Spinning In

February 1, 2003

Since the dawn of flight we've lost aircraft to stalls and spins. Such accidents continue to be among general aviation's leading problem areas, and even when airplanes are designed to have gentle stall characteristics, pilots still find ways to lose lift catastrophically.

Pilots should think angle of attack, since airspeed is an indirect measure of an approaching stall. A review of any basic aviation text helps if you are a bit foggy on this. On every flight we operate close to the critical angle of attack to get airborne, climb out, and return for landing. How close one gets to stalling and how one gets close make all the difference between a happy ending and a disaster. That was amply demonstrated in a recent accident involving a Cirrus SR22, which we'll get to momentarily.

The distinction between stall and spin complicates matters somewhat. A spin is an aggravated stall where one wing produces more lift than the other, resulting in autorotation. The critical point is that if the aircraft isn't stalled first, it can't spin. In light, Normal category single-engine aircraft, a spin in an incipient or early phase is defined, for FAA certification purposes, as one turn or three seconds of rotation. Recovery must be accomplished in not more than one additional turn.

But don't try this at home — "unapproved for spins" means exactly that. The fact that an airplane has demonstrated compliance for certification doesn't mean that the average pilot will survive an inadvertent spin. Four consecutive steps have to occur: The pilot must recognize that the aircraft is spinning and in which direction, and he must know exactly what to do and execute the procedure correctly the first time because the time available is about three seconds in many airplanes. After that, the aircraft may become unrecoverable. As the accident rate shows, this ability doesn't matter because in most cases the inadvertently spun aircraft is below recovery altitude.

In the case of the Cirrus SR20 and SR22 and Lancair Columbia 300, however, the FAA allowed an alternative way to meet the spin criteria with an exemption from the one-turn spin recovery requirement. The alternative focused on enhanced stall characteristics. Why the exception? The answer is in the statistics and the aerodynamics.

The difference in altitude required to recover from stalls and spins is significant. Most airplanes recover from a "normal" stall in several hundred feet, assuming the pilot recognizes it and takes prompt corrective action. Variables such as weight, aerodynamic design, power setting, load on the wing (Gs), and center of gravity (CG) have an effect, which can be pronounced. In an incipient spin recovery the pilot's operating handbooks (POH) of many aircraft are not very clear about altitude loss. Based on anecdotal observation and the few POHs that do provide data, plan on 1,000 to 1,500 feet as the bare minimum altitude loss, assuming that the pilot was right there with a textbook recovery. Under the best conditions it probably takes at least three to five times as much altitude to recover from an incipient spin as from a stall.

The AOPA Air Safety Foundation reviewed 465 fatal stall-related accidents that occurred during the past 10 years — an average of almost one fatal accident per week. (You may also search general aviation accidents in ASF's online database.)

We looked at each final accident report and categorized the phase of flight (such as takeoff, approach, climb, or cruise) shown in Table 1, and then we used witness statements or the accident investigator's estimate of the altitude when the stall occurred. As you can see from Table 2, in almost 80 percent of the accidents the precipitating event, the stall, occurred below 1,000 feet agl. Spin recovery from those altitudes is highly unlikely. The percentage could be higher, but because in 13 percent of the accidents the altitude was unknown we gave them the benefit of the doubt.

The phase of flight shows maneuvering to be the clear loser, with takeoff coming in a distant second. In the maneuvering-flight category — which the NTSB defines as aerobatics, low pass, buzzing, pull up, turn to reverse direction, or engine failure after takeoff — 70 percent of accidents occurred below 1,000 feet agl and half of those were below 250 feet agl.

The FAA's Small Aircraft Directorate, which issued the Cirrus SR20 and SR22 type certificates, looked at more than 1,700 stall/spin accidents dating back to 1973 and concluded that 93 percent of those airplanes were at or below pattern altitude — too low for spin recovery. The current one-turn spin recovery requirement remains essentially unchanged since 1945, so inquiring minds asked what would happen if the departure from controlled flight, the stall, was made more difficult. Would fewer accidents occur? If the stall is prevented the spin can't happen. More important, how many lives could be saved if the aircraft's stall characteristics were friendlier?

