ELFLYSVE: Probability distributions for air traffic at Swedish airports

About This dataset was produced as part of the ELFLYSVE research project, which investigates the potential of electric aviation in Sweden. One of the project’s work packages focuses on techno-economic modeling and optimization of airport energy systems, specifically their design and sizing to meet the requirements for electric aircraft (EA) charging. Since there is currently no commercial EA traffic in operation, the project developed a methodology to generate synthetic flight timetables. These timetables are based on probability distributions derived from historical air traffic data in Sweden. The historical flight data was provided by Luftfartsverket (LFV), Sweden’s air traffic management service provider, and includes all scheduled flights (anonymized aircraft identifiers) to or from Swedish airports for the years 2019–2023 (approximately 1.2 million flights). Methodology The methodology takes the perspective of an individual airport (i.e. incoming and outgoing aircraft), rather than trying to plan for a network of airports and routes for EA which is a highly complex and uncertain task. From the historical flight data, representative distributions were created for Number of arrivals Arrival time Turnaround time This open data repository contains five types of datasets, provided both for Sweden as a whole and for individual Swedish airports (ten datasets in total): (1) Weekday arrivals,(2) Monthly arrivals,(3) Arrival minute,(4) Turnaround time (full)(5) Turnaround time (clustered) The (full) notation in dataset (4) indicates normalization based on the entire turnaround dataset. The (clustered) version in dataset (5) uses normalization within clusters, where a cluster is defined as an hour interval. This clustering accounts for the dependency between turnaround time and arrival time (e.g., arrivals near midnight typically have longer turnaround times than those during morning peak hours). Dynamic programming was applied to the full turnaround dataset to identify clusters with similar turnaround times. These datasets serve as input for synthetic timetable generation, followed by EA route assignment and charging load simulations. Further details on the methodology and results will be available in a journal paper (submission in January, 2026). Data description (1) Weekday arrivals This dataset contains the normalised probability distributions of number of arrivals per day of the week: Sweden: norm_distribution_weekday_arrivals_SE.csv Individual airports: norm_distribution_weekday_arrivals_per_airport.csv Column Description Data type weekday Day of the week (0 = Monday, 6 = Sunday) Integer probability (or ICAO-code) Normalised arrival probability (for individual airports, this column contains ICAO codes) Float (2) Monthly arrivals This dataset contains the normalised probability distributions of number of arrivals per month of the year: Sweden: norm_distribution_monthly_arrivals_SE.csv Individual airports: norm_distribution_monthly_arrivals_per_airport.csv Column Description Data type month Month of the year (1 = January, 12 = December) Integer probability (or ICAO-code) Normalised arrival probability (for individual airports, this column contains ICAO codes) Float (3) Arrival minute This dataset contains the normalised probability distributions for arrival minute of the day: Sweden: norm_distribution_arrival_minute_SE.csv Individual airports: norm_distribution_arrival_minute_per_airport.csv Column Description Data type minute Minute of the day (0-1439) Integer probability (or ICAO-code) Normalised arrival probability (for individual airports, this column contains ICAO codes) Float (4) Turnaround time (full) This dataset contains the normalised probability distributions of turnaround time: Sweden: norm_distribution_turnaround_minutes_SE_full.csv Individual airports: