Independent driving pattern factors and their influence on fuel-use and exhaust emission factors

Abstract

This study is aimed at finding independent measures to describe the dimensions of urban driving patterns and to investigate which properties have main effect on emissions and fuel-use. 62 driving pattern parameters were calculated for each of 19 230 driving patterns collected in real traffic. These included traditional driving pattern parameters of speed and acceleration and new parameters of engine speed and gear-changing behaviour. By using factorial analysis the initial 62 parameters were reduced to 16 independent driving pattern factors. Fuel-use and emission factors were estimated for a subset of 5217 cases using two different mechanistic instantaneous emission models. Regression analysis on the relation between driving pattern factors and fuel-use and emission factors showed that nine of the driving pattern factors had considerable environmental effects. Four of these are associated with different aspects of power demand and acceleration, three describe aspects of gear-changing behaviour and two factors describe the effect of certain speed intervals.

Introduction

Driving pattern affects the emission and fuel-use of vehicles together with other variables such as vehicle, engine and fuel types. Driving pattern is generally defined as the speed profile of the vehicle, but can be expanded to include other parts of driving behaviour, such as gear changing. The concept of driving pattern does not normally include trip generation, choice of travelling mode or route choice. The latter factors are of great importance for the emission and fuel-use and they should generally be included in any holistic analysis of the environmental impact. A model for vehicular emissions and fuel consumption is presented in Fig. 1.

Newman and Kenworthy (1984) drew attention to the “abuse of driving pattern research” when researchers, planners and decision makers draw conclusions on an overall planning level from results gained on a detailed micro-level. Thus, comparing the emission factors of vehicles in traffic jams to that in smoothly flowing traffic has motivated the expansion of roads in urban areas. This expansion is closely related to the general trend of increasing urban sprawl, which leads to more car trips and higher traffic loads and, finally, higher emission and fuel-use. Yet, the criticism of driving pattern research is concerned more with how the results are interpreted and applied than questioning the need for more knowledge. According to Cost 319 (1999) action driving pattern research in general and especially studies in various countries are of high priority to gain a better geographical representation of driving behaviour.

Driving patterns can been described by many parameters; average speed being the most common. Kuhler and Karstens (1978) introduced a set of 10 driving pattern parameters: average speed, average driving speed (excluding stops), average acceleration (for all acceleration phases when a>0.1), average deceleration (for all deceleration phases when r<−0.1), mean length of a driving period (from start to standstill), average number of acceleration–deceleration changes (and vice versa) within one driving period, proportion of standstill time (v<3 km/h, $\text{|a|<0.1}\text{m}\text{/}\text{s}{}^2$), proportion of acceleration time $\text{(a>0.1}\text{m}\text{/}\text{s}{}^2\text{)}$, proportion of time at constant speed $\text{(|a|<0.1}\text{m}\text{/}\text{s}{}^2\text{)}$, and proportion of deceleration time $\text{(a<−0.1}\text{m}\text{/}\text{s}{}^2\text{)}$. These criteria were used to compare driving cycles, i.e., standardised driving patterns used to test vehicle emission and fuel consumption in laboratories. Andr$\text{e}\text{́}$ (1996) reviewed the parameters used to characterise driving patterns and driving cycles. He found that the most common parameters were duration, average speed, acceleration standard deviation, positive kinetic energy, idle period, number of stops per kilometre, running speed (excluding stops), average values of speed, acceleration and deceleration, average duration of running periods, the number of acceleration and deceleration shifts and relative and joint distribution of speed, acceleration and deceleration. Fomunung et al. (1999) used variables representing surrogates for inertial power (acceleration×speed) and drag power (acceleration×speed²) in a model for NO_x emission. Ericsson (2000a) used a set of 26 driving pattern parameters divided into level measures, oscillation measures and distribution measures to characterise different properties of driving patterns.

Many of the parameters that have been used to model emissions and fuel consumption are probably correlated in real traffic. The inclusion of many correlated driving pattern parameters in such a model will induce several problems: the effects will be difficult to present in a condensed way, will be difficult to estimate, and if effects can be estimated, they will be difficult to interpret correctly. The effect of one parameter may in reality be cancelled out by that of another parameter correlated to the first having the opposite effect.

The aim of this study was to find independent measures describing the properties of driving patterns and to investigate which of them has main effect on emission and fuel consumption per kilometre. A set of 19 230 driving patterns from real-traffic urban driving was examined in three steps. In step 1, the driving patterns were described using 62 parameters. 44 parameters were used to describe the occurrence, frequency and levels of speed and acceleration/deceleration of the driving patterns, and 18 parameters were used to describe the engine speed and choice of gears. In step 2, factorial analysis was performed on all 62 parameters to explore the underlying properties or factors that summarise driving patterns. Here, the 62 initially calculated parameters were reduced to 16 independent factors, each describing one dimension of the driving pattern. In step 3 the emission and fuel-use factors were calculated for the driving patterns of the Volvo and of the VW Golf, consisting of a subset of 5217 cases. The driving patterns of these cars were chosen because calibrated emission models were available for these two car models. Both emission models use a mechanistic approach and are designed to simulate the forces that influence the car and the resulting effect on the emission and fuel consumption. The detailed speed profiles of the models in addition to the engine speed and gear level are used as vehicle input variables. Using regression analysis, the relation between the 16 factors describing the driving patterns, on the one hand, and the emission and fuel-use factors, on the other, was examined. Nine of the 16 factors were found to have considerable effect on emission and/or fuel consumption per kilometre.