An overview of experimental results and dispersion modelling of nanoparticles in the wake of moving vehicles.

Abstract

Understanding the transformation of nanoparticles emitted from vehicles is essential for developing appropriate methods for treating fine scale particle dynamics in dispersion models. This article provides an overview of significant research work relevant to modelling the dispersion of pollutants, especially nanoparticles, in the wake of vehicles. Literature on vehicle wakes and nanoparticle dispersion is reviewed, taking into account field measurements, wind tunnel experiments and mathematical approaches. Field measurements and modelling studies highlighted the very short time scales associated with nanoparticle transformations in the first stages after the emission. These transformations strongly interact with the flow and turbulence fields immediately behind the vehicle, hence the need of characterising in detail the mixing processes in the vehicle wake. Very few studies have analysed this interaction and more research is needed to build a basis for model development. A possible approach is proposed and areas of further investigation identified.

Full text

Page 1

1
An overview of experimental results and dispersion modelling of
nanoparticles in the wake of moving vehicles
Matteo Carpentieri, Prashant Kumar* and Alan Robins
Faculty of Engineering and Physical Sciences, University of Surrey, Guildford GU2
7XH, UK
Capsule
The transformation of nanoparticles and the flow characteristics in both the near and
far wake regions must be understood in order to develop mathematical models.
Abstract
Understanding the transformation of nanoparticles emitted from vehicles is essential
for developing appropriate methods for treating fine scale particle dynamics in
dispersion models. This article provides an overview of significant research work
relevant to modelling the dispersion of pollutants, especially nanoparticles, in the
wake of vehicles. Literature on vehicle wakes and nanoparticle dispersion is reviewed,
taking into account field measurements, wind tunnel experiments and mathematical
approaches.
Field measurements and modelling studies highlighted the very short time scales
associated with nanoparticle transformations in the first stages after the emission.
These transformations strongly interact with the flow and turbulence fields
immediately behind the vehicle, hence the need of characterising in detail the mixing
processes in the vehicle wake. Very few studies have analysed this interaction and
more research is needed to build a basis for model development. A possible approach
is proposed and areas of further investigation identified.
Key words: Dispersion model; Nanoparticles number concentration; Particle size
distribution; Street canyons; Near and far wake; Vehicle exhaust emission

*Corresponding author. Division of Civil, Chemical and Environmental Engineering,
University of Surrey, Guildford GU2 7XH, UK; Tel.: +44 1483 682762; Fax: +44 1483
682135. Email addresses: P.Kumar@surrey.ac.uk; Prashant.Kumar@cantab.net

Page 2

2
1.
Introduction
Atmospheric nanoparticles have attracted substantial attention from the
scientific community due to their negative impacts on human health (Donaldson et al.,
2005), urban visibility (Jacobson, 2005) and global climate change (IPCC 2007).
Recent toxicological and epidemiological evidence indicates that number
concentration is one of the most important metrics to assess effects on human health
(ICRP, 1994; Oberdorster, 2000; Davidson et al., 2005; Donaldson et al., 2005). Road
vehicles are the dominant source of nanoparticles, contributing up to ~86% of total
particle number concentrations in the urban environment (Johansson et al. 2007; Pey
et al. 2009). More than 99% of the particles, by number, in the atmospheric urban
environment are in the <300 nm size range (Kumar et al., 2008a–d; Kumar et al.,
2009a). Exhaust emissions from vehicles can increase ambient number concentrations
of nanoparticles by two orders of magnitude or more (104–106 # cm–3) relative to the
background level (103–104 # cm–3) (Kumar et al., 2008c, d; Kumar et al., 2009b). It is,
therefore, important to examine emissions from individual vehicles under real driving
and dilution conditions. However, the situation becomes complex when fine spatial
scale studies of nanoparticle dispersion are contemplated (e.g. in a vehicle wake)
because of the limited time response of most available instrumentation. Our
preliminary results (Kumar et al., 2009c) showed that the particle number distribution
and concentration of nanoparticles changes rapidly in the wake of a moving vehicle
due to the influence of a number of transformation processes (i.e. coagulation,
condensation, deposition and nucleation). This is because, as the exhaust dilutes and
cools, volatile precursors present in sufficient concentrations may become sufficiently
supersaturated to nucleate, grow and undergo gas–to–particle conversion (Kittelson et
al., 2006a). This is the key difference between modelling dispersion for standard

