different vehicle operation (and different emission
levels) can have the same average speed. Clearly,
all the types of operation associated with a given
average speed are not accounted for by the use of a
single emission factor. This is a particular problem
at low-medium average speeds, for which the
range of possible operational conditions associated with a given average speed is great. In addition, for modern vehicles a large proportion of the
total emission during a trip can be emitted as very
short, sharp peaks, often occurring during gear
changes and periods of high acceleration, and
this detail is lost when only average speed is
considered. The concept of “driving dynamics”
has therefore been developed to enable model
developers to describe vehicle operation using
additional parameters [64]. In qualitative terms,
dynamics can, perhaps, be thought of as the
“aggressiveness” of driving. Quantitatively, the
term refers to the variation in various properties
or statistical descriptors of a driving pattern. One
of the more useful parameters appears to be relative positive acceleration (RPA) [65]. However,
most model users have no straightforward means
of quantifying such parameters.
In “traffic situation” models, emission factors
are referenced to specific traffic situations that are
known by the model user. A typical example is the
Handbook of Emission Factors [66]. The traffic
situations are characterized by the features of the
sections of road concerned (e.g., “motorway with
120 km/h limit,” “main road outside built-up
area”). The driving dynamics are not quantified
by the user, but are defined through a textual
description (e.g., “free flow,” “stop and go”).
However, requiring the user to define the traffic
situation using a textual description of dynamics
may lead to inconsistencies in interpretation.
Multiple-regression models are another comparatively recent development, and the most
prominent example is the Dutch VERSIT+
model [67]. The model contains test data for a
large number of cars over a range of different
driving cycles. Within the model, each driving
cycle is characterized by various descriptive
parameters (e.g., average speed, RPA, number
of stops per km, etc.). A regression model is fitted
to the average emission factors and cycle
parameters for the various driving cycles, giving
the descriptive variables which are the best predictors of emissions (the group of descriptors
being different for each type of vehicle and
each pollutant). The model requires a driving
pattern (speed-time profile) as the input, from
which it calculates the same range of descriptive
variables and estimates emissions based on the
regression results.
In modal models emission factors are allocated to the specific modes of vehicle operation
encountered during a trip. In the simplest type,
vehicle operation is defined in terms of a relatively small number of modes – typically idle,
acceleration, deceleration, and cruise. For each
of the modes the emission rate is fixed, and the
total emission during a trip, or on a section of
road, is calculated by weighting each modal
emission rate by the time spent in the mode
[68, 69]. This approach has usually been used
to determine the impacts of traffic control measures and signal improvements [70].
Some more detailed modal models aim to give
a precise description of vehicle emission behavior
during a series of short time steps (often 1 s).
These are commonly referred to as “instantaneous” models, although several alternative
terms can be found in the literature, including
“microscale,” “continuous,” and “online”
[71]. In the simplest type of instantaneous
model, emission rates are defined for different
combinations of speed and acceleration, usually
in the form of a matrix [72, 73]. Theoretically, the
finer the resolution of the emission factor matrix
the higher the model accuracy, but as the resolution improves the complexity of the calculations
increases. Some models use the product of speed
and acceleration instead of the acceleration alone
[74]. A more flexible approach relies upon emission maps that relate to engine power [75,
76]. This allows all relevant factors to be taken
into account, including rolling resistance, aerodynamic resistance, acceleration, gradient, transmission losses, and the use of auxiliaries.
Instantaneous models should have a number of
advantages. Firstly, emissions can be calculated
for any operational profile, and thus new emission
factors can be generated without the need for
Air Quality, Surface Transportation Impacts on
59
levels) can have the same average speed. Clearly,
all the types of operation associated with a given
average speed are not accounted for by the use of a
single emission factor. This is a particular problem
at low-medium average speeds, for which the
range of possible operational conditions associated with a given average speed is great. In addition, for modern vehicles a large proportion of the
total emission during a trip can be emitted as very
short, sharp peaks, often occurring during gear
changes and periods of high acceleration, and
this detail is lost when only average speed is
considered. The concept of “driving dynamics”
has therefore been developed to enable model
developers to describe vehicle operation using
additional parameters [64]. In qualitative terms,
dynamics can, perhaps, be thought of as the
“aggressiveness” of driving. Quantitatively, the
term refers to the variation in various properties
or statistical descriptors of a driving pattern. One
of the more useful parameters appears to be relative positive acceleration (RPA) [65]. However,
most model users have no straightforward means
of quantifying such parameters.
In “traffic situation” models, emission factors
are referenced to specific traffic situations that are
known by the model user. A typical example is the
Handbook of Emission Factors [66]. The traffic
situations are characterized by the features of the
sections of road concerned (e.g., “motorway with
120 km/h limit,” “main road outside built-up
area”). The driving dynamics are not quantified
by the user, but are defined through a textual
description (e.g., “free flow,” “stop and go”).
However, requiring the user to define the traffic
situation using a textual description of dynamics
may lead to inconsistencies in interpretation.
Multiple-regression models are another comparatively recent development, and the most
prominent example is the Dutch VERSIT+
model [67]. The model contains test data for a
large number of cars over a range of different
driving cycles. Within the model, each driving
cycle is characterized by various descriptive
parameters (e.g., average speed, RPA, number
of stops per km, etc.). A regression model is fitted
to the average emission factors and cycle
parameters for the various driving cycles, giving
the descriptive variables which are the best predictors of emissions (the group of descriptors
being different for each type of vehicle and
each pollutant). The model requires a driving
pattern (speed-time profile) as the input, from
which it calculates the same range of descriptive
variables and estimates emissions based on the
regression results.
In modal models emission factors are allocated to the specific modes of vehicle operation
encountered during a trip. In the simplest type,
vehicle operation is defined in terms of a relatively small number of modes – typically idle,
acceleration, deceleration, and cruise. For each
of the modes the emission rate is fixed, and the
total emission during a trip, or on a section of
road, is calculated by weighting each modal
emission rate by the time spent in the mode
[68, 69]. This approach has usually been used
to determine the impacts of traffic control measures and signal improvements [70].
Some more detailed modal models aim to give
a precise description of vehicle emission behavior
during a series of short time steps (often 1 s).
These are commonly referred to as “instantaneous” models, although several alternative
terms can be found in the literature, including
“microscale,” “continuous,” and “online”
[71]. In the simplest type of instantaneous
model, emission rates are defined for different
combinations of speed and acceleration, usually
in the form of a matrix [72, 73]. Theoretically, the
finer the resolution of the emission factor matrix
the higher the model accuracy, but as the resolution improves the complexity of the calculations
increases. Some models use the product of speed
and acceleration instead of the acceleration alone
[74]. A more flexible approach relies upon emission maps that relate to engine power [75,
76]. This allows all relevant factors to be taken
into account, including rolling resistance, aerodynamic resistance, acceleration, gradient, transmission losses, and the use of auxiliaries.
Instantaneous models should have a number of
advantages. Firstly, emissions can be calculated
for any operational profile, and thus new emission
factors can be generated without the need for
Air Quality, Surface Transportation Impacts on
59