Estimating Energy Savings from Bus Improvement Options

The potential to achieve significant energy savings in the road transport sector can be a powerful driver to promote bus transport, especially bus rapid transit (BRT) development. This research introduces a spreadsheet tool for making realistic estimates of energy savings due to increased use of buses, with an explicit inclusion of the effect of congestion on traffic flow and fuel consumption. Based on scenarios developed around projected growth in trip demand, changes in vehicle technology, lane expansion, and modal distribution of trips, the model determines typical daily profiles for fuel consumption by vehicle types. A case study has been performed on an urban corridor in the city of Kuala Lumpur to compare energy usage among three scenarios: business as usual, conventional bus lane, and full-scale BRT implementation. The BRT provides significant energy savings over both alternatives, with the greatest savings achieved when locating the BRT in a newly constructed lane. Introduction The transport sector is the world’s primary consumer of petroleum products, accounting for 58 percent of global total final consumption in 2004 (IEA 2007). For developing and developed countries alike, the greatest share of this sector’s Journal of Public Transportation, Vol. 11, No. 3, 2008 20 fuel consumption comes from road transport (U.S. DOE 2007; NEB 2005; IEA 2007; UN ESCAP 2005). In light of concerns about oil price volatility, domestic energy security, and the environmental impact of burning fossil fuels, the road transport sector has been subject to increasing scrutiny over how its energy consumption can be reduced. Energy efficiency has now become a key component in recent initiatives to promote sustainable transport, particularly in the urban context. These include numerous regional and multinational efforts as well as initiatives spearheaded by individual municipalities (see examples in GDRC 2007). Among the different tools promoted to reduce road transport energy consumption, improvement of public bus systems is commonly recognized as a costeffective option that can be implemented in the very near term (Hensher 2007). Under favorable conditions, increasing the modal share of public bus over private transport can achieve significant benefits in both reduced energy consumption and improved air quality due to the higher energy efficiency per passenger-km of bus transit (Romilly 1999; Shariar and Kahn 2003; Hossain and Kennedy 2006). Enhancements to public buses, such as increased frequency, reserved bus lanes, and full-scale bus rapid transit (BRT), can increase transit ridership as long as a supportive transport policy framework is in place. It has been claimed that highquality BRT systems that replace conventional on-street bus services should be as effective as rail-based systems in generating patronage (Graham 2005). In addition, increased modal shift in favor of public transport can result in fewer cars utilizing the same road space with a possible speed advantage and fewer flow breakdown situations. Fewer vehicles moving at a higher speed have an important bearing on the fuel consumption of the urban corridor. To date, most analyses of energy consumption by the transport sector utilize a top-down approach that draws on fuel consumption statistics at the national or regional level. While this method can provide a gross indication of total energy demand by different modes, it cannot capture the effects of modal shift at the operational level, such as changes in traffic congestion and trip travel times. For example, a long-term energy planning tool such as LEAP (SEI 2006) can set a sectoral target of energy consumption and can estimate the energy consumption across various vehicle categories, but it cannot estimate the impact of initiatives that alter traffic flow in localized areas. Incorporating operational details and arrangements into energy consumption estimates is more significant when it comes down to a project implementation level. As an example, approval and implementation of a bus improvement project could receive a significant boost Estimating Energy Savings from Bus Improvement Options 21 with a more realistic estimation of potential energy savings. Utilization of microsimulation models along with the incorporation of vehicular emission functions may prove to be highly demanding in terms of technical know-how and cost implications for many cities. This article introduces a simpler spreadsheet tool that can facilitate the estimation of potential energy savings with efficient public transport alternatives under various traffic scenarios. Development of such a model requires information on vehicular emission and their dependence on driving behavior (i.e., instantaneous speed, acceleration, and idling). A number of research studies (Brzezinski, Enns, and Hart 1999; Biggs and Akcelik 1986; Post et al. 1984) have proposed various mobile source emission models that calculate fuel consumption as an intermediate output to determine total vehicular emissions. While some of these models incorporate realistic driving cycles, changes in travel demand, vehicle aging and other effects, the majority cannot be used to assess the energy use impact of more dynamic variables (i.e., idle time, acceleration/deceleration, etc.) that depend on vehicle operating conditions (Barth et al. 1996). A comprehensive report on emission inventory methodologies (EEA 2005) suggests two methods for including vehicle speed effects on mobile emissions. First, driving conditions can be categorized according to road type (i.e., urban, rural, or highway) and the emissions estimated based on speed-dependent emission factors and a mean vehicle speed for each category. Alternatively, speed-dependent emission functions can be integrated over speed-distribution curves that cover the entire range of driving conditions and their probability of occurrence. In the end, the authors suggest that the added complexity of including speed-distribution curves for calculating mobile source emission inventories may not be justified due to the high uncertainty in estimating vehicle emission factors. This uncertainty may result from a wide discrepancy of emission factors for vehicles of different type and age (Ntziachristos and Samaras 2000). On the other hand, in estimating CO2 emissions