The Role of Non-Volatile Memory in Aircraft Investigations: Uncovering Hidden Data, Study notes of Aviation

Download The Role of Non-Volatile Memory in Aircraft Investigations: Uncovering Hidden Data and more Study notes Aviation in PDF only on Docsity!

Remembering Before the Crash:

How Non-Volatile Memory Can Change The Course

Of an Investigation

Elise Marie Vondra

University of Southern California

ISASI Rudolf Kapustin Memorial Scholarship Submission 2019

In the last 20 years, investigators have made air travel increasingly safe though learning from past accidents. 541 fatalities in 34 commercial aircraft accidents occurred between 1990 and 1999, and those numbers dropped to 67 fatalities in 8 accidents in the years 2007- 2017 (1). Investigators painstakingly find, sort and analyze data from the aircraft in order to determine the cause. In the aftermath of an accident, however, media coverage focuses mainly on two major types of data, the Flight Data Recorder (FDR) and Cockpit Voice Recorder (CVR), publishing articles such as “Crashed Lion Air Plane’s Cockpit Voice Recorder Found” (2). The FDR and CVR seemingly give investigators enough data to work with, yet an increasingly important and more nuanced source of data comes from Non-Volatile Memory (NVM) sources on the aircraft. NVM data is stored within the chip of a component, is utilized in avionics such as the GPS, altimeters, and pressure sensors, and survives without a power source. NVM is able to record data from more flights than current FDRs and CVRs. Currently the widely-used Honeywell Connected Recorder-25 (HCR-25) FDR and CVR only store the last 25 hours of flight data (3). In contrast, an NVM chip on a Cabin Pressure Controller, such as that used on Helios Airways Flight 522, can store over 300 hours of data (4). As avionics advance, the use of NVM on components and their available storage will increase. The investigator must adapt to find and utilize this data, not only after an accident occurs, but also in preventative measures. Three accidents, Helios Airways Flight 522 in 2005, Turkish Airlines Flight 1951 in 2009, and a General Aviation flight in Papua New Guinea in 2009, highlight the importance of NVM, and how it can be taken advantage of in accident investigations. En route from Larnaca, Cyprus to Prague, Czech Republic, a Helios Airways Boeing 737- 300 lost contact with Air Traffic Control (ATC) mid-flight. Loss of cabin pressure caused both pilots to lose consciousness. The aircraft flew until fuel starvation, eventually impacting terrain 33 km northwest of Athens. The Air Accident Investigation and Safety Board (AAIASB) of the Hellenic Ministry of Transport and Communications concluded that the cabin pressurization mode was set to manual control, and both pilots failed to identify any subsequent warnings. The AAIASB complimented FDR and CVR data with data from the two Cabin Pressure Controllers’ NVM. The No. 2 slave controller survived the accident, and investigators sent it to Nord-Micro,

device that can be installed in smaller aircraft that do not operate under air carrier regulations. Investigators recovered the device at the accident site and sent to the manufacturer for analysis. This device stores data regarding takeoffs, landing, groundspeed, and GPS position and altitude. The ATSB, the governing body in Papua New Guinea, used this data to recreate the flight path, with data taken about every 7 minutes to track the flight up until the crash, and not rely on other, less factual sources of data, such as ATC recordings or eye witnesses (6). In each of these cases, NVM provided additional data and context to accident investigators. While more limited FDR recording data may describe what occurred during the accident, NVM can provide information from many previous flights, revealing trends showing why the accident ultimately occurred. In both Turkish Airlines Flight 1951 and Helios Airways Flight 522, NVM data provided trends that investigators utilized and applied to existing and future aircraft. As more sensors become digitized, and potentially contain NVM, investigators can, and must, find, read, analyze, and act upon the data contained in them. The investigator can also look to other sources of NVM that are not installed in the airframe. For example, passengers’ cell phone data may be used to track GPS and accelerometer data. This has been used by the NTSB in railroad accidents, specifically, where the NSTB analyzed data from 80 different passenger devices in the investigation of the Amtrak 188 Derailment in Philadelphia in 2015 (7). This technique can be implemented into aviation applications so that investigators can collect more data leading to the incident. Personal devices may also give investigators an eye witness view of the flight. For example, if a passenger filmed the cabin in extreme turbulence, this could give investigators an idea of flight conditions before the accident occurred. New measures are being taken to improve flight data by component manufactures. General Electric is placing their new Enhanced Airborne Flight Recorder (EAFR) into the Boeing 787 and 777x fleet. The EAFR will allow 50 hours of recorded memory, compared to the 25 hours of the HCR- 25 (8). Although this is an improvement, the EAFR does not replace the detailed and larger storage of NVM on other smaller components.

A component that would greatly benefit from NVM analysis is Angle of Attack (AoA) sensors on aircraft. With the current controversy surrounding the Boeing 73 7 - MAX 8 accidents involving Lion Air in 2018 and Ethiopian Airlines in 2019, investigators would gain information from looking at NVM of AoA sensors of all previous MAX 8 flights, and other airframes that use an AoA indicator. However, currently, the main source of data on AoA sensor failures is pilot driven NASA reports in the Aviation Safety Reporting System (9). Reliable, robust, and factual data from NVM would assist investigators, manufacturers and carriers in improving safety and reliability of new sensors. As well as providing the aircraft investigators with data that would otherwise not be available to due to regulations, as in the DHC- 6 - 300 case, NVM also allows larger scale trends to be seen in investigations. In the Turkish Airlines Flight 1951 accident, the radio altimeter NVM provided more context to the crash, showing a pattern of incorrect altimeter readings that went uncorrected. Adding NVM to new components can also assist in using past flights for current accidents. Learning from these accidents, a challenge to investigators, and especially operators, is to collect and analyze this data more frequently. Taking proactive measures in data analysis of components will not only assist carriers in preventative measures, but also can be applied at a multi carrier scale, thus assisting component manufacturers in innovating safer technology, thus assisting the trend of creating air travel safer for all.