Optimizing the structure of train line plan to improve the capacity of High Speed Railway

Transcription

1 Optimizing the structure of train line plan to improve the capacity of High Speed Railway Lei NIE, Professor School of Traffic and Transportation Beijing Jiaotong University

2 OUTLINE High speed railway network in China Operation performance of HSR in China The characteristics of train line plan Capacity improvement

3 Capacity Shortage of Railway Transport Goods by Railway ( ) Wood: 85% Crude oil: 85% Coal: 60% Steel etc.: 80% Passenger in Railway Station Coal Truck on highway Traffic jam in city, almost no subway The busiest railway in the world: 6% (route length), 25%(converted ton-kilometer) in 2005.

4 The Development of Network Railway, Inner river, Pipeline length(10 thousand km) railway inner river pipeline

5 2014, over 16,000 km high-speed railway in operation, 14.3% in total length 2015,over 19,000 km high-speed railway in operation, 15.8% in total length of km 2016, the operational route lengths of high speed railway has reached 22,000 km, the longest operational route in the world. 5

6 Mid-term and Long-term Railway Network Plan (2017) Year 2020,Route Length 150 thousand km High Speed Railway 30 thousand km, 20% in total, cover 80% of big cities 4 N-S+4W-S 8 N-S+4W-S 6

7 OUTLINE High speed railway network in China Operation performance of HSR in China The characteristics of train line plan Capacity improvement

8 Shorten the Time-space Gap 11h41m 4h48m 20h31m 7h59m city integration before after Beijing Shanghai 11h15m 4h42m before after Beijing Guangzhou 11h23m 6h10m before after before after Beijing Xi'an Beijing Dalian 12h31m 4h24m 9h48m 4h56m before after Shanghai Changsha before after Shanghai Wuhan 9h17m 3h30m before after Harbin Dalian

9 Passenger Flow Volume 2016, 4300 EMU train/day Rapid increase of EMU train passenger Year 2013,670 million, 32.4% Year 2014, 908 million, 40.0% Year 2015, 1106 million, 46.5% Year 2016, 1443million, 52% 30% increasing 9

10 Market Share of Passenger Transportation (Year 2015) Passenger volume (10,000 persons) Passenger traffic turnover (100 million person-kilometers) Total volume Waterway Railway Civil aviation Highway 1,943,271 27, ,484 43,618 1,619,097 30, , Passenger volume (10,000 persons) 27,072 2% 253,484 13% 43,618 2% Passenger traffic turnover (100 million personkilometers) % % % 1,619,097, 83% % Waterway Civil aviation Railway Highway Waterway Railway Civil aviation Highway

11 Transportation during the Spring Festival: The world's largest migration Year Start and finish date Total passengers Raiway (100 million) (100 million) (100 million) Highway Waterway (10 thousand ) Civil Aviation (10 thousand )

12 Revenue of Chinese HSR Beijing-Shanghai high speed railway: First year: 11 billion RMB Second year: 17 billion RMB Third year: 25 billion RMB Without considering depreciation,6 HSRs revenue and expenditure can be balanced Beijing-Tianjing: 120 km, Beijing-Nanjing: 301 km, Beijing-Shanghai: 1318 km, Shanghai-Hangzhou: 202km, Nanjing-Hangzhou:249km, Guangzhou-Shenzhen:126km, From:Huaxia Times From:web information

13 OUTLINE High speed railway network in China Operation performance of HSR in China The characteristics of train line plan Capacity improvement

14 0:00-6:00 Maintenance time High speed railway(beijing South-Shanghai Hongqiao) 14

15 The characteristics of train line plan non-cyclic timetable Train frequency Train stops Train category G, D Origin and destination

16 Train frequency between two stations 16

17 Train departure time distribution at stations 17

18 Complex operation Large-scale network: 22000km 45000km EMU train station: 770 EMU train, 4632 train paths Train operation distance: <100km >2500km >2000km, about 108 train paths 2760km, from Beijing to Kunming 16h24min, from Chengdu East to Fuzhou 18

