Transportation of hazardous materials
Research into the routing of hazardous materials tends to concern evaluating the environmental impact of different routes between two locations rather than scheduling routes between several locations. A main area of interest is the risk associated with a route. Erkut and Verter (1998) define risk as a combination of, for each section of the route, the probability of an accident and the population size affected. The different real-world routes generated by several measures of risk from this basic definition are examined. Karkazis and Boffey (1995) consider further issues on accident impact such as the weather. An overview of such risk evaluations is given by List et al (1991). Kara and Verter (2004) propose a system whereby a regulator restricts access to sections of the road network on the basis of risk assessments and the haulier subsequently optimizes routing decisions over the remaining network.
Environmental impact
The environmental impact of a fleet will be affected by factors other than the routes and schedules used, such as the size of the vehicles and the type of fuel used. Practical measures, such as the way the vehicles are driven, can have an impact on emissions. In the UK the SAFED programme provides driver training to encourage safe and fuel-efficient driving through a wide range of factors. Issues include aerodynamics and loading, braking technique, the use of gears, cruise control and the determination of optimal speeds. Companies can provide efficiency awards to drivers who achieve targets such as using less fuel. These methods emphasize the importance of the fleet’s efficiency to the drivers who will be required to implement the results of any more complex analysis. Such measures have been shown to reduce fuel consumption by between 1.9 and 13.5 per cent (DfT, 2006) in one study and by 4.35 per cent in a before-and-after study of Greek bus drivers (Zarkadoula, Zoidis and Tritopoulou, 2007).
Emissions auditing
Emissions auditing is the process of calculating the amount of greenhouse gas or other pollutants released into the atmosphere by a given activity. When estimating vehicle emissions, a variety of factors can be taken into account, including load weight and distribution, vehicle age, engine size, vehicle design, driving style, road gradient and speed. Speed is the major factor with reference to vehicle routing, and a route generated while optimizing distance may emit more CO2 or other polluting gases due to slower speeds than a longer alternative route.
A simple method of estimating emissions from a vehicle is to take the distance of the planned journey and assume an average driving speed or fuel consumption per mile/km. Such an approach is included in the model of Dessouky, Rahimi and Weidner (2003). However, this assumption implies a linear relationship between such an estimate and the total distance travelled that makes minimizing emissions in such a way equivalent to minimizing distance. A more detailed approach would break each journey down by road type (eg highways, major roads, minor roads, residential streets) and assume an average speed/fuelconsumption for each type. Such an approach is already used in many
software packages to estimate driving times. However, speed, particularly within city centres, has been shown to vary substantially during the course of the day (Eglese, Maden and Slater, 2006; van Woensel, Creten and Vandaele, 2001). Figure 10.1 shows how the average speed for a particular section of a primary road varies. Any estimate of emissions that fails to take this variation into account will be limited in its accuracy. Furthermore, failure to consider congestion reduces the robustness of computer-generated schedules when implemented in the real world. Palmer (2007) has shown that routes took 10 per cent longer in practice than the estimates provided by computer software. Any environmental gains from the use of vehicle routing and scheduling software may be lost due to schedules being infeasible in practice or even ignored by the drivers through lack of faith in their predicted timings.
Congestion
It has already been mentioned that the average speed on a road will vary at different times of day. The main cause of this variability is congestion. Congestion prevents a vehicle from driving at an optimum speed and subsequently has a negative impact on total vehicle emissions. McKinnon (2007) identifies exposure to congestion as one of the key freight variables that the UK government needs to manage in order to reduce CO2 emissions. Figure 10.2 shows the relationship between vehicle speed and fuel consumption (which varies directly in proportion to CO2 emissions). As speed decreases below the optimal level, considerably more fuel is used. To make matters worse, congestion forces driving in a stop–start manner, which results in increased fuel consumption and emissions as the vehicle
accelerates and brakes instead of travelling at a steady speed. This means that estimates of fuel consumption based on vehicle test cycle data may not accurately represent the fuel used in typical driving conditions, as discussed in McKinnon and Piecyk (2008). Modern technology, particularly with the advent of GPS devices, allows the monitoring of vehicles. Data from vehicles are stored and then transmitted to a central location and analyzed. Typically, speed and location (accurate to a particular section of road) are recorded; however, modern devices also include information on fuel flow. It is hoped that in the future information on fuel consumption collected in this way will aid emissions auditing. However, in the meantime, data on vehicle speeds have been compiled so that the average speed for a section of road at each time of the day is known. This provides a way of measuring the congestion that occurs on a daily basis.This data has been used by Eglese, Maden and Slater (2006) and Maden (2006) to find solutions to VRSPs that minimize the total driving time. The aim of this approach is to produce more reliable vehicle schedules, but a potential environmental benefit is the construction of routes that tend to avoid congestion and the emissions produced in slow-moving traffic. Such an approach will also provide more robust schedules that will reduce overtime and improve customer satisfaction through more punctual deliveries and collections.
Conclusions
Modern computer software systems are able to produce efficient sets of vehicle routes for road freight deliveries that produce economic savings and environmental benefits compared with manual planning systems, particularly when the customers and demands vary from day to day. Modern developments in tracking technology are opening up new opportunities to improve vehicle routing and scheduling further by taking account of expected congestion, and to modify routing plans dynamically by taking into account current traffic conditions. Baumgaertner, Léonardi and Krusch (2008) describe a qualitative survey of trucking companies and software providers that is used to assess the importance of computerized vehicle routing systems and other technologies to reduce fuel consumption and CO2 emissions.
Vehicle routing and scheduling is only one of many factors that will influence the economic and environmental performance of a distribution system, but good routing and scheduling have the potential to contribute to reductions in greenhouse gas emissions and other pollutants.