Additional Participating Entities:
Customs and Border Protection; Washington, District of Columbia
Defense Support Services/PAE; Arlington, Virginia
Bureau d’Enquêtes et d’Analyses (BEA); Le Bourget, FN
Airbus Helicopters; Grand Prairie, Texas
Turbomeca; Grand Prairie, Texas
Aviation Accident Factual Report - National Transportation Safety Board: https://app.ntsb.gov/pdf
Docket And Docket Items - National Transportation Safety Board: https://dms.ntsb.gov/pubdms
United States Department of Homeland Security: http://registry.faa.gov/N578AE
NTSB Identification: ERA14TA096
14 CFR Public Aircraft
Accident occurred Friday, January 17, 2014 in Houlton, ME
Aircraft: EUROCOPTER AS350B3, registration: N578AE
Injuries: 2 Uninjured.
NTSB investigators may not have traveled in support of this investigation and used data provided by various sources to prepare this public aircraft accident report.
HISTORY OF FLIGHT
On January 17, 2014, about 2040 eastern standard time, a Eurocopter AS350B3, N578AE, operated by U.S. Customs and Border Protection (CBP), was substantially damaged following a reported engine compartment explosion at Houlton International Airport (HUL), Houlton, Maine. The two commercial pilots were not injured, and night visual meteorological conditions prevailed. The helicopter was operating on a company flight plan for the local public use flight.
According to the copilot's written statement, the helicopter had just returned to HUL after a search and rescue mission, with the copilot flying. The crew then commenced takeoff and landing practice for night flight and night vision goggle (NVG) recurrency training.
The copilot stated that he had completed a single approach and three landings to runway 23, with one landing at the beginning, one near the middle, and one toward the end. He then took off again, hovered, and began a transition to forward flight with forward cyclic and a slight amount of increased collective. As he was beginning to apply cyclic, he heard and felt a loud explosion from the rear of the helicopter, followed by a severe vertical vibration, and the engine noise became "very loud." The helicopter began to experience cyclic controllability problems along with some yaw instability as well. The pilot in command (PIC) called out the rotor rpm warning horn, but the copilot was unsure whether it was constant or intermittent due to the engine noise. Helicopter control continued to rapidly decay for the next 5 to 10 seconds, at which time the PIC took control.
The copilot also noted that he subsequently made two mayday calls over the radio [per the PIC, via the footswitch], and almost immediately, the helicopter began "severe" pitch and roll oscillations. During some of the oscillations, the left side door came open, but the copilot was able to get it closed again. About that time, he also noticed that the red "GOV" light [manual mode engaged and loss of automatic governing failure] was illuminated. After 10 to 20 seconds, the PIC was able to regain some control of the helicopter, there was a decrease in engine and rotor noise, and the PIC was able to land the helicopter beyond snow banks at the end of the runway. After performing an emergency shutdown, the PIC said he thought the helicopter was on fire, and although the FIRE light was not illuminated, there was an orange glow reflected in the snow. Upon exiting the helicopter, the copilot saw flames coming from the engine compartment; he tried to extinguish the fire with a portable fire extinguisher, but without effect. The local fire company arrived about 10 minutes later and subsequently extinguished the flames.
According to the PIC, after the explosion, the main rotor (NR) overspeed warning sounded and a vertical vibration developed. At that point, the helicopter had not yet begun yaw oscillations, so the PIC felt they still had tail rotor thrust. He could not quite hear if the NR warning was intermittent or continuous (low NR) and told the copilot they could have low rotor rpm. He believed that the copilot then lowered the collective slightly in response to his statement, but the noise increased and the oscillation began. The PIC then took control of the helicopter. As he did, he observed two amber caution lights and what he believed were two red warning lights. Severe vertical vibrations and almost uncontrollable yaw oscillations continued, as did a high NR warning.
The PIC then focused on trying to keep the helicopter's skids level, not hitting the ground, and not flying out of ground effect. He could not ascertain airspeed, and there were three instances when he estimated that the helicopter entered 30- to 40-degree banks.
Throughout the event, the PIC could not adjust collective without inducing "extreme" attitude excursions. He also could not maintain the helicopter in a position where he could roll off the throttle. Then, after about 30 seconds, the attitude excursions began to "calm down," and the pilot was able to land the helicopter beyond the snow bank. As the helicopter touched down, the PIC noted that the red FIRE light was not illuminated, and that the original two red lights he saw were actually an amber ENG CHIP light and the red GOV light. After the event, and reviewing training materials, the PIC was able to estimate that the amber lights he saw were the FUEL P and DOOR lights.
