TY - GEN
T1 - Assessing last mile delivery strategies - A hybrid solution approach
AU - Muriel, Juan E.
AU - Zhang, Lele
AU - Fransoo, Jan C.
AU - Perez-Franco, Roberto
N1 - Publisher Copyright:
© 2021 Proceedings of the International Congress on Modelling and Simulation, MODSIM. All rights reserved.
PY - 2021
Y1 - 2021
N2 - Urban freight is growing faster than other transport activities, and its adverse effects bring consequences to people, the environment and the liveability of cities. Although understanding its dynamics has become a priority for governments, the multiplicity of actors with conflicting objectives makes it a significant urban planning challenge. This paper develops a hybrid (simulation-optimisation) methodology to evaluate the impact of different last mile delivery strategies over the network traffic flow. The model is focused on the use of one type of on-street parking infrastructure: the loading zone (LZ). We refer to on-street LZs as parking areas that occupy space directly on the road lane. The design and management of parking systems, especially on-street LZs, is considered one of the most powerful traffic control measures with a substantial influence on the efficiency of the urban freight system. The methodology considers the decision-making process made by the road users, their interaction, and the variability of stochastic parameters (traffic conditions, competition, cruising, and illegal parking). The framework combines a stochastic cellular automata (CA) traffic microsimulation, with a metaheuristic and a commercial solver. The CA model has two layers, the lower layer describes the road network, its entry and exit points, LZ locations, traffic demand, speed limits, intersections, and traffic light settings. The upper layer manages the agents (private vehicles (PV) and delivery vehicles (DV)), their size, speed, motion, lane changing, vehicle routes, illegal parking decisions and the duration of the delivery stops. To optimise the routes of DVs, we use a greedy randomised adaptive search procedure - GRASP to solve a two-level (trucking and walking) optimisation problem and the CPLEX optimiser to re-optimise mid-route decisions. The model was developed in the Java programming language. The methodology is applied to a CBD network simulating realistic conditions and evaluate three urban logistics strategies: Alternative LZ, Illegal Parking and Last Delivery. Although the results conclude that to minimise the impact of DVs the best strategy is Illegal Parking, important considerations need to be addressed. For instance, the location of the LZs in the study network were equally spread over an edge, with two illegal parking areas. This aspect certainly limited the well-known congestion consequences of illegal parking. For example, LZs that are located close to up- or down-stream intersections with more illegal parking will certainly spread the congestion shockwave to adjacent edges, attaining different results. A similar situation occurs when delivery vehicles parked illegally are blocking the access to buildings, side streets or public transport. The fact that the Illegal Parking strategy derived better results in this study may seem counterintuitive and is likely to be controversial. However, a logical explanation is that the city might be better off by relaxing parking restrictions to help DVs finishing their routes faster, than by tightening regulations that increase DVs cruising time and cause more congestion. Since the benefits and responsibility for the implementation of more relaxed illegal parking measures lie entirely in the city's hands, is an additional incentive to consider it as a feasible traffic management policy.
AB - Urban freight is growing faster than other transport activities, and its adverse effects bring consequences to people, the environment and the liveability of cities. Although understanding its dynamics has become a priority for governments, the multiplicity of actors with conflicting objectives makes it a significant urban planning challenge. This paper develops a hybrid (simulation-optimisation) methodology to evaluate the impact of different last mile delivery strategies over the network traffic flow. The model is focused on the use of one type of on-street parking infrastructure: the loading zone (LZ). We refer to on-street LZs as parking areas that occupy space directly on the road lane. The design and management of parking systems, especially on-street LZs, is considered one of the most powerful traffic control measures with a substantial influence on the efficiency of the urban freight system. The methodology considers the decision-making process made by the road users, their interaction, and the variability of stochastic parameters (traffic conditions, competition, cruising, and illegal parking). The framework combines a stochastic cellular automata (CA) traffic microsimulation, with a metaheuristic and a commercial solver. The CA model has two layers, the lower layer describes the road network, its entry and exit points, LZ locations, traffic demand, speed limits, intersections, and traffic light settings. The upper layer manages the agents (private vehicles (PV) and delivery vehicles (DV)), their size, speed, motion, lane changing, vehicle routes, illegal parking decisions and the duration of the delivery stops. To optimise the routes of DVs, we use a greedy randomised adaptive search procedure - GRASP to solve a two-level (trucking and walking) optimisation problem and the CPLEX optimiser to re-optimise mid-route decisions. The model was developed in the Java programming language. The methodology is applied to a CBD network simulating realistic conditions and evaluate three urban logistics strategies: Alternative LZ, Illegal Parking and Last Delivery. Although the results conclude that to minimise the impact of DVs the best strategy is Illegal Parking, important considerations need to be addressed. For instance, the location of the LZs in the study network were equally spread over an edge, with two illegal parking areas. This aspect certainly limited the well-known congestion consequences of illegal parking. For example, LZs that are located close to up- or down-stream intersections with more illegal parking will certainly spread the congestion shockwave to adjacent edges, attaining different results. A similar situation occurs when delivery vehicles parked illegally are blocking the access to buildings, side streets or public transport. The fact that the Illegal Parking strategy derived better results in this study may seem counterintuitive and is likely to be controversial. However, a logical explanation is that the city might be better off by relaxing parking restrictions to help DVs finishing their routes faster, than by tightening regulations that increase DVs cruising time and cause more congestion. Since the benefits and responsibility for the implementation of more relaxed illegal parking measures lie entirely in the city's hands, is an additional incentive to consider it as a feasible traffic management policy.
KW - city logistics
KW - illegal parking
KW - Traffic simulation
KW - urban logistics
UR - http://www.scopus.com/inward/record.url?scp=85177086968&partnerID=8YFLogxK
U2 - 10.36334/modsim.2021.M1.muriel
DO - 10.36334/modsim.2021.M1.muriel
M3 - Conference contribution
AN - SCOPUS:85177086968
T3 - Proceedings of the International Congress on Modelling and Simulation, MODSIM
SP - 764
EP - 770
BT - Proceedings of the 24th International Congress on Modelling and Simulation, MODSIM 2021
A2 - Vervoort, R. Willem
A2 - Voinov, A. Alexey
A2 - Evans, Jason P.
A2 - Marshall, Lucy
PB - Modelling and Simulation Society of Australia and New Zealand Inc. (MSSANZ)
T2 - 24th International Congress on Modelling and Simulation, MODSIM 2021
Y2 - 5 December 2021 through 10 December 2021
ER -