19 OUTLINE High speed railway network in China Operation performance of HSR in China The characteristics of train line plan Capacity improvement

20 Case 1 Non-cyclic operation

21 A model for analyzing the influence of train line structure on section capacity Non-cyclic timetable: A mixed integer programming model Objective The minimum occupied time: train path compression min Z=max{a i I} min{d j I} min Z= I i 1 i, d j, o i a i, des i i

22 Model Constraints Running time a d r x x is 1 is is is is is 1 is 1 a d r x x y is 1 is is is is is 1 is 1 is Dwell time d a w x is is is is d a w x is is is is Headways s d d M 1 O HD js is ij s d d MO HD s is js ij s a a M 1 O HA s js 1 is 1 ij s 1 a a MO HA s is 1 js 1 ij s 1 Overtaking N s 1 s Oij Oij 1 j 1, j i Train order 列车前后行关系 O s ij O s ji 1 Cross-line train 跨线车的固定到发 k d k is is is Departure time control t d t is is is Maintenance time window d a is is SL e SL b

23 Algorithm The Branch and Bound based on the optimal estimation To solve the large-scale problem: The Segmentation and Scroll strategy used to draw train timetable piecewised CPLEX solver, and the visualization and index statistics of train timetable are realized by MATLAB. HA Freetrain Fixedtrain

24 Case study: Non-cyclic timetable 北京南 廊坊 天津南 沧州西德州东 济南西 泰安曲阜东滕州东枣庄 徐州东 宿州东 1318km, 23 stations G, D Headway: 4min Dwell time at big station: 2min the other station: 1min 大节点 ( 省会城市 ) 小节点 蚌埠南定远镇江南滁州常州北丹阳北无锡东南京南昆山南苏州北 上海虹桥 Beijing-Shanghai

25 Case study: Non-cyclic timetable timetable at Xuzhou-Bengbu 上海虹桥 昆山南苏州北无锡东 常州北丹阳北镇江南 南京南 滁州 定远 蚌埠南 宿州东 车站 徐州东 枣庄滕州东 天窗 曲阜东 泰安 济南西 德州东 沧州西 天津南 廊坊 北京南 0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 24:00 时刻

26 Case study: Non-cyclic timetable timetable at Xuzhou-Bengbu Earliest departure Latest arrived Occpucation time /min real world 6:40 23: optimized 6:48 23: Travel time/min Travel speed /(km/h) Technical speed /(km/h) real world optimized Reduce occupation time by 21min train Travel /min Travel speed /(km/h) Technical speed/(km/h) RW Opt RW Opt RW Opt G D

27 Case 2 Cyclic operation

28 Analysis on section capacity for cyclic timetable (TR. Part C 2016) Integrating capacity analysis with timetabling can reveal the influence of the structure of train line plans and operating on improving capacity utilization. For most capacity analyses and cyclic timetabling methods, the cycle time is a constant. A minimum cycle time calculation (MCTC) model based on the periodic event scheduling problem (PESP) for a given train line plan. A non-collision constraint and a series of flexible overtaking constraints (FOCs) are constructed based on variations of the original binary variables in the PESP. Because of the complexity of the PESP, an iterative approximation (IA) method for integration with the CPLEX solver is proposed. 28

29 Problem and Model Our model based on the PESP and the model in Sparing and Goverde (2013) further ongoing study of our previous paper (Zhang and Nie (2016) on Transportation Part C) Objective: minimize the cycle time T Input: periodic line plan, operation parameters and service requirements Output: minimum cycle time T (important), periodic timetable 06/04/

30 Methods and Algorithms The (ILP) model was coded by MATLAB R2012a and solved by Cplex 12.3 Our extended iterative approximation methods can help Cplex solver reducing the computation time 06/04/2017 Xin Zhang, Lei Nie, School of Traffic and Transportation, Beijing Jiaotong University 30

