Modelling empty runs in regional freight models

1
Modelling empty runs in regional freight models

• Ole Kveiborg and Mikal Holmblad
• Danish Transport Research Institute
• Technical University of Denmark

COST-WATCH, Torino October 2007
2
Empty trips the missing 25

• In Danish National road freight transport
  statistics empty trips account for 25
• Trips not veh. km
• Focus on
• Reduction in CO2 emissions and energy use
• Externalities
• Competition
• Congestion
• Optimisation
• We do not have much knowledge about empty runs
• The relation between runs with and without load
• Empty running is not the difference between runs
  in opposite directions
• Empty and loaded vehicles cross in opposite
  directions
• What are the causes for the trips without load?

3
The contents

• How can empty trips be modelled?
• Empty trips and trip chains
• Model estimation results
• Two models compared

4
Generating empty trips

• Freight flows not balanced between A and B
• But over long run vehicles return to their base
• National trips close to out-return in one day
• A larger un-balance lead to more empty trips
• Empty trips in both directions

5
Generating empty trips

• Empty running is wasteful
• Trucking industry try to get load on all trips
• Contracts
• Through distribution centresor co-operation with
  other companies
• Trip chaining
• Reduced load
• Eksternal regulation
• Drive/rest legislation
• Maximal allowable load
• Distance
• Competition in trucking industry
• Type of goods
• Type of vehicle

6
Modelling empty running

• Empty running has not been a focla area in
  transport models
• Simple ad hoc models
• Knowledge of empty running only at aggregate
  level
• Influential factors are not known and have only
  had limited influence in models
• The systematic patterns often not modelled
• Contrary to the large effort used at modelling
  trips with load

7
Modelling empty trips I

• A simple model
• Empty running is a fixed factor related to trips
  with load

?

• Problem
• Increasing load from A to B increase empty trips
  from A to B (and not from B to A as desired)

8
Modelling empty trips II

• A simple model (Nortman og van Es, 1978)
• Empty running related by a fixed factor on trips
  in opposite directions

?

• Problem
• More load from A to B does not influence empty
  trips between A and B (a reduction in empty
  trips)
• Empircal works have indicated that total number
  of trips is independent of loads

9
Modelling empty trips III

• A model based on probability I (Hautzinger, 1984)
• Empty trips depend on proportion of loads in both
  directions

• Problem
• Assumes that trips are always out- and return
  trips

10
Modelling empty trips IV

• Freight transport is more complex
• Trip chains

11
Modelling empty trips V
12
Modelling empty trips VI

• A model based on probability II (Holguin-Veras og
  Thorsen, 2003)
• Empty trips between i and j is sum of direct
  return trips and trips being part of trip chains

• Empty trips in trips chains is a conditional
  probability

13

• Prob. That a trip from h to j continues to i
  depends on
• Distance between i and j
• Distance already travelled
• The amount of load between zones

14
An alternative compromise

• This model take distance into account, but is
  independent of zones

15
Data

• Travel diary
• Loaded and empty trips (county level)
• Total loads
• Travelled distance
• Vehicle types
• Here solo and articulated

16
Data empty trips depending on distance
17
Data compared to total number of trips
18
Results
19
Results explanatory power
20
Results explanatory power
21
Results distance only
22
Conclusion

• Important to focus on empty trips
• Most models use simple relations
• Possible to include more realism in models using
  existing data
• Apparently good model fit
• Large variation in precision on individual
  observations
• Empty running largest on short distances
• However, distance non-significant
• Mostly included in zonal relations
• Large zones means too little variation in
  distance
• Remove zonal relation leads to small influence
  from distance

23
Thank youQuestions?

• Ole Kveiborg
• OK_at_DTF.DK