Optimizing Urban Freight Deliveries: from Designing

In this article, we present efforts towards building a city-wide freight delivery rounds optimization system aimed at leveraging city traffic information to optimize professional vehicle rounds. This project, called SmartDeliveries, has been developed as part of the OptimodLyon project. We present the system’s architecture and general objective. Then we present early evaluations which have led us to identify three main technical challenges to address, but also to quantify the potential productivity gains for its users: 18% in distance savings and 11% in time. This could lead, if the system were to be generalized at the scale of the city, to a reduction in as much as 5% in traffic, and to corresponding savings in emissions, which are the primary motivation of the project leaders. Finally we discuss the remaining challenges to address, from a technical as well as an organizational standpoint, to introduce real life implementation of the system. INTRODUCTION Context Traffic management in major cities faces well-known but major challenges. While mobility is directly tied to economic growth and prosperity, road networks are saturated. Costly attempts in the 1960’s and 1970’s to adapt European cities to massive car influx have yielded negative externalities, demonstrating the lack of sustainability of the approach. The current approaches in Intelligent Transportation Systems (ITS) favour, among others, to rely on better information processing from all the actors involved to develop alternate mobility patterns, for instance encouraging the use of public transports and cycling, or deferring rush hour travels. OptimodLyon (http://optimodlyon.fr/en/) is a three year project supported by the French agency for the environment, in which eight industry partners, among which IBM, four academic institutions and the metropolitan authority of Lyon (called Urban Community of Lyon) work together to improve urban mobility through better collection, processing and distribution of mobility information. The flagship application of OptimodLyon is a real-time, predictive, multi-modal journey planner. This application lets travellers plan their journey through a combination of car, public transportation and bike sharing with the best available information. The expectation is to enable a reduction of 1% of the car modal share on the city, resulting in saving 25 kTe carbon of emissions per year and significant traffic improvements during rush hours, at a very reasonable cost compared to the amount of public works that are usually needed to obtain similar levels of decongestion. Planned, professional urban mobility Despite environmental concerns, there is a large portion of road traffic that will not lend itself easily to modal transfer. According to the (Academie des Technologies, 2009) report, 35% of road trips carry goods rather than people. If we account for the important portion of movements that involve moving both goods and persons to deliver a value added service (such as maintenance and construction visits, catering, sales visits...), an even larger portion of existing road traffic needs to be accounted for. We call this portion of the traffic we focus on “professional urban mobility” and we estimate (counting campaigns performed internally) that it covers at least 50% of daytime urban traffic. Most policies aimed at limiting private vehicle use affect almost equally professional urban mobility, which could impact economic activity negatively. Any policy aimed at limiting the negative externalities of urban car travel must therefore target this portion of the traffic properly. Businesses are also acutely aware and affected by the cost of mobility and have direct incentives to rationalize it. For instance, fuel represents 18% of the total costs of a small freight operator employing 200 persons (undisclosed personal communication). Current estimates (CERB, 2007) consider that 20 minutes lost in urban traffic amount to about eight euros for such an operator, which represents, by-and-large, its gross financial margin for a half-day vehicle round. We can therefore consider there is a synergy of goals between public authorities and the private businesses in curbing congestion and optimizing trips. The shared objective of both public and private stakeholders is to tend towards the minimization of vehicle-kilometers and time spent on the road. Our work stems from the observation that a large portion of professional mobility demand is actually planned. Somewhere, in the information system of the companies, the intent to move is known in advance, at a close informational distance from an internet connection. If the city can securely obtain this information and use it to improve its traffic forecast, providing in return relevant suggestions that can reduce the cost of mobility, both parties could benefit from this exchange of information. Synergy between cities and freight operators The expected gains of a cooperative data exchange for cities and road users are as follows:  Freight trip planning data constitutes a new source of information for traffic regulation, providing an advance view of freight mobility, which represents a sizable portion of traffic.  Data exchange provides a new regulation tool, enabling to divert a sizable portion of traffic outside of congested areas.  Because the service may measurably save time and money to its users, it can be turned into a revenue source for the city to leverage its traffic regulation investments. For its users, we expect a high profitability, as distance and time savings shall directly impact their revenue. Finally, the costs of exploiting such a solution shall be fairly small, as all it requires is an information infrastructure whose core hardware aspects are already in place: internet, mobile phones, and traffic data. Of course this is still an idealized view of the system we propose. In the remainder of this paper, we will present an implemented prototype, then discuss the evaluations we have performed with three companies and 4000 time-stamped and geolocalized tours. These tests led us to identify three major technical challenges, as well as helped us quantify the potential benefits of the system. In the final section of this paper, we will discuss the remaining challenges, which are equally divided between technical and organizational considerations. OVERVIEW OF THE PROPOSED SYSTEM SmartDeliveries (Baudel et al, 2013) is a software service and a set of applications comprising the following elements (Figure 1):  A situation monitor maintains a model of the current and forecast traffic conditions for the whole city. It connects to the city information services (such as data.grandlyon.com) to gather planned events and public works information, real time and forecast traffic conditions, weather data. It provides means to plan routes and current and future travel time estimates.  A demand monitor gathers users’ mobility plans. It is accessed either via a web application (for occasional users) or a dedicated web service (connecting to the firm’s information services). The plans are provided as plain csv files comprising a list of destinations, together with constraints on the trips, such as time and precedence constraints.  A routes monitor creates, optimizes and follows all planned trips in real time, adapting to the observed route taken (as the driver stays in control of the plans), and provides optimization suggestions and alerts when there are traffic events from the city or deviations from the initial plans. It connects to a mobile application that acts as a route planner and provides traffic alerts. Figure 1: Smart Deliveries architecture Once fully implemented, this architecture shall enable the computation of global optimization plans, enabling to spread traffic in time and space when potential delivery parking spots congestion is detected. The key implemented components for now are the travel time estimator and the time-dependent, robust optimization service, which will be described hereafter.