Modelling the ecological and economic footprint of last-mile parcel deliveries using open data

A case study for Lyon

Sebastian Hörl

6 September 2023

6 September 2023

hEART 2023

LEAD Project

• H2020 Project from 2020 to 2023
   
• Six living labs with different innovative logistics concepts
  • Lyon, The Hague, Madrid, Budapest, Porto, Oslo
  • One partner for implementation and one for research each
     
• Development of a generic modeling library for last-mile logistics scenario simulation and analysis

https://www.leadproject.eu/

Lyon Living Lab

• Implementation of an urban consolidation centre (UCC) to collect the flow of goods and organize last-mile distribution
   
• Due to data availability
  • Focus on large-scale analysis
  • Focus on B2C parcel deliveries
     
• Research questions
  • What is the (approximate) impact of daily parcel deliveries in a Metropolitan area like Lyon in terms of emissions and energy consumption?
  • Which effects do specific interventions have?

Methodology: Overview

Methodology: Synthetic travel demand

Synthetic travel demand has been generated for Lyon in order to perform agent-based mobility simulation of all residents' movements.

0:00 - 8:00

08:30 - 17:00

17:30 - 0:00

0:00 - 9:00

10:00 - 17:30

17:45 - 21:00

22:00 - 0:00

Methodology: Synthetic travel demand

Synthetic travel demand has been generated for Lyon in order to perform agent-based mobility simulation of all residents' movements.

Methodology: Synthetic travel demand

+

=

Synthetic travel demand has been generated for Lyon in order to perform agent-based mobility simulation of all residents' movements.

Methodology: Synthetic travel demand

=

+

=

Synthetic travel demand has been generated for Lyon in order to perform agent-based mobility simulation of all residents' movements.

Methodology: Synthetic travel demand

eqasim-org/eqasim-java

eqasim-org/ile-de-france

Methodology: Parcel demand

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

Methodology: Parcel demand

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

Methodology: Parcel demand

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

Methodology: Parcel demand

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

Methodology: Parcel demand

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

• Based on synthetic population, find average number of purchases delivered to a household defined by socioprofesional class, age of the reference person and household size per day.

Methodology: Parcel demand

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

• We now the average number of parcels, but we do not now the distribution of the number of parcels for a household on an average day.
   
• We know it must be non-negative, and we know the mean.
   
• Without additional data, we assume maximum entropy distribution, which is Exponential in this case.

Methodology: Parcel demand

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Methodology: Parcel demand

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Methodology: Parcel demand

Gardrat, M., 2019. Méthodologie d’enquête: le découplage de l’achat et de la récupération des marchandises par les ménages. LAET, Lyon, France.

+

Based on sociodemographic attributes of the households, parcels are generated for the city on an average day.

Methodology: Study area

How many parcels need to be delivered on one day?

Methodology: Study area

Perimeter

• City of Lyon
• Grand Lyon metropolitan region (dashed)
• Bordering municipalities including relevant logistics infrastructure

Demand

• 1.6M persons
• 790k households
• 16,252 parcels

How many parcels need to be delivered on one day?

Methodology: Distribution centers

From where do operators delivery the parcels?

Methodology: Distribution centers

Approach

• Facilities per operator extracted from SIRENE
• Geolocated using public BAN API
• La Poste: Facilities with 20+ employees

From where do operators delivery the parcels?

Methodology: Market shares

How many parcels are delivered by each operator?

Methodology: Market shares

How many parcels are delivered by each operator?

Approach

• For some operators, we know the annual national volumes from gray literature
   
• We know that La Poste (Colissimo + DPD + Chronopost) add up to about 65% of all parcels in France
   
• For the rest, we approximate their market share using their annual turnover values

Methodology: Market shares

How many parcels are delivered by each operator?

Methodology: Assignment

How many parcels are delivered by each distribution center?

Methodology: Assignment

How many parcels are delivered by each distribution center?

Approach

• For each parcel, sample an operator
   
• Find the operator's distribution center that is closest (shortest distance) to the parcel

Methodology: Assignment

Approach

• For each parcel, sample an operator
   
• Find the operator's distribution center that is closest (shortest distance) to the parcel

Outcome

• Nine centers with 300+ parcels
• Remaining centers with less than 300

How many parcels are delivered by each distribution center?

