HETEROGENEITY IN HYBRID INVOLVED CRASH SEVERITIES: AN EXPLORATORY ANALYSIS USING THE HIERARCHICAL MIXED LOGIT MODELWITH HIERARCHICAL HYBRID VEHICLE ATTRIBUTES 

ABSTRACT

 

In recent years, hybrid vehicles have been increasingly accepted by consumers because of environmental, economic and nonrenewable (e.g. fossil fuels) resource concerns.  In addition to the differences in fuel source, hybrid vehicles are different from the traditional fuel-engine vehicles in terms of ride characteristics, ride noise, vehicles weight, etc. For example, the power of hybrid vehicles comes from the combination of a fuel engine and an electric engine; when hybrid vehicles are using an electric engine, overall traffic noise is much lower than it would be with a traditional fuel engine vehicle. In addition, because of the extra battery, hybrid vehicles are always heavier than fuel engine vehicles in the same class. Factors such as weight distribution, vehicle noise interactions with driver and other vehicles can be sources of potential shifts in severity distributions involving hybrid vehicle crashes.  In particular, how the attributes of a hybrid vehicle affect our inferences on crash severity propensities is a largely unaddressed issue in the literature.  With the emergence of the hybrid vehicle market, this issue is bound to become prominent in our address of severe crashes in the nation, especially the FHWA’s target of lowering fatalities.

This thesis attempts contribute some insight into the impact of hybrid vehicle attributes on crash severity propensities by using a random parameter (mixed) logit model to predict the crash severity in crashes involving hybrid vehicles. Three levels of severity are considered: (a) property damage only; (b) possible injury; and (c) injury. The injury category combines the fatality, severe (incapacitating) injury and evident injury categories.  Using 5 years (from 2006 to 2010) of statewide data from reported crashes in Washington State involving hybrid vehicles, this thesis develops a mixed logit model for the severity of crashes involving hybrid vehicles, by considering factors such as roadway conditions, environment factors, driver and passenger attributes and vehicle characteristics. The mixed logit model is the state of the art in modeling crash severity.  However, the extant literature does not include a form of the mixed logit model where the random crash severity parameters are evaluated hierarchically.  The hierarchical model allows for the identification of factors that can influence of the mean of the random parameters in the mixed logit.  In this thesis, the hierarchical influences consist of hybrid vehicle attributes, due to the fact that at least one of the vehicles involved in the crash dataset is a hybrid.  The research results shows that the hierarchical mixed logit is a plausible approach for gaining insight into the particular impact of hybrid vehicle attributes on crash severity parameters.

 

 

 

TABLE OF CONTENTS

List of Figures ………………………………………………………………………………………………………….. v

List of Tables …………………………………………………………………………………………………………… vi

Acknowledgements …………………………………………………………………………………………………… vii

Chapter 1 Introduction ………………………………………………………………………………………………. 1

Chapter 2 Literature Review ………………………………………………………………………………………. 4

Chapter 3 Empirical Setting ……………………………………………………………………………………….. 9

Chapter 4 Methodology …………………………………………………………………………………………….. 14

Chapter 5 Model Estimation ………………………………………………………………………………………. 17

5.1 Model at drivers and occupants level ………………………………………………………………. 17

5.2 Model at crash level ……………………………………………………………………………………… 35

Chapter 6 Conclusions and Recommendations ……………………………………………………………… 41

References ………………………………………………………………………………………………………… 45

Appendix. A Vehicle information Data ………………………………………………………………… 47

Appendix. B Master variable list …………………………………………………………………………. 55

Appendix. C Random parameters and heterogeneity in means variable list ……………….. 89

Appendix. D Mixed logit model of crash severity involving hybrid vehicles at crash

level ………………………………………………………………………………………………………….. 101

Chapter 1 

 

Introduction

In the past 10 years, with requirements of environmental protection, fuel efficiency, as well as the evolution of hybrid vehicle manufacturing, hybrid vehicles from different manufactures (e.g. Toyota, General Motor, Ford, etc.) have been accepted by more and more consumers, with respect to much stronger fuel efficiency performance than traditional fuel engine vehicles, better ride comfort with lower ride noise, better vehicle safety features for drivers and passengers, more environmentally friendly, perceived resale value and a better than expected reliability in long term use.