The PIC further stated that the use of NVGs did not hamper his vision in the cockpit.
The 2003 Eurocopter AS350B3 helicopter was powered by a single Turbomeca, Arriel 2B engine driving a three-blade main rotor system which rotated clockwise (viewed from above), and a conventional tail rotor. It had skid-type landing gear and dual flight controls, and had been updated with night vision goggle (NVG)-compatible cockpit lighting.
Engine speed was governed by a "Digital Engine Control Unit" (DECU), a single-channel fuel governor design that performed fuel regulation, engine parameters management and failure recording; a Hydro-Mechanical Unit (HMU) with manual backup system that included a pump/metering unit; and electrical and mechanical links between the helicopter, the DECU and HMU.
The helicopter's collective twist grip had been modified. The mechanical stop of the original twist grip was replaced by one that automatically freed the grip and allowed the pilot to manually modify fuel flow rate when the red GOV light was illuminated. If the red GOV light illuminates, an aural "GONG" is triggered.
Per Eurocopter Letter-Service No. 1702-71-05, "In the event of a total…governor failure detected by the DECU, the engine fuel flow is frozen at its value at the moment of the failure. The engine power is therefore maintained. In stabilized flight, there is no urgency to modify the flight parameters and to adjust the twist grip. On the contrary, rapidly reducing collective pitch, without a synchronized reduction of the twist grip, will create rotor overspeed."
There were no flight or cockpit voice recorders on the helicopter. There were, however, some recorded maintenance parameters within a "Vehicle and Engine Multifunction Display" (VEMD) and the DECU.
The VEMD had a multifunction screen installed on the instrument panel designed to
manage essential and non-essential vehicle and engine data. The VEMD was a dual-channel system that could store flight reports, failure messages with associated parameters, and undated over-limit reports. Recordings would begin when Ng (engine gas generator speed) increased to 10% or when NR (main rotor rotational speed) increased to 70 rpm, and ended, or were "closed out" when Ng decreased to 10% or NR decreased to 70 rpm.
The DECU stored numbered failure blocks.
Runway 23 was 5,015 feet long and 100 feet wide, at an elevation that averaged about 485 feet.
HUL weather, recorded at 2053, included clear skies, wind from 200 degrees true at 5 knots, visibility 10 statute miles, temperature 1° C, dew point 0° C, and an altimeter setting of 29.83 inches Hg.
WRECKAGE AND IMPACT INFORMATION
According to the responding Airbus investigator, although the helicopter had been moved to the hangar before his arrival, observations provided by CBP indicated that the helicopter had initially come to rest about 250 feet beyond the end of runway 23. The pieces of engine cowling that were not consumed by the fire had been removed and were in the hanger with the helicopter.
The majority of the helicopter structure was not burned or damaged from the event. The forces on the airframe at the time of landing were reported to be relatively normal. There were no reported rotor blade strikes. All of the dynamic and static components of the helicopter were accounted for.
Main rotor and tail rotor drive train continuity were confirmed. The emergency fuel handle and the rotor brake handle were observed to be pulled to the rear. The emergency power switch was up and engaged and the twist grip was in the MIN position.
All of the right side belly panel latches were undone except for one latch.
The fire damage was mostly confined to the "bathtub" area of the engine; however, there was thermal structural damage below the bathtub which was considered substantial damage to the helicopter per 14 Code of Federal Regulations Part 830.
There was no apparent trace of the engine's sand filter. The engine's free turbine blades were separated from the free turbine disk, and the containment shield was distorted and pierced, consistent with the effects of the blade-shedding phenomenon. There was FOD damage to the inlet compressor blades. The intake funnel was partially consumed, the adjusted valve assembly, bleed valve, and exciter box were melted, and the igniter case was melted with holes revealing the internal components.
The engine and some additional components were removed from the helicopter and shipped to Turbomeca USA, where additional examinations occurred with Federal Aviation Administration (FAA) participation and NTSB oversight. There, the engine was photographed prior to disassembly, and all external piping, accessories and wire harnesses were removed. A large amount of debris was noted inside the engine; however, it could not be determined whether the debris was ingested or forced into the engine by firefighting efforts.