NASA asked the same question in the late 1970s. Paul Stough and Dan DiCarlo reviewed several aerodynamic approaches to the problem in a recent paper, "Spin Resistance Development for Small Airplanes — A Retrospective." If the wing tips stall last, the pilot can maintain lateral control well into the stall. In a well-behaved airplane, the inboard section of the wing stalls first and may cause buffeting and pitching without rolling off. If the pilot is paying any attention at all, the shaking, the decay of control response, and the pitch movement should provide ample warning to reduce the angle of attack and start flying again before the aircraft departs controlled flight.

Stough and DiCarlo discovered that a "drooped leading edge on the outboard wing panel delayed tip stall to a very high angle of attack and resulted in a relatively small penalty in cruise." A discontinuous outboard leading-edge modification acted as a vortex generator and delayed the stall from progressing to the wing tips. NASA flew the modified wings to almost double the normal critical angle of attack. Four aircraft, a Beechcraft Musketeer, Piper Arrow, Cessna 172, and Grumman American Yankee, were tested in an original and modified configuration. In addition to becoming more controllable in the stall, the aircraft became highly spin resistant. The results are shown in Table 3.

Why settle for spin resistance and not spin-proofing? Spin-proofing is described as a "very lengthy and technically difficult process." If you can eliminate 95 percent of the spins, does it make economic or common sense to greatly increase the cost and complexity for a very small reduction in accident potential? The FAA and the industry agreed that it did not. The paper established criteria to determine when an aircraft could be deemed spin resistant and thus excused from the normal one-turn recovery requirement. According to Stough and DiCarlo, "Since a spin-resistant airplane should not be spinnable under both normal and moderately abnormal circumstances, use of words such as departure and recovery was thought to be inappropriate. In the stall- and post-stall flight regimes, a spin-resistant airplane should have gentle and well-behaved stall characteristics while allowing the pilot to retain full control of all three axes of flight throughout the angle-of-attack range achievable."

Recognizing that the skill of pilots getting into inadvertent stalls might not emulate that of a NASA test pilot, a significant allowance is made for average technique. To qualify under the new FAA spin-resistant criteria, excerpting from the regulation, "at the applicant's option, the airplane may be demonstrated to be spin resistant by the following: (i) During the stall maneuvers...the pitch control must be pulled back and held against the stop. Then, using ailerons and rudders in the proper direction, it must be possible to maintain wings-level flight within 15 degrees of bank and to roll the airplane from a 30-degree bank in one direction to a 30-degree bank in the other direction; (ii) reduce the airplane speed using pitch control at a rate of approximately 1 knot per second until the pitch control reaches the stop; then, with the pitch control pulled back and held against the stop, apply full rudder control in a manner to promote spin entry for a period of 7 seconds or through a 360-degree heading change, whichever occurs first. If the 360-degree heading change is reached first, it must have taken no fewer than 4 seconds. This maneuver must be performed first with the ailerons in the neutral position, and then with the ailerons deflected opposite the direction of turn in the most adverse manner. Power and airplane configuration must be set in accordance with Sec. 23.201(e) without change during the maneuver. At the end of 7 seconds or a 360-degree heading change, the airplane must respond immediately and normally to primary flight controls applied to regain coordinated, unstalled flight without reversal of control effect and without exceeding the temporary control forces specified...and (iii) compliance must be demonstrated with the airplane in uncoordinated flight, corresponding to one ball-width displacement on a slip-skid indicator, unless one ball-width displacement cannot be obtained with full rudder, in which case the demonstration must be with full rudder applied."