The field of crash modeling began with the modeling of the relationship between roadway characteristics and crash frequencies.  The Poisson and negative binomial models of the

1990s (Jones et al., 1991; Shankar et al., 1995; Poch and Mannering, 1996; Milton and Mannering, 1998) were the front end of this trend; in the mid to late 90s, the techniques evolved to include crash data with zeros, through the use of zero-inflated negative binomial models (Shankar et al., 1997; Carson and Mannering, 2001; Lee and Mannering, 2002); and through negative binomial with random effects models to account for panel data (Shankar et al., 1998); through the Conway–Maxwell–Poisson generalized linear models (Lord et al., 2008); the random parameter negative binomial models (Sharma, 2007; Anastasopoulos and Mannering, 2009) and dual-state negative binomial Markov switching models (Malyshkina et al., 2009).

When comparing the hybrid vehicle with traditional fuel-engine vehicle, it is worth noting that, the hybrid vehicle has some characteristics, which the traditional fuel-engine vehicles does not have. First, the power of hybrid vehicles comes from the combination of two different resources, an electric engine and a fuel engine. Second, with an extra electric engine in the car, the structure of hybrid vehicles is different from traditional fuel-engine vehicles. Third, since hybrid vehicles need an extra much bigger battery to provide the power for the electric engine.  Compared to fuel-engine vehicles in the same class, hybrid vehicles are heavier and the weight distribution is different.  For example, the curb weight of 2015 Toyota Camry hybrid’s weight distribution is 59% to the front and 41% to the rear, while the traditional Camry of the same vintage has a 63/37 weight distribution split.  The hybrid weighs 3,565 lbs. while the gas engine model weighs 3,460 lbs.

From a user perspective, some drivers prefer the much quieter ride experience of the hybrid. The quieter ride complicates the severity of interactions with external sources such as pedestrians and other vehicles.  Further, when comparing the electronic assist systems between hybrid and traditional fuel-engine vehicles in the same class, hybrid vehicles are usually equipped with more driver assist system features as well as active and passive safety systems.  There appears to be no explicit safety motivation for this characteristic among hybrids, the motivation primarily being a marketing one.  However, when one considers a fuel conscious driver being exposed to better safety technology, compensating behavior can come into play, causing hybrid drivers to interact with their vehicle and external sources in a manner not consistent with traditional vehicle drivers.

Due to the various factors highlighting the differences in hybrids versus traditional gas engine models, it appears the heterogeneous sources driving factors that influence crash severity in hybrid involved crashes can be significant.  To analyze this in detail, this thesis explores the analysis of the distribution of injury level of the most severely injured person in hybrid involved crashes reported for Washington State.  As a first step, this thesis outlines an analytical framework to evaluate the applicability of the mixed logit model to model the heterogeneous influences of hybrid vehicle attributes on crash severity.  A comparative analysis of traditional vehicles and hybrid vehicles in terms of their relative performance from a severity standpoint is beyond the scope of this thesis.

The rest of this thesis is organized as follows: a literature review is presented followed by a brief description of the data we use in this thesis. Then, the modeling methodology and detailed discussion about model estimation and their implications for crash severity distribution analysis is presented. Finally, the thesis ends with conclusions and directions for future research.

HETEROGENEITY IN HYBRID INVOLVED CRASH SEVERITIES: AN EXPLORATORY ANALYSIS USING THE HIERARCHICAL MIXED LOGIT MODELWITH HIERARCHICAL HYBRID VEHICLE ATTRIBUTES 

Sharing is caring!