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.
C-1 Accident Report Data To understand the safety concerns in the NAS that could be caused by energy facilities and production, historical acci dent data were reviewed and analyzed. In this task, analysis of historical accident data is based on compiling aviation acci dent reports and databases. Energy facilities considered in this research include solar power facilities, wind turbines and traditional power plants. According to the initial database review, there are very few accident/incident records directly related to energy facilities. Thus, we adjusted the approach and searched for accidents due to potential safety impacts that could be caused by energy facilities and production listed in ACRP Synthesis 28. C.1 Data Source There are three major aviation safety databases: Aviation Safety Reporting System (ASRS)1, Aviation Safety Network (ASN)2 and Aviation Accident Database (AAD)3. In addition, the FAA is constructing the aviation safety information ana lysis and sharing system (ASIAS). ASRS is a repository of voluntary reports and collects confidential safety information from frontline personnel in aviation. ASN is supported by the Flight Safety Foundation, and AAD is supported by the National Transportation Safety Board (NTSB). The character istics of the three databases are quite different (see Table CÂ1). Figure CÂ1 shows the number of records in the three safety databases from 2007 to 2012. In this Guidebook, the researchers chose ASRS as the safety database for further investigation. A pilot search experiment shows that there are very few accidents directly related to energy systems but there are many accidents caused by safety impacts (e.g., glare or turbulence) which could be caused by energy facilities. C.2 Text Mining of Safety Data The safety data are semiÂstructured data, which include structured parts and unstructured parts. The structured parts usually show place, time, and types of flight and so on. Unstructured data are free text. Structured data can be expressed as a relational table with fields and can be easily summarized and classified. Classification and query of the structured part are well developed in present aviation safety databases. The analysis process of unstructured data highly depends on researchersâ experiences and knowledge. Text mining tools are being developed to improve the efficiency of human analysis on unstructured data. Functions of these tools include Clustering, Document Retrieval, and Classi fication. In this research project, a new document retrieval method was developed to help retrieve accurate information from unstructured data in the large size air traffic safety data base. The first step of the new method was sample analysis, including establishment of keywords dictionary, and calcula tion of weights of keywords, Critical Point and Max Distance. The Critical Point expresses average keywords frequency of samples. Max Distance is the maximum distance from samples to the Critical Point. The second step was to search similar records with samples, which were based on the Distance Rule. Distance Rule is arithmetic to calculate the distance to Critical Point, considering not only the frequency of keywords but also the weights of keywords. C.3 Case Study The new method was applied to search for incidents caused by the effects of glare. Figure CÂ2 presents the number of glare incidences reported in the ASRS annually from 2007 to 2012. The total number of reports over the 6 year period is A p p e n d i x C 1Aviation Safety Reporting System Database, http://asrs.arc.nasa.gov/search/ database.html. Accessed on October 10, 2012. 2Aviation Safety Network, http://aviationÂsafety.net/index.php. Accessed on December 8, 2012. 3Aviation Accident Database, http://www.ntsb.gov/aviationquery/index.aspx. Accessed on October 10, 2012.
C-2 106. Note that the glare came not only from energy facilities (e.g., concentrated solar power [CSP]), but also from a variety of sources, such as the reflection of lights from the glass walls of buildings. Further investigations were conducted to understand the time and flight phases when the glare incidents occurred. Figure C-3 shows that most glare incidents happened either in the day or at night but very few of them occurred during dusk or dawn. In addition, Figure C-4 compares the incidents occurring in different flight phases in the day or at night. It shows that during the day, incidents are evenly distributed in different flight phases, but at night, the incidents are more likely to occur during approach and taxi phases. This observation provides further insights for aviation safety management. Figure C-1. Comparison of data size (2007â2012). 0 5 10 15 20 25 2007 2008 2009 2010 2011 2012 Figure C-2. Glare incidences by year. 41% 45% 8% 6% Daylight Night Dusk Dawn Figure C-3. Glare incidences by time of day. Table C-1. Comparison of air traffic safety databases. ASRS ASN AAD Size of data Large Small Medium Covered Regions Worldwide Worldwide US Search Func ons Excellent Limited Good Export Format CVS, Word, Excel None XML, Text Figure C-4. Statistics of glare incidents by flight phases in the day and at night.