31 Case study: Cyclic timetable Station 1 Station 2 Station 3 Station 4 Station 5 Station 6 f 1 =3 f 2 =2 f 3 =2 f 4 =1 f 5 =1 f 6 =1 Fig. 12. Hypothetical case : Line plans Fig. 14. Examples of timetables different colors represent different train lines; the blue numbers indicate the numbers of the lines; K = 0; 31

32 Case study: Cyclic timetable min T (min) / km/h / km/h / km/h / km/h / km/h M K (min) Fig. Influence of the regularity constraint and the train speed gap on the minimum T (the MCTC model with the FOCs; the average computation time for all cases is 4,420 seconds). 32

33 Case study: Cyclic timetable min T (min) K=0 K=6 K=12 K=M (average computation time is 3,798 seconds) % 20% 40% 60% 80% 100% The proportion of fast trains ( / km/h) Fig. Influence of K and the proportion of fast trains on the minimum T ( MCTC model with the FOCs ). 33

34 Real world Case: Cyclic timetable Fig. 19. Time-space diagram of the solution for the realworld test case with the FOCs 34

35 Cross-line train Trains which pass/across at least two different railway lines at connection/border stations A train time window is the time span that the train can depart or arrive in, and depends on the requirements of timetabling (usually for passenger transfers in stations) [14:30, 14:40]

36 A train time window for cross-line train Naming rules of the experiments: four factors of train time window are included s Stations Number Span Regularity -s(origin) -m(intermediate) -one time window for one cross-line train -δ in regularity constraint T δ x f ij T + δ, ij f ij (i, j) A regular * means the case without time windows, i.e. the basic case Xin Zhang, Lei Nie, School of Traffic and 06/04/2017 Transportation, Beijing Jiaotong University 36

37 Case study Beijing-Shanghai High Speed Railway Only the impacts of the number of time windows are analysed (time window in origin stations, ten minutes span, δ=0) Time period of line plan is two hours 06/04/2017 Beijing South Langfang Tianjin South Cangzhou West Dezhou East Jinan West Dingyuan Chuzhou Nanjing South Zhenjiang North Dangyang North Changzhou North Wuxi East Suzhou North Kunshan South Property Real-world case Number of stations 23 Number of trains 18 Number of lines (train stop schedules) 17 Line plan 17 types of train stop schedule Taian Qufu East Tengzhou East Zaozhuang Xuzhou East Suzhou East Bengbu South Shanghai Hongqiao 37

38 The structure of train line plan Cross-line train: 69.7% Section Total trains Cross-line % 北京南 - 廊坊 % 廊坊 - 天津南 % 天津南 - 沧州西 % 沧州西 - 德州东 % 德州东 - 济南西 % 济南西 - 枣庄 % 枣庄 - 徐州东 % 徐州东 - 宿州东 % 宿州东 - 蚌埠南 % 蚌埠南 - 南京南 % 南京南 - 上海虹桥 %

39 Case study Box-plot: number of time windows In fact, the proportion of the number of the cross-line trains is more than 50% (for B-S HSR)! nearly half of all trains! In sum, the minimum T increases as the number of time window (the crossline trains) increases. The uptrend is higher at first and when eight cross-line trains are considered, but relatively stable for other values. 06/04/

40 Case study Box-plot: number of time windows The overlaps show that even the number of time windows increases, they still have a chance to obtain low minimum cycle time. Therefore, the significance of the time window location is highlighted. Xin Zhang, Lei Nie, School of Traffic and 06/04/2017 Transportation, Beijing Jiaotong University 40

41 Optimize the structure of Train line plan Reduce Cross-line train Categorize G,D train speed Optimize the Structure Regularize Stop Pattern Utilize Triangle area Reduce Train Stops

42 Future huge traffic demand Year 2016 Passenger:2.75 billion, 2.0 times/person Year 2020 Demand:1.45 billion people 4 times =5.8 billion persons? Objective:4 billion (National Railway Cooperation) 42

43 Thanks!? Question