Development of a Runway Veer-Off Location Distribution Risk Assessment and Reporting Template (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Veer-Off Reporting and Data Collection As noted earlier, the basis for developing veer-off location models was information obtained from aircraft accident and incident reports available from several databases in the U.S. and countries with aviation accident rates comparable to those verified in the U.S. The most important information collected was the charac- terization of the aircraft path during the runway veer-off. The pathway information was essential to develop the probabil- ity distributions associated with the longitudinal and lateral deviations when the plane was off the runway. This chapter presents the sources and type of information collected for the development of risk models. Data Required for Modeling Veer-Off Distances Previous mathematical models developed in the U.S. and elsewhere to characterize the probability of an aircraft devi- ating a certain distance from the runway during the veer-off have used either the final location of the aircraft or the largest deviation during the runway excursion. The basic approach in this effort was to describe the air- craft veer-off path and use this information to characterize the subareas contiguous to the runway and that were chal- lenged by the veer-off rather than using a single location. The proposed approach is a significant improvement to veer-off modeling. It required an in-depth evaluation of report nar- ratives to obtain the aircraft travel path for as many events as possible. More importantly, this approach also helped to characterize the probability distribution of the subareas over the length of the runway. The veer-off path of the aircraft and its stopping location depend on several factors that can be divided into the follow- ing categories: • Location where the aircraft departed the runway; • Speed of aircraft when leaving the runway; • Runway surface conditions (e.g., dry, wet, contaminated, etc.); • RSA surface conditions; • Presence of obstacles (e.g., NAVAIDs, ditches, uneven ter- rain, snow banks); • Condition of landing gear (e.g., retracted, partially collapsed); • Aircraft direction during the veer-off (e.g., straight, side- ways, ground looped); • Bearing capacity of RSA terrain during the incident; and • Pilot role in contributing to the event or in attempting to avoid it. A combination of factors is usually present in all events making it very difficult to accurately model these events. In particular, human factors are extremely complex to model. For this reason, this study did not focus on the causal fac- tors of the particular veer-off. Rather, the focus was placed on identifying evidence to characterize the chances that an air- craft veering off the runway will travel over certain subareas of the RSA. The data obtained on aircraft veer-off paths from actual veer-off accidents and incidents were used to charac- terize the probability distribution of the aircraft veer-off path occurring in segments of the RSA. The aircraft veer-off path usually cannot be completely characterized from the data provided in the accident/incident report. The investigation reports do not always provide specific location references. Some reports may provide the excursion pathway in a diagram or a picture, while others do not. It was necessary to make assumptions and make inferences based on information contained in the narrative of the report when pos- sible. For example, average aircraft deceleration and narrative of the aircraft speed when going off the runway helped identify the subarea in which the plane may have departed the runway. As expected, the information required to depict the actual wreckage path was rarely available. For this reason, report nar- ratives were reviewed and interpreted. All available references, C H A P T E R 3

6including historical satellite pictures, were used to infer the approximate veer-off path of the aircraft. Location References Used Longitudinal Distances Longitudinal distances were measured from the runway threshold for landing and from the beginning of the takeoff roll for takeoff operations. These two points normally coin- cide except for cases in which the threshold may be displaced. In a few cases the aircraft started its takeoff roll from an exist- ing taxiway intersection other than at the beginning of the runway and in these cases the distance was measured from the taxiway intersection with the runway. Lateral Distances The reference for measuring the lateral distances is the side edge of the runway. Justification to use the edge instead of the runway axis is presented below. Why Aren’t Lateral Deviations Measured from the Centerline? Another reference alternative evaluated for mea- suring lateral distances was the runway centerline; however, there were no data on aircraft wander that could be used to characterize the probability distribution both on the runway area and in the RSA. Moreover there is another important justifica- tion to use the runway edge, rather than the run- way centerline. It is important to note that the runway and the RSA may have very different types of surface and that the transition between the two areas may have a discontinuity in the pavement. Because aircraft control and braking can be significantly differ- ent whether the aircraft is moving on the runway paved surface or outside on the unpaved RSA, it is fair to assume that the probability distribution char- acterizing aircraft wander on the runway should not be extrapolated to outside the paved area. Therefore, as illustrated in Figure 2, there may be two very different probability distributions for the characterization of lateral distances, one covering the runway paved area, and another covering the RSA. In Figure 2, one of the distributions may be used to characterize aircraft wander during normal operations. The flat curve represents the probability distribution if the aircraft departs the runway paved area. Even if there is a paved shoulder area, it may have a small drop and most importantly, it is the area where runway lights are installed and in many cases struck by aircraft veering off the runway. In summary, a probability distribution to represent aircraft wander with distances measured from the runway centerline would incorporate an error due to the aircraft response on two different types of surface, on and off the runway. In addition, the purpose of this study is to model veer-offs and it is not necessary to model aircraft deviations in the paved area of the runway. Therefore, the edge of the runway was selected to measure lateral devia- tions during the lateral runway excursions. Runway Probability Distribution During Normal Operations LeftRight L=0L=0 RSA Probability Distribution for Veer-offs Interface (Discontinuity) Runway/RSA Figure 2. Use of runway edge to measure lateral distances (L = deviation from runway edge).

