Quantifying the factors which affect a train's operational energy consumption. James Pritchard Research Fellow in Rail Energy Systems

Transcription

1 Quantifying the factors which affect a train's operational energy consumption James Pritchard Research Fellow in Rail Energy Systems

2 Outline Why is understanding energy consumption important? What are some of the problems with existing data? Empirical data analysis Conclusions

3 Background: Energy consumption Energy consumption is directly linked to emissions There are additional incentives to reduce it $

4 Background: Road vs Rail (2012)

5 Background: Road vs Rail (2012)?

6 Background: A key question?

7 Problems with existing data Much of the available data are from simulations o May not be 100% representative of the real-world o Heavily dependent on input parameters Available empirical data are limited in scope o Often a very limited sample o Often restricted to one route The choice of metric is not always consistent or appropriate o e.g. passenger loading data may be assumed

8 Problems with existing data

9 Factors which may cause variation Type of train Type of service Features of the infrastructure Driving style Time Period (linked to passenger loadings and weather)

10 Empirical Data Analysis

11 Empirical Data Analysis Train Operator 1 Train Operator 2 Data from a fleet of intercity trains, all of the same class Data from three classes of suburban and semi-urban electric trains > 2 years worth of data ~ 1 month worth of data Energy/GPS data recorded at 5 minute Energy/GPS data recorded at one intervals minute intervals Schedule and train allocation data Schedule and train allocation data Comprehensive OTMR data set No additional data about speed, distance or on board functions Energy readings matched to journeys undertaken kwh per seat-km was chosen as the metric

12 Empirical Data Analysis Train Type Seating Density [# seats / train length [m]] Suburban Suburban Suburban Intercity Intercity All Train 1 Train 2 Train 3 Train 1 Train 2 trains Number of journeys sampled Mean distance between scheduled stops [km] Median distance between scheduled stops [km] Standard deviation of mean stop spacing Mean net energy consumption [kwh per seat-km] Median net energy consumption [kwh per seat-km] Standard deviation of mean net energy consumption

13 Empirical Data Analysis Train Type Seating Density [# seats / train length [m]] Suburban Suburban Suburban Intercity Intercity All Train 1 Train 2 Train 3 Train 1 Train 2 trains Number of journeys sampled Mean distance between scheduled stops [km] Median distance between scheduled stops [km] Standard deviation of mean stop spacing Mean net energy consumption [kwh per seat-km] Median net energy consumption [kwh per seat-km] Standard deviation of mean net energy consumption

14 Empirical Data Analysis: All Data

15 Empirical Data Analysis: Intercity Train 1

16 Intercity Train 1: Initial Observations Mean Energy Consumption [kwh per seat km] Standard Deviation Average from literature Calculated from empirical data There is variation Empirical variation is not as big as that found in literature Important to know what is being measured o Influence of regenerative braking system important o Gross energy consumption was found to be kwh per seat-km empirically

17 ANOVA: Intercity Train 1 ScheduleID o Unique integer identifying a particular schedule DriverID o Unique integer to differentiate between drivers TimePeriod o Weekend (Saturday and Sunday 0600 to 2300) o Morning Peak (Weekdays 0700 to 1000) o Evening Peak (Weekdays to 1900) o Night ( ) Month

18 ANOVA: Intercity Train 1 Explanatory Variable Adjusted R2 Ranking ScheduleID DriverID TimePeriod Month Number of Explanatory Variables Explanatory Variables Adjusted R2 2 ScheduleID + DriverID ScheduleID + DriverID + Month ScheduleID + DriverID + Month + TimePeriod 0.475

19 ANOVA: Route and Service Adjusted R2 Explanatory Variable ScheduleID All Data All Suburban Suburban Suburban Suburban Data Train 1 Train 2 Train Intercity Train

20 ANOVA: Route and Service RouteClassification o Up, Down, Regional ScheduleLength o Distance in km MeanStopSpacing o Distance in km MeanStopDensity o # of stops per km ScheduleMinutes ServiceClassification

21 ANOVA: Route and Service (all trains) Number of Explanatory Variables Explanatory Variables Adjusted R2 1 MaxServiceSpeed RouteClassification ScheduleLength ServiceClassification ScheduleMinutes 0.453

22 Variation with mean stop spacing

23 ANOVA: Type of train TrainType MassPerSeat SeatingDensity DesignSpeed TrainClassification TrainLength FleetAge UnitCount # of units running together RegenAvailable working regen. braking

24 ANOVA: Type of train Explanatory Variable Adjusted R2 TrainType MassPerSeat SeatingDensity DesignSpeed TrainClassification TrainLength FleetAge UnitCount RegenAvailable 0.005

25 ANOVA: Type of train Explanatory Variable Adjusted R2 TrainType MassPerSeat + SeatingDensity 0.418

26 Conclusions Energy consumption can vary, but the observed variation in large empirical data set found to be less than variation across existing literature. Choice of metric is important, as is knowing what s being measured. Understanding aspects of the route and schedule is important they do have an impact. Particularly important to adjust figures for suburban trains according to stop spacing. The train can be characterised by the mass per seat and the seating density if the type is unknown.

27 Conclusions Future work Can schedules and calling patterns be optimised to reduce energy consumption? Can we combine data with work on passenger flows to produce accurate per-passenger figures for comparison with other modes?

28 Thank you Any Questions?