Eco-Drive Experiment on Rolling Terrain for Fuel Consumption Optimization

Chapter 1. Introduction

Accomplishments in individual vehicle control have laid the foundation for more advanced control that governs interactions among multiple connected vehicles (CVs) and can produce resultant effects on highway traffic performance. Efforts have been made to extend adaptive cruise control (ACC) to cooperative adaptive cruise control (CACC) to further improve vehicle-following efficiency through multivehicle communication that takes stability (Vugts, 2010), traffic throughput (Van Arem, 2005), and energy and environmental impacts (Malakorn & Park, 2010; Ma et al., 2016) into account. Limited attempts have been made to extend these developments to other, more complex infrastructure geometries, such as ramp merges (Park & Smith, 2012) and intersections (Drenser & Stone, 2012; Zhou et al., 2016; Hu et al., 2015). However, most of these studies apply simulation to evaluate the effectiveness and benefits of control algorithms. Field experiments must now be undertaken to show that new technologies function as expected in practice and to collect data on control system performance under nontheoretical conditions (e.g., mixed traffic, system delay, or inaccurate input data) as basis for further system improvement.

Eco-drive is one of the many research topics that address the issue of vehicle fuel efficiency. CV data are now being leveraged to allow vehicles to cooperate better within the current and future environments in terms of traffic conditions, signal timing, and terrain information. This study investigates the use of vehicle automation and mobile communication technology to derive the maximum benefits from eco-drive. The concept of vehicle-to-infrastructure (V2I)-based eco-drive is illustrated in figure 1. Traffic management centers (TMCs) maintain databases of all roadway profiles (i.e., location, horizontal and vertical curves, work zones, etc.). In this concept, TMCs predefine a list of roadway segments on which automated eco-drive is recommended or enforced. These segments are selected because of their potential for significant fuel savings according to segment characteristics. Once the eco-drive vehicle receives information from the TMC, an onboard computer equipped with a preloaded algorithm will design a recommended trajectory (i.e., speed profile) for the vehicle to traverse the entire rolling segment. This algorithm should also account for other factors such as vehicle operating capability, driver comfort, safety, and speed limits. Advanced algorithms may also account for the existence of a front vehicle (via vehicle-to-vehicle (V2V) communication) and downstream traffic congestion through the addition of speed harmonization (Ma et al., 2016) components to the algorithm. But these advancements are out of the scope of this study and will be left for future exploration. In this paper, the term “eco-drive” is used to refer to this specific concept of V2I-based eco-drive on rolling terrains.

Note that this roadway profile information is usually collected through roadway survey and design documents from State department of transportation (DOT) construction divisions, which are generally only available for newly constructed, major roads. In the future, with increasingly accurate and prevailing vehicular onboard sensors, CVs (eco-drive or not) can potentially send real-time information related to roadway geometry (e.g., latitude, longitude, altitude) to TMCs to update roadway profile databases, particularly in cases where changes in geometry occur, or to collect information on roadways where no profile data are available. This study also explores advanced Global Position System (GPS) service to extract roadway profile data.

Source: FHWA
Figure 1. Concept of V2I-based eco-drive on rolling terrains.

Given that roadway geometry data collected via connected vehicle technologies are much more accessible than those from design/survey documents, researchers believe that CV geometry data are more likely than survey data to be used as input for potential eco-drive applications. Therefore, this experiment purposefully used CV geometry data as input. It is one of the many very important designs this experiment adopts to ensure the eco-drive application is tested under the most realistic environment.

Past research validated that a 6 percent increase in roadway grade resulted in a 40 to 94 percent increase in fuel consumption (Park & Rakha, 2006). Another study confirmed that fuel economy on flat routes is superior to that on rolling or mountainous routes by approximately 15 to 20 percent (Boriboonsomsin & Barth, 2009). However, in theory, if no energy is wasted, vehicles driving on rolling terrain should consume the same amount of fuel as vehicles driving on flat roads. The only difference between the two is the fact that the vehicles on rolling terrain constantly have energy transferring between potential energy and kinetic energy. Therefore, these studies concluded, the increase in fuel consumption resulted in additional unnecessary waste, which can be avoided or reduced by optimizing vehicle states. Some studies have investigated vehicle speed and powertrain optimization (Hellström, et al., 2010), but these approaches are over simplified or are not yet ready for real world implementation. For example, these particular approaches only consider constant slope scenarios. Further, the algorithms used—such as dynamic programming—are computationally intensive and difficult to apply in real time.

A recent study shows that using CV technology on a hybrid electric vehicle (with speed and powertrain optimization algorithms) could gain up to 17 percent fuel savings on rolling terrain (Hu et al., 2016). Further research for regular gasoline engines shows the benefit of the proposed optimal controller is significant compared to cruising on rolling terrain at a constant speed, with fuel saving ranging from 11.7 to 16.3 percent (Hu et al., 2016). Both studies show great potential in significantly reducing fuel consumption for a stretch of roadway with changing terrain, and the proposed algorithms using the Relaxed Pontryagin’s Minimum Principle (RPMP) are computationally efficient and applicable in real time. While these new algorithms prove effective in simulation with many assumptions, it is necessary to test these algorithms in real-world scenarios to better understand the algorithm performance, and thus improve them to optimally control vehicles for eco-drive.

Following this introduction, this report provides a brief review of the innovative vehicle control platform, including algorithm design and system logic. Then the experimental design for the field experiment and testing environment is described, followed by a discussion of the experiment. The last section discusses conclusions and future research.