Page 3

3
gaseous pollutants and nanoparticles. The very short time scales of this evolution (less
than a few seconds) call for a more accurate description of the mixing process in the
vehicle wake than available today.
A number of methods exist for modelling the dispersion of passive pollutants away
from a roadway, though they very rarely can treat vehicle wakes or include any
adjustment due to this effect; for example, CALINE4 (Benson, 1989), CPBM
(Yamartino and Wiegand, 1986), ADMS-Urban (CERC, 2006) and OSPM
(Berkowicz, 2000). Most of these models treat the emission as a continuous line
source, either without any adjustment or by a simple enhancement of turbulence levels
(Baker, 2001), despite the significant effect on the dispersion of pollutants that vehicle
wakes have. There have been a modest number of studies concerning the aerodynamic
behaviour of ground vehicles and the effect of ground structures on drag (Eskridge et
al., 1979; Eskridge and Hunt, 1979; Baker, 2001). However, little attention has been
paid to the wake properties that are connected with the dispersion of nanoparticles. On
the other hand, a number of studies have been dedicated to measurements of
nanoparticles along the roadside and in street canyon (Charron and Harrison, 2003;
Gidhagen et al., 2004; Brugge et al., 2007; Kumar et al., 2008a; Kumar et al., 2008d;
Kumar et al., 2009b). However, the processes of transport and transformation that
relate these observations to vehicle tailpipe emissions remain largely unknown.
The objectives of this study are to synthesise significant past research work relevant to
nanoparticle dispersion modelling in the vehicle wakes, together with highlighting
associated research gaps. Section 2 presents a critical discussion of the characteristics
of vehicle wakes and their mixing properties. Section 3 gives an introduction to the
transformation processes of nanoparticles emitted by vehicles. The most relevant

Page 4

4
literature on this topic is then reviewed in the following sections, organised in three
distinct categories: field measurements (Section 4), wind tunnel experiments (Section
5) and mathematical modelling (Section 6). Section 7 summarises the previous
sections, identifies future research needs and sets up a research strategy for
developing mathematical nanoparticle dispersion models and forming appropriate
assumptions on the role of fine scale particle dynamics in existing dispersion models.
Such models would assist regulatory authorities in designing sensible future actions to
regulate nanoparticle emissions.
2.
Flow and mixing in the wake of moving vehicles
A detailed knowledge of flow and mixing processes in vehicle wakes is
necessary in order to cope with the very short time scales of nanoparticle evolution
processes; an adequate treatment of such processes is essential for developing reliable
mathematical models. In this study, we are particularly interested in the case of a
moving vehicle with little or no cross–wind. When a vehicle moves forward at
velocity V through the undisturbed air, the flow separates from the bluff rear end in
the wake. The wake as a whole may generally be divided into two distinct regions: the
near wake and the main or far wake (Fig. 1). The near wake is the region of separated
flow close behind the body. For a road vehicle, it is characterised by both a region of
recirculation and the formation of a pair of streamwise longitudinal vortices (Hucho,
1987). The far wake is an area of general turbulence downstream of the near wake,
which has little discernible flow structure (Hucho, 1987).
2.1 Near-wake
The near-wake of a car consists of two components: a large scale recirculation
region immediately behind the vehicle and a system of longitudinal trailing vortices