The HMU was removed and packaged for shipping to France for further examination. The engine was then separated into its various modules and components.
The reduction gearbox (module 5) was removed and examined. The input pinion alignment marks were found to be aligned. Continuity was confirmed through the reduction gear train. No further disassembly of the module was performed.
The free turbine (module 4) was removed. The free turbine blades were found shed from the turbine disk at the over-speed notch in the blade roots. (The design of the blade is such that at an overspeed of 140–150% N2, the turbine blades will liberate at a machined notch to prevent a turbine disk rupture about 170%.) The disk could still be rotated on the bearing with minimal effort. Module 4 could not be disassembled normally due to distortion from the blade shedding.
The gas generator (module 3) was fully disassembled. Foreign object debris (FOD) was observed on several of the centrifugal compressor blades. The compressor cover did not show any signs of contact with the centrifugal compressor. A substantial amount of burnt material was found inside the housing but it was impossible to determine what had been ingested by the engine or had been forced inside by the firefighting efforts.
The axial compressor module (module 2) was removed from module 3. The axial compressor was removed from the housing. FOD was present on each of the 13 compressor blades concentrated mostly on the leading edges near the tips. A substantial amount of burnt material was found inside the housing.
Continuity was confirmed through the accessory gear train and no disassembly of the module 1 was performed.
The "adjusted valve" and the bleed valve were observed to be melted. Similarly, the high energy box exhibited significant thermal damage. The wiring on the right side of the engine was burnt but appeared mostly intact on the left side. The accessories on the left side appeared intact as well, but was covered with a black sooty deposit. No discrepancies (other than deposits due to fire) were found when removing the pipes.
Drops of molten metal, consistent in appearance with aluminum, could be seen in several places on the lower part of the engine.
TESTS AND RESEARCH
Onsite, the responding investigators powered up the VEMD, and noted the following codes for Flt # 2791in the unit's "Maintenance Mode." They also noted "Test Reference" numbers, and the screens were photographed.
The VEMD had not been "closed out" for the flight and as such, the Flight Report and over limits were not recovered at that time. The VEMD was ultimately shipped to the Bureau d'Enquêtes et d'Analyses pour la Sécurité de l'Aviation Civile (BEA), where the two internal boards were visually inspected and found to be in "good condition." It was later powered up on an Airbus Helicopters bench, and the screens were photographed. Results, along with Airbus Helicopters explanations included:
Data indicated that the last recorded flight, numbered 2791, lasted almost 1 hour and 10 minutes. Sixteen failures and several over-limits were recorded during the flight. The 16 failures were categorized into 3 groups: The two first failures (no identification label) occurred at the beginning of the flight and were not correlated to the event. The next eight failures were recorded during the flight, and the last six failures correlated to the VEMD having been powered up on the helicopter after the accident flight.
Eight undated over-limits were also recorded:
- A T4 MED (temperature at the exhaust end of the gas generator chamber) over-limit that lasted 19 seconds. The T4 MED temperature threshold would have reached values above 865°C during the starting phase or values between 915°C and 941°C during the flight phase.
- A T4 HI over-limit that lasted 8 seconds and reached values above 941 °C.
- An NG over-limit that lasted 40 seconds and reached a value of 104 %.
- An NF (free turbine rotation speed, computed in rotor rotation speed) over-limit that lasted 25 seconds and reached a value of 510 rpm (510 rpm is the maximum value recorded by the VEMD).
- Four NR (rotor rotation speed) over limits which maximum value reached 496 rpm.
Pertinent Test Reference points included:
Test Ref 9 Time: 0:00:00
Test Ref 129 Time: 1:08:26
Test Ref 131 Time: 1:08:56
Test Ref 133 Time: 1:08:56
Test Ref 44 Time: 1:08:56
Test Ref 53 Time: 1:09:26
Test Ref 126 Time: 1:09:27
Test Ref 55 Time: 1:09:49
Test Ref 129 was a "stepper motor or resolver failure," and a red GOV warning light would have illuminated. It would have been a permanent failure that indicated a frozen stepper motor position for the gas generator. The pilot would have had to manage NR and fuel flow manually through the twist grip, which is automatically unlocked when the red "GOV" occurs.