7 In addition, the runway distance available was divided into 20 subsections, 10 on each runway side, as shown in Figure 3. The length of each subsection varied according to the normal- ization procedure used for measuring longitudinal distances, as explained later in Chapter 5. The procedure was necessary to characterize the lateral probability distributions for each subsection evaluated and obtain the probability distribution for the entire runway. Aviation Accident and Incident Databases Veer-off data was collected from several databases in the U.S. and abroad; however, close to 90% of the information was retrieved from U.S. sources, particularly from the data- bases managed by the NTSB, FAA and NASA. International databases managed by accident investigation bureaus of ten different countries were also sources of information for major incidents and accidents. These countries have aviation accident rates similar to that of the U.S. The following is a list of the databases from which data were collected: • National Transportation Safety Board (NTSB)—Aviation Database, • Federal Aviation Administration (FAA) Accident/Incident Data System (AIDS), • FAA/NASA Aviation Safety Reporting System (ASRS), • Transportation Safety Board of Canada (TSBC)—Aviation Investigation Reports, • Australian Transport Safety Bureau (ATSB)—Aviation Safety Investigations and Reports, • France Bureau d’Enquêtes et d’Analyses pour la Sécurité de l’Aviation Civile (BEA)—Rapports d’Enquête, • UK Air Accidents Investigation Branch (AAIB) - Publica- tions and Search Reports, • New Zealand Transport Accident Investigation Commission (TAIC)—Aviation Occurrence Reports, • Air Accident Investigation Bureau of Singapore (AAIBS)— Reports Available, • Ireland Air Accident Investigation Unit (AAIU)— Investigation Reports, • Spain Comisión de Investigación de Accidentes e Incidentes de Aviación Civil (CIAIAC)—Investigación, • South African Civil Aviation Authority (SACAA)— Accidents and Incidents, and • Dutch Safety Board (DSB)—Investigation and Publication. Additional information about these databases is provided in Chapter 4. Database Statistics Veer-off records were identified from various sources and information was collected to develop the location probabil- ity models in this study. Records were consolidated, dupli- cate records were removed and, even during the modeling process, additional data to fill the gaps were collected, when possible. A summary of data used for modeling is provided below: • Period: 1982 to 2011; • A total of 1,144 veer-off records were identified in the data- bases: 345 veer-off accidents and 799 veer-off incidents; • 901 veer-offs occurred during landing and 243 veer-offs during takeoffs; • There were 1,072 records from U.S. databases (NTSB, AIDS, and ASRS) and 72 records from 10 international data- bases; and • There were 577 records with sufficient information to char- acterize or infer the veer-off path. Direction of Operations Subarea 1L Subarea 1R Subarea 2L Subarea 2R Subarea 3L Subarea 3R Subarea 4L Subarea 4R Subarea 5L Subarea 5R Subarea 6L Subarea 6R Subarea 7L Subarea 7R Subarea 8L Subarea 8R Subarea 9L Subarea 9R Subarea 10L Subarea 10R Distance Available = D 0.1D 0.2D 0.3D 0.4D 0.5D 0.6D 0.7D 0.8D 0.9D 1.0D Figure 3. Subsections of the RSA – example for normalization with runway distance available.

8Figure 6. Veer-off records from 1982 to 2011—runway side. Figures 4 through 7 provide a summary of veer-off cate- gories for the records identified by the research team. Please note that the total number of records for a certain category depends on whether the information was available for the record. For example, the total number of records with veer- off path information is 577, which represents approximately 50% of the total number of events identified and included in the accident/incident database. As seen in Figure 6, most veer-offs occurred on the left side of the runway. A hypothesis test indicated that the difference between left and right is statistically significant. Figure 7. Records with information on veer-off path. Figure 5. Veer-off records from 1982 to 2011—type of operation. Figure 4. Veer-off accident and incident records from 1982 to 2011.