). Avoiding such problems is why some aircraft static ports are heated and why manually operated drains are installed at low points in the static lines.

Because the margin between approach speed and stall speed is oftentimes just a few knots, an accurate airspeed reading is critical. If the airspeed indicator is reading too high, the pilot might fly too slowly and risk stalling – even though content that he’s indicating “on speed.” An airspeed indicator overreading 10-15 knots and the slender margin between a stall and any maneuvering for final line-up could be all that’s needed to push unsuspecting pilots into a low altitude unrecoverable stall. The NTSB database documents a history of inexplicably low airspeeds on descent and approach. If instrument error is at play here, it is not an unusual situation. It might be a silent killer.

Management Morass

NTSB findings (extracts):

• At the time of the accident, Aviation Charter was not operating in accordance with its weight and balance load manifest procedures, it did not have adequate stall recovery guidance, it did not have consistent deicer boot operational guidance, and it did not have an in-range [approach and landing] checklist.
• Aviation Charter was not adequately making company pilots aware of its standard operating procedures.
• At the time of this accident, Aviation Charter was not training pilots on crew resource management (CRM) in accordance with its [FAA-approved] CRM training module. Source: NTSB

Message from the Director General

“The number of accidents occurring in air taxi operations continues to be of great concern … What is required is to improve our attitudes toward safe operations and to adopt a better safety culture within the air taxi industry.”

The Safety of Air Taxi Operations (SATOPS) task force report (extracts):

• Recommend pilots stop pushing the weather.
• Recommend Transport Canada [counterpart to FAA] develop a standard for human factors and decision making training.
• Encourage air taxi operator management to attend the company aviation safety officer (CASO) course.
• Recommend air taxi operator management attend the CASO course and implement the principles learned … in their company.
• Recommend air taxi operators establish a company safety program that is fully supported by management.
• Recommend Transport Canada require air taxi operators to submit relevant statistics to determine where accidents are occurring and to target areas where resources could be allocated for accident prevention programs.
• Recommend Transport Canada conduct in-flight inspections of air taxi aircraft.
• Recommend Transport Canada conduct more random audits and inspections.
• Recommend Transport Canada provide more regulatory compliance presence, especially in northern and remote areas.

Source: Transport Canada, SATOPS final report, spring, 1998

Accident Rates Compared
	Part 121 scheduled service (airlines)	Part 135 non-scheduled service (on-demand air taxis)
5-year average 1998-2002 Accidents/100,000 flying hours	0.248	2.132
Flying hours/ accident	1 per 403,225	1 per 46,904
Sources: http://www.ntsb.gov/aviation/Table6.htm, http://www.ntsb.gov/aviation/Table9.htm