DCA14FA058

the failure of the flight crew to adequately monitor the system for proper operation and manually activate the system during the flight in icing conditions. Contributing to the accident was the crew's limited training on the manual operation of the anti-ice system and the nonactivation of the automatic ice detection system for reasons that could not be determined.

After receiving intermittent localizer indications on the airplane’s first approach to the airport, the flight crew, conducted a go-around, and air traffic control cleared the flight for a second approach. The captain indicated that, while the airplane was level at about 2,000 ft on the base leg, the flight entered clouds. The first officer stated that she noted moisture on the windshield wiper and the captain indicated that the wind screen was wet. The cockpit voice recorder (CVR) recorded the captain and first officer briefly discussing ice; however, the airplane’s ice protection system, which was set to the automatic mode, did not operate automatically, and the crew did not activate the system manually. The crew did not see the ice light come on and there were no icing messages on the engine indicating and crew alerting system. As the first officer was applying control inputs to adjust for a crosswind, a rapid roll to the right occurred, which resulted in a wing strike and substantial damage to the airplane. About forty minutes after arrival at the gate, an examination of the airplane found an accretion of ice on the leading edge of both wings. The aircraft performance study, which correlated icing charts with the airplane’s flight profile, determined that the airplane spent over 20 minutes at altitudes where icing was probable during both approaches. The study concluded that the right roll was not commanded by the flight crew but likely due to ice buildup. Although the vertical load factor did not indicate that the airplane experienced a full aerodynamic stall, the ice buildup likely created enough flow separation on one wing for it to lose lift during the flare, without affecting the control of the aircraft in a measurable way during the approach. After the accident, the ice detection and anti-ice systems were tested at the aircraft level with no anomalies identified. The ice detectors were also functionally tested by the manufacturer at the component level with no anomalies identified that could have contributed to the event. A review of FDR data revealed that no failures were recorded for the ice detection system during the accident flight. Additionally, the system operated as expected during a manual preflight test and detected icing conditions during the previous flight. A review of the maintenance records did not reveal any systematic problems with the ice detection system. Therefore, it could not be determined why the ice detection system did not detect the presence of icing conditions even though the airplane accreted ice during the approach. This possibly could have been due to variations in static air temperature that prevented the ice that accumulated on the sensors from reaching the alert threshold or the occurrence of meteorological conditions out of the 14 Code of Federal Regulations Part 25 Appendix C during approach, or a combination of these two factors.” Although the ice detection system did not automatically activate the ice protection system, the CVR recorded a brief discussion during the final approach indicating that the crew was aware that the airplane was picking up “a little bit” of ice. According to the Trans States Airlines EMB145 Airplane Operations Manual (AOM) and Standard Operating Procedures (SOP), even though the airplane is equipped with an ice detector, the crew was responsible for monitoring icing conditions and for manual activation of the ice protection system when necessary. Therefore, the crew recognized that the airplane was operating in icing conditions and accumulating ice and should have manually activated the ice protection system. It is likely the crew's overreliance on automation for the activation and proper operation of the ice and rain protection system resulted in their failure to adequately monitor the system and respond appropriately when it did not activate automatically. Although the AOM and SOPs indicated that the crew is responsible for monitoring icing conditions and for manual activation of the ice protection system when necessary, there was no information in Trans States Airlines ground training modules that presented the crew as being responsible for monitoring and activating the ice and rain protection system when no warnings or cautions were received from the EICAS. Additionally, manual ice detection methods for flight crews to use when flying in potential icing conditions were not specifically referenced during ground training. The Trans States Airlines manager of flight standards said that manual selection of the anti-ice system was not emphasized in training like the automatic mode of operation was during flight operations. It is possible that because the manual operation of the airplane's ice protection system was not emphasized during training, the crew may not have recognized the need to perform this task. Trans States Airlines issued an operations bulletin after the accident that stated interim procedures for crewmembers to follow when operating in potential in-flight icing conditions. The bulletin called for active monitoring of the deicing/anti-icing equipment and, if it did not activate, to accomplish the QRH's Ice Detectors Fail procedures.