As is usually the case with airplanes, there are compromises. Items (i) and (iii) are not too difficult, but (ii) is a bear. To grossly oversimplify, the aerodynamics that increase resistance to stalls to make them docile also make the wing resistant to recovery in certain spin regimes. The spin matrix, such as power on, power off, loading, and CG, is broad and the aircraft must demonstrate compliance in each area. It can be done but at a significant operational price. For example, an extremely narrow CG range could make it easier to meet requirement (ii) but then we might all be flying two-place aircraft.

According to Stough and DiCarlo, "Both the Cirrus and Lancair were certified using spin-resistance certification standards; however, neither was certified as fully spin resistant." Cirrus, which had already made the decision to include a standard parachute system to solve other safety problems, proposed this as an equivalent level of safety. If the pilot somehow managed to get beyond the enhanced stall characteristics and into a spin, there was a way to escape.

Now let's look at a preliminary NTSB report of a fatal Cirrus stall/spin accident that occurred last spring. The two private pilots, co-owners of a new SR22, had taken delivery of the aircraft less than a week prior to the accident.

The pilot in the left seat held an instrument rating for single- and multiengine airplanes. His logbook showed 311 hours of total flight time. The other pilot was instrument-rated in single-engine airplanes and his Cirrus client profile datasheet stated he had 475 hours of flight time. A contract flight instructor, who provided factory training, had flown with both pilots and estimated they had about 20 to 30 hours each in the SR22.

The airplane was equipped with a Cirrus Airframe Parachute System (CAPS). According to the SR22 pilot's operating handbook, "CAPS [is] designed to bring the aircraft and its occupants to the ground in the event of a life-threatening emergency. The system is intended to save the lives of the occupants but will most likely destroy the aircraft and may, in adverse circumstances, cause serious injury or death to the occupants. The CAPS consists of a parachute, a solid-propellant rocket to deploy the parachute, a [manually activated] rocket activation handle, and a harness imbedded within the fuselage structure.... CAPS is initiated by pulling the activation T-handle installed in the cabin ceiling on the airplane centerline just above the pilot's right shoulder."

The weather was good VFR with light winds. Prior to the accident a CFI reported seeing the Cirrus doing touch and goes. Radar data indicated that the target departed the airport, climbed to 5,500 feet, and then headed toward the accident area, maintaining between 5,200 feet and 5,700 feet. En route, it made a left 90-degree turn, followed by a right 90-degree turn. It then made an approximately 360-degree right turn, followed by a 360-degree left turn.

The target then continued the left turn, making smaller turns, until it reached the airspace over the accident site. Target altitude readouts in the vicinity of the accident site showed that the Cirrus started at about 5,500 feet and then lost more than 2,400 feet in less than 30 seconds before radar contact was lost.

A witness one-half mile to the north who was accustomed to airplanes performing maneuvers in the area noticed that the pilot would "cut the engine," then descend, and pull up, recovering with full power. The airplane performed the maneuvers for about five minutes, and the witness saw the maneuver repeated "three or four times." The witness stated that he was fairly sure the airplane "probably did a turn" at the end of the pull-ups, but he wasn't sure which direction the airplane may have turned. After the airplane completed its last pull-up, the witness noticed that it entered another dive. The airplane "suddenly went into a spiral and went straight down. He seemed to keep a constant speed on descent and looked like he was in slow-motion spinning. He continued nose down to the tree line and straight down to the ground. I did not hear his engine at all once he went into the spiral. I did not think he had an engine problem and [he] was intentionally cutting the power of his plane and then giving it full power on the climbout."

A second witness noted the airplane had "plenty of altitude." The airplane "peeled off to the left," and the witness "remembered seeing the bottom of the aircraft." The airplane passed through about 180 degrees of turn, then leveled off, "and right after it came back to level flight it stalled." The airplane "went into a nose-dive spin and then a flat spin into the ground." It "tumbled in a downward spiral, which turned into a flat spin...spinning on its own axis, slightly nose down, like a turning top."

