ORNL researchers propose optimization framework for use in real-time feedback systems to improve driving styles with reduced fuel consumption
|Cumulative fuel consumption of the original and optimized Japan 10-15 driving cycle. Source: Malikopoulos and Aguilar. Click to enlarge.|
Studies have concluded that optimizing a driver’s driving style can reduce fuel consumption and emissions by up to 40%; exactly how to achieve that optimization across a large and diverse driving population remains an area of active investigation—and one of great opportunity.
Dr. Andreas Malikopoulos at Oak Ridge National Laboratory and colleague Juan Aguilar have developed an optimization framework based on their assessment of driving style factors that have a major impact on fuel economy. The framework can be used to develop real-time feedback systems to enable drivers to alter driving styles in response to actual driving conditions to be more fuel efficient and environmentally friendly, the two suggested in a paper presented at the 2012 15th International IEEE Conference on Intelligent Transportation Systems (ITSC 2012) in September.
This paper has two main objectives: (a) to investigate those driving factors that have a major impact on fuel economy and (b) to optimize driving styles with respect to these driving factors. In this context, we formulate an optimization framework that aims to modify driving styles with respect to key driving factors.
...Individual driving styles are different and rarely meet the driving conditions posited in testing (e.g., engine optimization with respect to steady state operating points or vehicle speed profiles for particular highway and city driving). The optimization framework adopted here facilitates better understanding of the potential benefits from employing a more conservative driving.—Malikopoulos and Aguilar
One of the most promising approaches identified in other work to optimize driver behavior is to provide the driver with immediate information about the effect of his or her behavior on fuel consumption, they noted. In their work, Malikopoulos and Aguilar identified two factors that have a major impact on engine operation, and so on fuel economy: the stop factor and the coefficient of power demand.
The stop factor is defined as the amount of time during a driving cycle that the vehicle is stopped (i.e., time that the vehicle’s velocity is zero divided by the total duration of the driving cycle). This factor indicates idle engine operation over a driving cycle.
There is a linear correlation between the stop factor and fuel consumption, the authors note; as the stop factor increases, fuel consumption also increases.
The coefficient of power demand provides an indication of the transient engine operation since it is proportional to power demanded by the driver. The power demanded by the driver is proportional to the product of the vehicle speed and acceleration; this product is defined as the coefficient of power demand.
In their analysis, they also found a linear correlation between the coefficient of power demand and fuel consumption.
The stop factor represents a commuting aspect rather than a driving one—i.e., it is route-dependent, rather than dependent on driving behavior, and can be altered only by changing the route. Furthermore, hybridization of vehicles has aimed to address the stop factor by shutting off the engine when the vehicle is stopped and thus eliminating near-idle engine operation. Therefore, Malikopoulos and Aguilar focused on the problem of optimizing a driving cycle with respect to the coefficient of power demand.
To develop the optimization framework with respect to the coefficient of power demand, they used Autonomie—a Matlab/Simulink simulation package for powertrain and vehicle model development developed by Argonne National Laboratory. They built a set of polynomial metamodels to construct an explicit relation between fuel consumption and the coefficient of power demand and so to formulate the optimization problem analytically significantly to reduce computation time. The models can reflect the responses in fuel consumption produced by changes in the coefficient of power demand.
Applying the optimization framework to three driving cycles—Japan 10-15, combined FTP and HWFET, and FTP—they found that the optimized cycles:
Increased travel time 28% in the Japan 10-15 cycle, with a 22.3% improvement in fuel consumption;
Increased travel time in the combined FTP and HWFET by 9.8%, with a 15.9% improvement in fuel consumption; and
Increased travel time 14.1% in the FTP cycle, with a 23.2% improvement in fuel consumption.
In other words, a more conservative driving style that delays destination arrival had a significant impact on fuel consumption. One approach for using the framework to develop a real-time feedback system for a driver would be as follows, as the authors suggest in a separate paper under review for publication in IEEE Transactions on Intelligent Transportation Systems:
The driver first drives the desired route by employing his or her typical style. A flash drive plugged into the onboard diagnostics (OBD) of the engine records two required signals: (a) the vehicle speed profile, and (b) the fuel consumption rate.
After the trip, the driver downloads these two signals in an application running the optimization framework.
The user can select the criteria for optimizing his or her driving style in the optimization problem; for example, an optimized driving style should be no less than a certain value of speed from the original one, or the arrival time should not exceed a certain percentage of the original time.
Visual instruction area in red corresponding to the driver’s acceleration profile when operating the vehicle. Click to enlarge.
The new optimized vehicle speed profile forms the basis of a real-time feedback system— e.g., an actual acceleration profile (red square) should be less than or just overlap the optimal acceleration profile (green square) at each instant of time. (Diagram at right.)
The belief implicit here is that after repeating the same route by following these visual instructions, the driver will eventually learn the optimized driving style. Thus, eventually, the driver will learn the most efficient position of the accelerator pedal to achieve a certain amount of benefit in fuel economy.
...The long-term potential benefits of this driver feedback system are substantial. Drivers will be able to evaluate their driving behavior and learn how to improve their driving styles based on their own preferences. Realizing a more eco-driving style can contribute significantly to sustainable mobility.—Malikopoulos and Aguilar (under review)
Andreas A. Malikopoulos and Juan P. Aguilar (2012) Optimization of Driving Styles for Fuel Economy Improvement. (ITSC 2012 MA8.2)
Andreas A. Malikopoulos and Juan P. Aguilar (under review) An Optimization Framework for Driver Feedback Systems