Test Ref 131 was an NF "B" sensor failure ("B" and "A" sensors were used to govern the free turbine), which would have occurred when NF exceeded 141%. On another VEMD page, NF at the same moment indicated a maximum recorded value of 511 rpm, or an NR of 132% (the maximum value recorded by the VEMD overlimit page). In addition, NG was in exceedance, and T4 and TRQ were also high.
Test Ref 133 was an NF "A" sensor failure, which typically occurs when NF exceeds 141%. As with NF "B", on another VEMD page, NF at the same moment indicated a maximum recorded value of 511rpm, or an NR of 132% (the maximum value recorded by the VEMD overlimt page). In addition, NG was in exceedance, and T4 and TRQ were also high.
Test Ref 44 indicated invalid NF "A" information. The failure was intermittent.
Test Ref 53 reflected an invalidity of TRQ information received. On another VEMD page, NF at the same moment indicated a maximum recorded value of 511 rpm, or an NR of 132%. In addition, NG and T4 were in exceedance, and TRQ was high and NR was 0.
Test Ref 126 indicated a raw torque value failure.
Test Ref 55 indicated an out of range oil pressure.
The DECU was placed on the Turbomeca USA test bench and powered on. The DECU was found to be operational and the faults were then downloaded. Out of the 32-fault capacity, 4 blocks were associated with the accident flight which matched the initial information obtained from the VEMD. A stepper motor failure was recorded, which was linked to the stepper motor or to the stepper motor control and would have triggered the red GOV indication.
The DECU also underwent board inspection with no preexisting anomalies noted.
The DECU's stepper motor function was checked at different temperatures on a specific test rig and found to be within specifications.
- At room temperature (+20°C/+68°F).
- At -40°C (-104°F).
- At +50°C (+122°F).
An additional test was performed on a forced vibration table at room temperature (+20°C/+68°F) for 15 minutes with no anomalies noted.
Acceptance test procedures (ATP) found the unit compliant with all the manufacturer specifications. No fault was found during all the tests, and the stepper motor failure could not be duplicated.
The HMU could not be tested on a test bench due to the external thermal damage; however, electrical and functional checks of the stepper motor, the resolver and the neutral switch were performed which were found to be within specifications.
Testing also indicated that a rod normally connected to the collective pitch lever twist grip was out of the neutral position, consistent with the fuel flow being manually controlled while the HMU was on the helicopter.
A complete disassembly found the components in "very good" condition and did not reveal any anomalies that could have mechanically resulted in the red GOV indication.
Also noted, was soot and coagulated molten aluminum on the casing, consistent with the HMU having been in the vicinity of high heat.
The harness between the DECU, the stepper motor and the resolver was visually inspected with "severe" fire damage observed. The electrical connection between the DECU, the stepper motor and the resolver of the harness was tested using a repair shop test rig, with no fault was found. The cables were also bent and twisted manually without any fault noted.
The axial and centrifugal compressor wheels, gas generator turbine wheel, engine air intake duct, and sand filter pieces were forwarded to the NTSB Materials Laboratory. According to the resultant factual report, upon receipt, the axial compressor, centrifugal compressor, and gas generator turbine wheels were intact, while a large portion of the air intake duct and most of the sand filter were missing.
Compositional analysis was performed on impact deposits found on the axial compressor blades and impact deposits found on the gas generator turbine blades.
The axial compressor was made of a titanium alloy, but the areas of impact showed material transfer with composition of a nickel-base alloy with chromium, cobalt, iron, aluminum and probably molybdenum. The source(s) of the impact areas could not be determined.
Debris, consistent with charred engine cowling, was found throughout the engine components including glass fibers found on the turbine wheel. Energy dispersive x-ray spectroscopy (EDS) of the blades from the axial compressor and the gas generator turbine revealed spectra across many areas that included peaks of aluminum, silicon, calcium, antimony, oxygen, and carbon.
While the peaks of aluminum, silicon, calcium, oxygen, and carbon could be from many sources including sand, soil, the sand filter, or the engine cowling, antimony is an element that is only significantly present in the engine cowling and is commonly added to composite materials as a fire retardant.
The bulk of the deposits on the turbine blade generally showed EDS spectra with relatively high peaks of aluminum and oxygen. In comparison, the charred resin from the engine cowling showed relatively high peaks of carbon, oxygen, and phosphorus while the highest peaks for the glass fiber were silicon and oxygen. The relatively high peaks of aluminum and oxygen for deposits on the turbine blade leading edge were consistent with oxidized remnants of the sand filter and/or inlet duct.