Page 5

5
with unsteady fluctuations caused by a variety of effects such as the instability of the
separated shear layer and wake pumping (Ahmed, 1981; Hucho, 1987; Baker, 2001).
The initial dispersion of the exhaust plume in the near wake becomes important at the
local scale, when receptors (e.g., people) are at close distance (in the same street) as
the emission source. However, the level by which this will affect the local scale
dispersion depends by the local conditions. It is likely that in very dense traffic
conditions and deep street canyons the near wake may be neglected, but more studies
are needed to support this hypothesis.
Studies related to the characterisation of dispersion behaviour in the near–wake are
much rarer than those related to the far wake zone (see Section 2.2). Baker (1996)
describes a model based on the assumption that the pollutant emitted by the vehicle is
spread uniformly in the near wake (that is the plume is coincident with the vehicle
wake), using a Gaussian puff approach to calculate the concentrations further
downwind. This approach might be acceptable for passive gaseous pollutants, but
nanoparticles experience transformations on very short time scales (see Section 3) and
a more detailed characterisation of the near wake may therefore be necessary. More
recently, computational fluid dynamics (CFD) have been applied to the near-wake
dispersion of pollutants (Richards, 2002; Dong and Chan, 2006). The simulations,
however, are limited by the lack of experimental studies specifically aimed at deriving
suitable boundary conditions for the numerical calculations.
2.2 Far wake
The most well known and documented vehicle wake theory is that of Eskridge
and Hunt (1979). Based on the perturbation analysis of the equations of motion the
theory describes the velocity field far downstream of a single vehicle moving through

Page 6

6
still air. Using the assumptions of constant vehicle velocity, flat terrain and no wind,
expressions were developed for the velocity deficit far downwind of a vehicle i.e. the
far wake. A more recent study by Hider et al. (1997) saw the derivation of the same
expression using a different method. They also derived expressions for the lateral and
vertical velocity components.
Dispersion in the far wake is usually treated as a standard Gaussian plume (Baker,
1996; 2001; Richards, 2002). Very few models take into account the effect of the
vehicle wake in the dispersion process (see Section 6).
The influence of cross-winds has also been studied. The primary effect, either in open
terrain or within a street canyon, it is to translate the wake in the wind direction
(Eskridge and Hunt, 1979; Yamartino and Wiegand, 1986; Hider et al., 1997; Baker,
2001). However, there are other effects due to atmospheric turbulence that are
strongly dependent on the surrounding topography (open terrain or street canyon, for
example; see Baker, 2001).
Table 1 summarise some of the key flow and mixing characteristics to be considered
for developing a mathematical model for the vehicle wake (near and far wakes).
Clearly, further research is needed for the adequate characterisation of vehicle wakes
in dispersion models. In particular, almost no information can be found on the mixing
process in the near-wake, though this is the key region where the nanoparticle main
evolution processes occur (see Section 3).
3.
Short range dispersion of nanoparticles in the urban environment
The behaviour of particles in the atmosphere and within the human respiratory
system is determined largely, but not wholly, by their physical properties, which have

Page 7

7
a strong dependence upon particle size (AQEG, 2005). The smallest group of
particles, with diameters of ≤30 nm, are usually referred to as the nucleation mode
(Charron et al., 2008; Kumar et al., 2010a, b, c). Growth of nucleation mode particles,
primarily by vapour condensation but also as a result of coagulation processes, leads
to the formation of accumulation mode particles that are typically between 30 and 300
nm in size (Kumar et al., 2010b). Particles in the accumulation mode can have long
atmospheric lifetimes, typically 7-30 days in the absence of rain, much longer than the
short-lived particles in the nucleation mode (Kumar et al., 2010b).
Nanoparticles emitted by cars are subject to dilution (mixing) processes very similar
to those described in the previous section for passive gaseous emissions. They are,
however, also subject to a number of other processes at various stages in their
evolution (e.g. nucleation, coagulation, condensation and deposition). Ketzel and
Berkowicz (2004) analysed the time scales associated with different transformation
processes. They concluded that dilution is the fastest transformation process at most
concentration levels, followed by condensation. Coagulation and deposition do not
usually play a significant role in defining particle number concentrations near streets.
However, they did not treat the very first stages of dispersion, where nucleation and
condensation are mainly responsible for the formation of nanoparticles. Details of the
interaction between dilution and other transformation processes at short distances
from the emission point were studied by Zhang and Wexler (2004). They found that
the exhausts emitted from different types of engines retained their own characteristics
in the first stage of the dilution process. Sulphuric acid–induced nucleation was found
to be the dominant particle production mechanism, followed by the condensation of
organic compounds, resulting in the rapid growth of nuclei mode particles and
relatively slow growth of accumulation mode particles. This study, however, did not