norm_distribution_turnaround_minutes_per_airport_full.csv See explanation in Methodology section for the "(full)" notation. Column Description Data type minutes Turnaround time expressed in number of minutes (0-1439) Integer probability (or ICAO-code) Normalised probability of turnaround time (for the individual airport dataset, this column is expressed as the ICAO-code of each airport) Float (5) Turnaround time (clustered) This dataset contains the normalised probability distributions of turnaround time within clusters: Sweden: norm_distribution_turnaround_minutes_SE_clustered.csv Individual airports: norm_distribution_turnaround_minutes_per_airport_clustered.csv See explanation in Methodology section for the "(clustered)" notation. Column Description Data type airport (only in "...per_airport_clustered.csv") ICAO-code of the airport String hour_interval Hour interval of the cluster (e.g. 5-18) String turnaround_minutes Turnaround time expressed in number of minutes Integer probability Normalised probability of turnaround time within the specified cluster Float Additional notes This research was funded by the Swedish Transport Administration (TRV 2023/34443).概述 本数据集源于ELFLYSVE研究项目，该项目旨在探究瑞典电动航空的发展潜力。项目下设的其中一个工作包聚焦于机场能源系统的技术经济建模与优化，具体而言，即针对满足电动飞机（electric aircraft, EA）充电需求的机场能源系统设计与规模配置展开研究。 鉴于当前尚无商业化运营的电动飞机航班，该项目开发了一套用于生成仿真航班时刻表的方法体系。此类时刻表基于从瑞典历史空中交通数据中提取的概率分布构建。历史航班数据由瑞典空中交通管理服务提供商Luftfartsverket（LFV）提供，涵盖2019至2023年间往返瑞典各机场的所有定期航班（已对飞机标识符进行匿名化处理），总航班量约120万架次。 方法学 本方法以单个机场（即该机场的进出港航班）为视角，而非针对电动飞机航线网络进行规划——后者属于高度复杂且充满不确定性的任务。研究团队从历史航班数据中提取得到三类代表性概率分布：到港航班数量、到港时刻以及地面周转时间（turnaround time）。 本开源数据集仓库共包含五类数据集，分别面向瑞典全国及单个瑞典机场（总计10个数据集）： (1) 工作日到港量（Weekday arrivals） (2) 月度到港量（Monthly arrivals） (3) 到港分钟分布（Arrival minute） (4) 地面周转时间（全量版）（Turnaround time (full)） (5) 地面周转时间（聚类版）（Turnaround time (clustered)） 数据集(4)中的“全量版”标注表示基于全部地面周转时间数据集进行归一化处理。数据集(5)的“聚类版”则采用聚类内归一化方式，其中聚类以小时区间为单位定义。该聚类处理考虑了地面周转时间与到港时刻之间的相关性（例如，午夜前后的到港航班通常比早高峰时段的到港航班拥有更长的地面周转时间）。研究团队对全量地面周转时间数据集应用动态规划（dynamic programming），以识别出地面周转时间相近的聚类簇。 此类数据集可作为仿真航班时刻表生成的输入数据，后续还可用于电动飞机航线分配以及充电负荷模拟。关于本方法及研究结果的详细内容将于2026年1月投稿的期刊论文中公开。 数据说明 (1) 工作日到港量（Weekday arrivals） 本数据集包含按星期几统计的到港航班数量归一化概率分布： 瑞典全国：norm_distribution_weekday_arrivals_SE.csv 单个机场：norm_distribution_weekday_arrivals_per_airport.csv | 列名 | 描述 | 数据类型 | |------|------|----------| | weekday | 星期几（0 = 周一，6 = 周日） | 整数型 | | probability（或国际民航组织（ICAO）代码） | 归一化到港概率（单个机场数据集的该列包含机场ICAO代码） | 浮点型 | (2) 月度到港量（Monthly arrivals） 本数据集包含按月度统计的到港航班数量归一化概率分布： 瑞典全国：norm_distribution_monthly_arrivals_SE.csv 单个机场：norm_distribution_monthly_arrivals_per_airport.csv | 列名 | 描述 | 数据类型 | |------|------|----------| | month | 月份（1 = 1月，12 = 12月） | 整数型 | | probability（或国际民航组织（ICAO）代码） | 归一化到港概率（单个机场数据集的该列包含机场ICAO代码） | 浮点型 | (3) 到港分钟分布（Arrival minute） 本数据集包含按当日分钟统计的到港时刻归一化概率分布： 瑞典全国：norm_distribution_arrival_minute_SE.csv 单个机场：norm_distribution_arrival_minute_per_airport.csv | 列名 | 描述 | 数据类型 | |------|------|----------| | minute | 当日分钟数（0-1439） | 整数型 | | probability（或国际民航组织（ICAO）代码） | 归一化到港概率（单个机场数据集的该列包含机场ICAO代码） | 浮点型 | (4) 地面周转时间（全量版）（Turnaround time (full)） 本数据集包含地面周转时间归一化概率分布： 瑞典全国：norm_distribution_turnaround_minutes_SE_full.csv 单个机场：norm_distribution_turnaround_minutes_per_airport_full.csv 关于“全量版”标注的说明详见前文方法学章节。 | 列名 | 描述 | 数据类型 | |------|------|----------| | minutes | 地面周转时长（单位：分钟，取值范围0-1439） | 整数型 | | probability（或国际民航组织（ICAO）代码） | 归一化地面周转时间概率（单个机场数据集的该列包含各机场的ICAO代码） | 浮点型 | (5) 地面周转时间（聚类版）（Turnaround time (clustered)） 本数据集包含聚类簇内的地面周转时间归一化概率分布： 瑞典全国：norm_distribution_turnaround_minutes_SE_clustered.csv 单个机场：norm_distribution_turnaround_minutes_per_airport_clustered.csv 关于“聚类版”标注的说明详见前文方法学章节。 | 列名 | 描述 | 数据类型 | |------|------|----------| | airport（仅在`...per_airport_clustered.csv`文件中存在） | 机场ICAO代码 | 字符串型 | | hour_interval | 聚类的小时区间（例如5-18） | 字符串型 | | turnaround_minutes | 地面周转时长（单位：分钟） | 整数型 | | probability | 指定聚类簇内的地面周转时间归一化概率 | 浮点型 | 附加说明 本研究由瑞典交通管理局资助（项目编号TRV 2023/34443）。