and fuel consumption, little variation exists among vehicles of similar engine size. In this case, including actual operating conditions in the energy estimation may be well justified. Such an effort can build from previous work to collect data on instantaneous speed-dependent fuel and emissions curves for different vehicle types (Rakha et al. 2000; Tong et al. 2000). The present study focuses on the development of a modeling tool that includes the above issues in estimating the energy savings for bus transit improvement options. When applied to bus systems, the improvements may range from a simple demarcation of an exclusive bus lane to fully segregated high-quality BRT Journal of Public Transportation, Vol. 11, No. 3, 2008 22 systems. With increased speed in a BRT lane, and potentially in other lanes if the number of private vehicles is reduced, BRT systems have the ability to decrease energy consumption by the transport sector significantly. However, due to technological, operational, and behavioral changes, forecasting the effect of bus system enhancements on energy consumption is not a simple task. For this reason, a decision support tool, the Sustainable Transport and Energy Planning (STEP) model, has been developed to assess the impact of such bus system implementation on the energy consumption along a defined corridor. The following section describes the modeling framework including model mechanisms, model data base, and calibrations. We then introduce transit and traffic scenarios for application of the model in a Kuala Lumpur corridor. Next, we describe the results of energy savings from model application. Discussions and implications of the results are presented in the concluding section. Modeling Framework The STEP model has been developed as a scenario-based planning tool to estimate the total energy consumption along a single traffic corridor for all road transport modes and vehicle types,1 with a special focus on bus transport. Scenarios are based on projections of passenger trip demand, mix of vehicle types, distribution between public and private modes, vehicle occupancy, and the number of lanes. The corridor itself is divided into sections, with each section characterized according to flow type, flow direction, number of lanes, and number of traffic signals or traffic circles. The model has two distinct parts, a traffic model and a fuel consumption calculator. For the former, projected passenger trip demand and other scenario parameters are used to determine typical daily profiles for average speed and volume along the corridor. It is advisable to construct three separate profiles for weekdays and weekends. For the latter, the fuel consumed by each vehicle category is calculated based on vehicle speed, travel time, volume, and fuel consumption curves for each vehicle type. In the case of dedicated bus lanes, volume, speed, and fuel consumption outputs are calculated separately for the bus lanes and nonbus lanes. The advantage of the present approach is that the effects of congestion and unstable vehicle flow are explicitly modeled and used in the estimations of fuel consumption and diverted demand (i.e., unmet demand diverted to an alternate route due to severe congestion). The model flow is summarized in Figure 1. Estimating Energy Savings from Bus Improvement Options 23 Figure 1. STEP Model Structure Traffic Data for Model Development Traffic data from a series of Automatic Incident Detection (AID) camera stations situated along the Cheras Road corridor were provided by the Urban Transportation Department, Kuala Lumpur City Hall. At each AID station, a video camera and image processing unit are used to record continuously the number of passing vehicles, average vehicle speed, vehicle density, and other parameters averaged Journal of Public Transportation, Vol. 11, No. 3, 2008 24 over three-minute periods for each lane and vehicle class. Three vehicles classes are distinguished according to vehicle length: large, medium, and light category. The data were used to formulate the traffic flow models and estimate the model parameters. Data was available from cameras located at eight positions along the investigated part of the Cheras Road corridor. Figure 2 shows the location of the cameras used in the present project. Figure 2. Camera Locations on Cheras Road, Kuala Lumpur Estimating Energy Savings from Bus Improvement Options 25 Traffic Model The objective of the traffic model is to estimate an hourly profile of realized vehicle volumes and travel times based on inputs of hourly passenger trip demand and various scenario parameters. The traffic model is actually a two-state model that estimates volumes and travel times separately for stable and unstable flow conditions. A new approach, described here, has been developed to estimate the percentage of time that traffic will reside in an unstable or stable state, based on the typical trip demand for a given period of the day. An expected value for the realized vehicle flow volume is found by summing stable and unstable volumes weighted by the probability of residing in either state. A similar approach is used to find the total travel time. Stable and Unstable Traffic Flow Models. The distinction between stable and unstable flow is based on the traffic flow speed. When the speed crosses below a threshold value, denoted here as the breakdown speed (sb), the flow is considered unstable, while it is considered stable for all greater speeds. Figure 3 shows sample observations of speed and volume averaged over three-minute periods for an uninfluenced section of an urban arterial road in Kuala Lumpur. The breakdown speed is set here to 40kmph. The upper portion of the data reveals an inverse relationship between speed and volume; as volume increases, the speed gradually declines. This region is considered stable. The lower portion reveals a region of unstable flow where both speed and volume are reduced due to interactions among a high density of vehicles. Figure 3. Traffic Volume as Function of Speed Journal of Public Transportation, Vol. 11, No. 3, 2008 26 Classical volume-delay models can be formulated to approximate the relationship between speed and volume in the stable region. The volume-delay model used to represent the traffic flow during stable conditions was the Bureau of Public Roads (BPR) model (Highway Capacity Manual 2000).