Page 8

8
include any effect due to the vehicle wake, which, as described in Section 2, strongly
modifies the flow and turbulence fields near the emission source. A summary of the
significance of different transformation processes is reported in Table 2; the table has
been compiled using information from Ketzel and Berkowicz (2004) and the other
references mentioned in this section.
4.
Field measurements
A number of field measurements of number and size distributions of
nanoparticles in urban areas have been made using instruments such as scanning
mobility particle sizers, electrical low pressure impactors, ultrafine particle
condensation counters, alone or in combination (Shi et al., 1999; Longley et al., 2003;
Wehner and Wiedensohler, 2003; Weber et al., 2006). A comprehensive review
covering this topic can be found in Kumar et al. (2010b). Most of these instruments
have low sampling frequencies relative to that required to characterise nanoparticle
dispersion phenomena in vehicle wakes, hence an instrument with a fast response is
required (Kumar et al., 2009c). The use of one such instrument (a Cambustion
DMS500) in a street canyon and a vehicle wake was recently reported by Kumar et al.
(2008a–d, 2009a–c). Yao et al. (2006) confirm the need for fast response
instrumentation. They measured the concentration of nanoparticles emitted by a
vehicle and applied different averaging times in the range from 1 to 120s and
concluded that time–averaging data from complex environments, such as street
canyons and tunnels, can affect their interpretation and lead to misleading or
meaningless deductions regarding particle evolution.
Field experiments purposely designed for measuring nanoparticle evolution in the
wakes of vehicles are rare and there’s a clear need for further research. Kumar et al.

Page 9

9
(2009c) measured nanoparticle concentrations in the wakes of a moving diesel
vehicle. In their preliminary experiments, a fixed measurement point was placed 20
cm above the ground and used to obtain nanoparticle number concentrations from
plumes dispersed by diesel vehicles passing over the sampling tube. The results from
this particular study are summarised in Fig. 2. An important finding was that the
effects of transformation processes, such as nucleation and condensation, were
generally complete within about 1 s after emission, showing the rapid evolution of
particle number and mass distributions in the near–wake. Because of very limited
observations, these preliminary results are useful for understanding the dynamics of
nanoparticles at very short time scales but not sufficient to enable sensible
mathematical models to be developed. There is still an need for detailed experiments
using the fast response instruments to estimate the evolution time of particle number
and mass distributions for various types of vehicles running at a range of speeds in
different urban settings and meteorological conditions, where background particle
number concentrations differ considerably. Additionally, it would be of considerable
value if detailed investigations of the transformation processes were also carried out
to help address important questions about the role of fine scale particle dynamics.
5.
Wind tunnel experiments
Wind tunnel experiments have been used extensively for determining the flow
characteristics of wake behind vehicles, such as Eskridge and Thompson (1982) and
Hackett et al. (1987), but studies for pollutant dispersion, particularly for
nanoparticles, are relatively rare. To our knowledge, there have not been any wind
tunnel experiments involving nanoparticle measurements so far. Results from such
studies in the controlled environment of the wind tunnel will be able to complement
the data from field studies and, in conjunction with flow and tracer gas measurements,

Page 10

10
greatly contribute to the development of mathematical models. There have been,
however, a number of wind tunnel studies that, even if not directly related to
nanoparticle dispersion, provide a useful basis for the development of future
experiments and mathematical models. Subsection 5.1 describes the extensive work
carried out for deriving parameterisation for traffic produced turbulence. This is
relevant to enhance our understanding of the mixing process in the near wake (see, for
example, Table 1). The following subsection describes the few attempts at
characterising dispersion behaviour in the wake of moving vehicles, in particular in
the near wake. The third subsection discusses improvements in the methodologies and
identifies further experimental work that can help in the development of models for
nanoparticle dispersion models in vehicle wake.
5.1 Traffic produced turbulence
Traffic produced turbulence (TPT) and its effects on the flow and mixing
processes has been the main focus of a number wind tunnel studies. For instance,
Kastner-Klein et al. (2000a, b; 2001a, b) studied different traffic configurations
(one-way and two-way) simulated by small metal plates moving on two belts along
the street in a wind tunnel model. Their main interest was in the interactions between
traffic– and wind–induced flows in a street canyon. The presence of traffic and its
arrangement were also shown to affect the concentration distribution along the
leeward canyon wall (Kastner-Klein et al., 2001b). The results were then used to
derive TPT parameterisations for street canyon models (Kastner-Klein et al., 2003; Di
Sabatino et al., 2003).
Khare et al. (2002) and Ahmad et al. (2002) used a similar approach, consisting of
moving belts carrying model vehicles, in their wind tunnel experiments. The system