Methodology: Cost structures

What influences operators decisions?

Methodology: Cost structures

What influences operators decisions?

Salaries

Vehicles

Distance

Daily cost

+

+

=

Methodology: Cost structures

What influences operators decisions?

Assumption (from grey literature)

• 1,300 EUR net per month
• 1,700 EUR gross per month
• 3,400 EUR staff cost per month
   
• 25 active days per month
• 136 EUR per day

Salaries

Vehicles

Distance

Daily cost

+

+

=

Methodology: Cost structures

What influences operators decisions?

Salaries

Distance

Daily cost

+

+

=

• We examined long-duration rental offers (LLD) of French vehicle manufacturers

Vehicles

Methodology: Cost structures

What influences operators decisions?

Daily cost

=

• Insight: Rental costs depend linearly on the transport volume

Salaries

Distance

+

+

Vehicles

Methodology: Cost structures

What influences operators decisions?

Daily cost

=

• Seven prototypical vehicle types: 3 sizes thermal or electric plus cargo-bike

Salaries

Distance

+

+

Vehicles

Methodology: Cost structures

What influences operators decisions?

Daily cost

=

• Distance-costs depend on consumption of fuel and electricity
• Multiplied by fuel or electricity prices (example 1.45 EUR/L and 9ct/kWh)

Salaries

Distance

+

+

Vehicles

Methodology: Cost structures

What influences operators decisions?

Daily cost

=

• Additional information from our manufacturer analysis: Emissions
• Assuming 90 gCO2eq/kWh for electric vehicles

Salaries

Distance

+

+

Vehicles

Methodology: Optimization

Minimize costs

Methodology: Optimization

Heterogeneous Vehicle Routing Problem

• Minimize cost per distribution center
• Operator can choose vehicles (7 types) and routes
   
• Operator must deliver all assigned parcels
• Maximum active time per day 10h
• Active time is travel time + 120s delivery + 60s pick-up per parcel
   
• Vehicles can not exceed capacity (we assume 10 parcels per m3)
• Multiple tours per vehicle are allowed
   
• Vehicles start and end the day at the distribution center

Minimize costs

Methodology: Optimization

Heterogeneous Vehicle Routing Problem

• Minimize cost per distribution center
• Operator can choose vehicles (7 types) and routes
   
• Operator must deliver all assigned parcels
• Maximum active time per day 10h
• Active time is travel time + 120s delivery + 60s pick-up per parcel
   
• Vehicles can not exceed capacity (we assume 10 parcels per m3)
• Multiple tours per vehicle are allowed
   
• Vehicles start and end the day at the distribution center

Minimize costs

Data

• OpenStreetMap network
• Extracted using osmnx
   
• Distance matrix between parcels and depot
• Travel time matrix using congestion factors
   

Solver

• Open-source
• VROOM

Visualisation platform

Scenarios and Results

• Baseline 2022: Estimate today's situation with today's price structure and demand
   
• Baseline 2030: Parcel delivery demand increased 100%
   
• Quantitative mitigation scenarios
  • Increasing the price of thermic vehicles
  • Implementing a substantial carbon tax
     
• Qualitative mitigation scenarios
  • Forbidding thermic vehicles in Lyon's LEZ
  • Forbidding thermic vehicles in the whole region

(different from abstract)

Scenarios and Results

(different from abstract)

Today vs. Future

Scenarios and Results

(different from abstract)

Today vs. Future

Scenarios and Results

(different from abstract)

Today vs. Future

Scenarios and Results

(different from abstract)

Today vs. Future

Scenarios and Results

(different from abstract)

Today vs. Future

Scenarios and Results

(different from abstract)

Today vs. Future

Scenarios and Results

(different from abstract)

Mitigation

Discussion & Open questions

• Validation
  • What data can we use to validate the model?
  • Validation of individual operators?
     
• Coherence
  • Are our cost structures coherent? Did we miss some aspects?
     
• Replicability
  • Data available anywhere in France (but ADM survey only for Lyon)
  • Theoretically applicable anywhere in France

Next steps

• Integration of new components
  • Integration of UCC
  • Pickup points
  • Parcel lockers
     
• Extension to Copenhagen (and B2B)
  • Part of the Horizon Europe project DISCO

Questions?

https://slides.com/sebastianhorl/heart23

sebastian.horl@irt-systemx.fr