A third witness, who observed the airplane with the second witness, noted that the airplane "rolled over once and then twisted, which looked to be intentional. Suddenly, the plane began doing a nose spin, which turned into a flat spin. It appeared as though the pilot lost control of the plane." No mention was made of a parachute deployment.

Except for some broken branches above the wreckage, and a small tree cut off next to it, there was no wreckage path through the trees. There was a post-crash fire. The CAPS parachute was found outside the airframe, in its deployment bag, in front of the right wing. The composite CAPS cover was found about 20 feet in front of the airplane, with no damage to its interior (kick plate) face. The solid propellant rocket was found on the ground, aft of the right wing, with cables leading to the wreckage. The propellant was expended, but it was unclear whether it was spent during an attempted deployment or because of the post-crash fire.

According to the SR22 POH, the airplane is not approved for spins, and the only method of spin recovery is activating the CAPS. If the airplane departs controlled flight, the CAPS must be deployed immediately. Spin entry is unlikely with proper airmanship, including the caveat never to abuse "the flight controls with accelerated inputs close to the stall." An abrupt wing drop in this case may lead to a spin or spiral, and it may be difficult to determine which. The POH notes that the minimum demonstrated altitude loss for a CAPS deployment is 920 feet from a one-turn spin, and pilots are cautioned not to "waste time and altitude trying to recover from a spiral/spin before activating CAPS."

CAPS was subject to an airworthiness directive that is not thought to be applicable to this aircraft because the aircraft was built after the AD was issued. We may never know if the pilots attempted to activate the parachute — or simply neglected to. There has been one successful CAPS deployment, and the pilot landed uninjured. That CAPS activation decision followed a flight control system (aileron) failure (see " Pilot Briefing: To Pull or Not to Pull the Chute," December 2002 Pilot). CAPS activation systems are tested while the aircraft is in production.

Making decisions from preliminary reports is risky, but we could probably agree that the pilot was engaged in some fairly enthusiastic maneuvering flight. Did the pilot go beyond the limits of the flight envelope and then hope to recover without having to destroy his new aircraft? Was there an attempt to activate CAPS at least 1,000 feet agl before impact?

The long-term accident record makes it clear that a large number of pilots do not respect or recognize stalls well enough to recover at the low altitudes at which they typically occur. The FAA responded to the accident statistics by allowing a new design approach that may ultimately save lives, but it's too soon to tell.

Is it unreasonable to expect pilots to follow explicit warnings about corners of the flight envelope and recovery procedures if the unthinkable occurs? Are the new certification requirements adequate? Stay tuned. ASF is following this one closely as is the FAA, the manufacturers, and the pilot community. In the interim, mind your angle of attack and your altitude.


Bruce Landsberg is the executive director of the AOPA Air Safety Foundation.


Fatal stall accidents

Table 1
Phase of Flight Total Percent
Approach 86 18.5%
Climb 11 2.4%
Cruise 15 3.2%
Go-around 30 6.5%
Maneuvering 187 40.2%
Takeoff 130 28.0%
Unknown 4 0.9%
Descent 1 0.1%
Crop-dusting 1 0.2%
  465 100%
Table 2
Altitude Total Percent
<250 ft 182 39.1%
251-500 ft 95 20.4%
501-750 ft 4 0.9%
751-1,000 ft 89 19.1%
> 1,000 ft 33 7.1%
Unknown 62 13.1%
  465 100%

Spin/attempt ratio

Table 3 Number of Spins
----------------------
Attempts
Aircraft Model Basic Airplane Modified Airplane
AA-1X (Yankee) 185
---- =96%
193
  0
---- =0%
 31
C-23 (Musketeer) 127
---- =98%
129
  7
---- =5%
134
PA-28RX (Arrow) 173
---- =83%
209
 13
---- =5%
244
C-172 (Skyhawk)  97
---- =59%
164
  0
---- =0%
 36