Page 11

11
was placed in various types of simulated atmospheric boundary layers, and the effect
of the traffic condition and the wind direction on the vertical spread of the exhaust gas
was examined. Since the main concern was the effect of traffic as a whole on the
general flow and mixing processes in the urban environment, the above mentioned
studies did not analyse in detail the characteristics of the wake behind a single moving
element. None of them studied the near wake region of the vehicles.
5.2 Dispersion of inert tracers in the vehicle wake
For the development of the ROADWAY model, Eskridge and Rao (1986)
determined the optimal turbulence scales in the far wake region of a moving vehicle.
This was determined by measuring concentrations of inert gaseous tracer in the far-
wake (30–60 car-heights behind) of a full scale model of a passenger car in the wind
tunnel. The near-wake was not the focus of their study, though. More relevant to the
near-wake was the study conducted by Clifford et al. (1997). They used three
passenger car models in a row in the wind tunnel, separated by half the car length
from each other and measured the concentration distribution of the tracer on the car
surface including air inlet positions. Results of their study highlighted the strong
influence in the mixing process of the car immediately behind the emitted plume. The
focus of their study, however, was on internal air quality rather than on the dispersion
process in the near-wake.
Richards (2002) and Baker (2001) analysed flow and dispersion characteristics in the
near-wake of a vehicle model with no cross-wind. Results from these wind tunnel
studies highlighted the close relationship between the inert tracer concentration field
and the velocity and turbulence fields. Concentration fluctuations, measured by a
flame ionisation detector (FID), were consistent with the fluctuations in the velocity

Page 12

12
field (obtained through a combination of particle image velocimetry, PIV, hot-wire
anemometry, HWA and flow visualisation methods). The time histories of
concentration had a “peaky”, intermittent nature. Kanda et al. (2006a) measured the
flow characteristics in the near-wake of small-scale models of a car and a lorry
emitting a thermally buoyant plume. They used PIV and laser Doppler anemometry
(LDA) to measure velocity and turbulence fields in the wind tunnel, and a FID to
measure tracer gas concentrations. They found that the buoyancy of the exhaust had
generally a minor effect on the dispersion behaviour. The results from this study were
then used as a basis for studying multi–vehicle configurations (Kanda et al., 2006b),
with an approach similar to the one adopted by Clifford et al. (1997) in the wind
tunnel.
The effect of cross-winds on a single wake generated by a model of lorry was studied
by Baker and Hargreaves (2001). The model was placed on a moving model rig and
was propelled across a wind tunnel in which an atmospheric boundary layer had been
simulated. Experiments were carried out with both open country and with a street
canyon wind simulations.
5.3 Discussion
In order to improve our understanding of the flow and mixing processes in the
near-wake, and also in the far-wake, more studies on the basic configuration of the
wake from a single vehicle are needed. As stated above, most of the previous studies
dealt with more complex situations such as multiple vehicles, cross-winds and
complex geometry (for example, street canyons). They are very useful for deriving
parameterisations of the traffic flow for use in larger scale models, but cannot be used
to characterise the single wake. The few studies specifically designed to study a single