The airplane was equipped with an FDR and a CVR. FDR data for the accident flight indicate that the ice detection fail parameter remained in the normal state for the entire flight. The ice condition parameter remained in the no-ice detected state for the entire flight. The FDR records these parameters from the No. 1 ice detection system only. A review of FDR data from the flight before the accident flight revealed that, before departure, the ice detection fail parameter switched to fail, the ice condition parameter switched from no ice to ice, and the stabilizer anti-ice command and wing anti-ice command switched to on. These parameter changes were consistent with a manual test of the No. 1 ice detection system. During the same flight, shortly after takeoff, the ice detection fail parameter remained in a normal status; the ice condition parameter switched from no ice to ice; and the stabilizer, wing, engine 1, and engine 2 anti-ice commands switched from off to on. These parameter changes are expected for a properly functioning No. 1 ice detection system that has detected icing conditions. FAA flight inspections of the MEM runway 36L localizer and glideslope and runway 36R localizer and glideslope were completed on October 29, 2014, and May 6, 2014, respectively. No anomalies were noted. Performance Study The NTSB conducted a performance study using FDR, CVR, and radar data. The study indicated that, while on the approach heading, the airplane experienced a 24 knt crosswind at 1,900 ft, dropping to 10 kts on the final portion of the approach. There was no data to support a sudden wind gust to have caused the sudden roll. The study correlated icing charts from the National Center for Atmospheric Research from the time of the accident with the airplane’s altitude and showed that the airplane spent an additional 19 minutes in an altitude region with an increased probability of icing during the go-around. The airplane’s automatic ice protection system did not activate, and the airplane’s de-ice systems were not on during the approach and landing. The right roll was not commanded by the pilots as the wheel position did not correspond with the roll. The recorded rudder pedal position was not consistent with the roll being initiated by excessive rudder deflection. During a full aerodynamic stall, the vertical load factor would drop due to the loss of lift, but the vertical load factor stayed between 0.9 and 1.1 g until after the roll event. Postaccident examination of the airplane revealed ice had built up on the leading edge of the wings, however, the vertical load factor record did not indicate that the airplane experienced a full aerodynamic stall. Ice buildup can cause aerodynamic stall as the ice disrupts airflow across aircraft lift surfaces. Buildup of ice on the leading edge can cause air flow to separate and lift to be lost across the whole or a portion of the wing. Aerodynamic Simulation Embraer conducted a simulation of the accident flight using its aerodynamic model of the EMB-145. The goal of the simulation was to quantify the rolling moment needed to match the aileron input and bank angle during the flare portion of the flight. While the simulation did show some differences between the simulation aileron and elevator inputs and the accident flight control surfaces, the discrepancies were small enough that they could have been due to the unavailable exact crosswind and side slip angle data. The simulation did not show a noticeable loss in roll authority or change in flight characteristics during the accident flight. However, ice could cause the airplane to roll by creating enough flow separation on one wing for it to lose lift without the initial ice build-up affecting the control of the airplane in a measurable way. Trans States Airlines Training Ice Protection System Training The EMB-145 ice protection system was covered in a Trans States Airlines ground training module in initial and recurrent training. The training addressed the automated detection and activation of the anti-ice system. There were no training references that presented the crew as being responsible to monitor and potentially activate the system when no warnings or cautions were received from the EICAS. Manual ice detection methods for flight crews to use when flying in potential icing conditions were not referenced during training. The adverse weather ground school training module had several slides that presented potential icing conditions on the ground and the associated crew procedures. The first officer stated that her training on the anti-icing system was that the system was supposed to let the pilots know when it failed to operate correctly. She said that they were not trained to turn on the system manually but to follow the QRH. She stated that because the crew did not know the anti-ice system did not activate on the accident flight, they did not follow the QRH for an anti-ice system failure. She did not recall the total air temperature (TAT) gauge reading during the final approach. (The QRH checklist for Ice Detector Fail called for the use of visual cues and temperature criteria to determine whether icing conditions existed. Ice Identification Training Trans States Airlines provided its EMB145 AOM, Volume 2, as a reference when the wing inspection lights were discussed in training.  A Trans States Airlines APD stated it was easy to tell if the airplane had ice by looking at the pattern on the unheated portion versus the heated portion of the windshield. In addition, a Trans States Airlines check airman said that the windshield wiper and windshield were standard ways to detect icing. The Trans States Airlines chief pilot said he identified icing by looking at the unheated portion of the windshield and the windshield wiper. Trans States Airlines Procedures Stabilized Approach Criteria Trans States Airlines EMB145 Standard Operating Procedures (SOP), section 1, Maneuvers and Procedures Guide, page 40, referenced an airspeed of 127 kts for a landing weight of 41,000 lbs and 45° of flaps and page 39, stated the following: "Stabilized Approach" the approach must be stabilized by 1,000 feet above field elevation when conducting visual and straight in instrument approaches in both IMC and VMC [visual meteorological conditions] weather conditions. During the final approach phase, when operating below stabilized approach height, in both VMC and IMC, on instrument and visual approaches, the following operational parameters must be maintained to be consider the approach stabilized. Sustained deviation from these parameters means the approach has become unstabilized and an immediate missed approach should be initiated. Either pilot may initiate the missed approach utilizing the callout "Go Around. ? In-Range and Before Landing checklists complete. ? Airplane properly configured; Final flap setting on circling approaches may be delayed as per EMB SOP Sec 1.5.9. ? Airspeed in the range Vref -5 knots to Vref +10 knots. VOR/LOC/FMS course deviation does not exceed one dot deflection. Glideslope deviation does not exceed one dot deflection. Descent rate does not deviate +/- 300 feet per minute (fpm) from planned descent rate and is no greater than 1000 fpm, unless specifically briefed. The airplane is descending along the proper descent path or is able to maintain obstacle clearance. Trans States Airlines Unstable Approach Data – General The Trans States Airlines director of safety stated that there was no data from their Aviation Safety Action Program or FOQA pointing to problems with stabilized approaches. He said their department looked at stabilized approaches, and the trend has been lower this past year. Additionally, a Trans States Airlines Check Airman stated that pilots could go-around and not fear that they would get a call from the chief pilot’s office under the company’s no-fault go-around policy. Tr