Page 13

13
wake, and the near-wake in particular (Baker, 2001; Richards, 2002; Kanda et al.,
2006a), have highlighted the differences in the plume behaviour depending on the
shape of the model and the boundary conditions, so more experimental studies are
needed to generalise their findings.
A possible improvement on past experimental studies would be the use of a rolling
floor in the wind tunnel. All the above mentioned studies suffered from the presence
of a boundary layer developing close to the ground, which can interfere with a
developing wake flow field and its measurement. Kanda et al. (2006a) tried to prevent
this problem by placing the model on an elevated table; nonetheless, a boundary layer
grows on the table top and reaches about 3 cm thick at about 1 m from the upwind
edge of the table, resulting in unrealistic velocity and turbulence fields (Kanda et al.,
2006a). Baker and Hargreaves (2001) solved the problem by using a ballistic model
within the wind tunnel.
In order to develop parameterised models specifically designed for nanoparticle
dispersion there is first a need to understand sufficiently the dispersion of inert gases.
However, as stated at the beginning of this section, measurements of nanoparticle
number concentrations in the wind tunnel, to be possibly compared with field
measurements and numerical simulations, will add valuable information.. In this case,
a few problems need to be overcome to model particle dispersion properly. In
particular, the presence of particle elements in the emitted plume adds further scaling
conditions to have similarity with the full scale situation. For example, see discussions
on similarity criteria in Kind (1986), Xuan and Robins (1994), Goossens and Offer
(1990) and Parker and Kinnersley (2004). The use of larger scale models is essential

Page 14

14
to meet the similarity criteria, besides providing measurements with better spatial
resolution.
6.
Mathematical modelling
As highlighted in the previous sections, further insight is needed on the
transformation processes and dynamics of nanoparticles to develop reliable
mathematical models for the estimation of their number concentrations in vehicle
wakes. This section provides an overview of approaches for modelling the dispersion
of pollutants in the wakes of moving vehicles, starting with the most recent advances
in fully computational models (CFD approach). These have their limitations due to
their complex nature and their requirement of intensive computational resources. The
main objective of this section is to discuss possible approaches for fast parametric
models that could be used as stand alone or as sub–modules within a multiscale air
quality model for routine air quality assessment and forecasts A simple mathematical
approach, which needs further development, is proposed as an example of such
approaches.
6.1 CFD modelling
With recent advances in computational modelling, it has become possible to
investigate flow and dispersion characteristics in the wakes of vehicles by using a
numerical approach, CFD. However, such methods cannot be used for operational
modelling (e.g., real time calculations, forecasting, long–term scenario analysis for air
quality assessment) because of their resource demands, though but they can be used,
like experiments, to gain insight on the phenomena involved, especially when
integrated with experimental techniques.

Page 15

15
As already discussed in Section 2.1, a number of approaches to CFD modelling of
gaseous pollutants in the wakes of vehicles has been attempted. These include
Reynolds–averaged Navier–Stokes (RANS) simulations (using the k-ε, RNG k-ε, or k-
ε/Chen turbulence models; Richards, 2002; Richards et al., 2000; Kim et al., 2001), or
large eddy simulations (LES; Dong and Chan, 2006; Chan et al., 2008). The latter
approach is more resource–intensive but has a better potential for unsteady
applications such as the dispersion of nanoparticles in vehicle wakes.
The evolution of nanoparticles can be implemented in numerical models by using the
general dynamic equation (GDE; Friedlander, 2000). This equation cannot easily be
solved, and several numerical techniques have been developed for this purpose,
including sectional methods (Garrick et al., 2006) and the method of moments
(McGraw et al., 1998). A recent successful application of such methods was carried
out by Chan et al. (2010); using an LES approach for turbulence, coupled with a
dispersion model based on the direct quadrature method of moments (DQMOM).
Their results offer an interesting analysis of the time scales of the processes involved
in nanoparticle evolution, yielding conclusions similar to those reported in Section 3
(see also Table 2), demonstrating the potential of the numerical approach.
A different numerical technique was adopted by Albriet et al. (2010). They coupled a
CFD RANS code (using the k-ε turbulence model) with a box modelling approach for
aerosol dynamics (in particular they adopted the Modal Aerosol Model, MAM).
Comparisons of the numerical results with measured values were promising, though
the authors pointed out the necessity of including the vehicle wake for a more correct
characterisation. This may prove to be difficult with the RANS approach, while LES
models may be more suitable for this purpose.