Content uploaded by Luc LR Int Panis
Author content
All content in this area was uploaded by Luc LR Int Panis on Apr 27, 2020
Content may be subject to copyright.
1
Estimating PM-emission reductions from speed
management policies
Luc Int Panis1*, Carolien Beckx1, 2, Ina De Vlieger1, Liesbeth Schrooten1
1 Integrated Environmental Studies, Flemish Institute for Technological Research (VITO)
2 IMOB Transportation Research Institute, Hasselt University
Wetenschapspark 5/6, B-3590 Diepenbeek, Belgium
*corresponding author: VITO IMS, Boeretang 200, B-2400 Belgium, luc.intpanis@vito.be
Fax +32 14 32 11 85 Tel +32 14 33 58 87
Abstract
Speed reduction measures have become an increasingly popular way to increase
traffic safety especially in urban areas. Recently many cities have converted
entire districts into 30 km/h zones. In many European countries the maximum
speed of haulage trucks is under discussion or review sometimes in combination
with a ban on overtaking. Reducing the maximum speed is perceived and
promoted by policy makers as beneficial to the environment because of reduced
fuel consumption and lower emissions.
These claims however have not been scientifically validated. They stem from the
popular believe that the widely used Copert-approach, which is scientifically valid
for average trip speeds, can be used to assess the environmental impact of
speed management policies at a local scale. It is obvious that speed reductions
in urban areas or on highways may have very different effects on PM emissions.
On the other hand the simplistic idea that speed reductions increase urban
emissions and decrease emissions on highways is probably wrong. Although few
experts would make this assumption explicitly, it is very frequently made implicitly
by the way that traffic and emission models are integrated. Integrating
macroscopic traffic models with emission functions based on average speed only
is clearly unsatisfactory. In addition, the lack of such functions for the PM
emissions of petrol fuelled cars is an important problem even with advanced
models such as VeTESS.
In this paper we study the problem of accurately estimating the effects of speed
managements policies on exhaust emissions of PM. Emissions for specific types
of vehicles were calculated with the microscopic VeTESS-tool using real-life
driving cycles and compared with results obtained using Copert-like
methodologies.
Our results indicate that emissions of most pollutants should not be expected to
rise or fall dramatically. Nevertheless the conclusion for emissions of PM could
be different. The effects of specific speed reduction schemes on PM emissions
from trucks are ambiguous, but VeTESS results indicate that the PM exhaust
from diesel passenger cars shows a significant decrease in urban areas
converted to 30 km/h zones. Exposure of residents to one of the most toxic
components of the urban air pollution mixture may therefore also decrease.
2
Unfortunately linking microscopic emission and traffic models raises other
concerns such as a lack of validation of the most prominent parameters:
acceleration and gear change behaviour.
Keywords
Speed management policy, PM-emissions, Traffic, Modelling
Introduction
Since September 1st 2005 zone 30 stretches are mandatory near all Belgian
schools, with some exceptions made for schools on the busiest regional roads.
The conversion of entire districts, streets or street sections into 30 km/h zones is
usually done in residential areas where the previous speed limit was 50 km/h
(e.g. city of Ghent; Int Panis et al, 2005). These measures, mainly aimed at
increasing traffic safety, are usually seen or even promoted by local authorities
as beneficial to the environment because of reduced fuel consumption and
emissions. The claims for these environmental benefits stem from the believe
that speed reduction measures in urban areas have similar benefits as those on
highways (Int Panis et al., 2006). However, in contrast to this popular believe,
wide spread emission estimation methods using quadratic functions such as the
Copert/MEET approach would lead us to believe that emissions may even rise
dramatically. Unfortunately the speeds typical for urban traffic (esp. congested
traffic) are very close to or lower than what is usually considered to be the
minimum average trip speed for which relevant estimates can still be made using
the Copert/MEET approach.
In June 2005 the Flemish transport Minister proposed to lower the maximum
speed for trucks on highways from 90 to 80 km/h. This resulted in an enormous
wave of critique from various stakeholders. Reference was made to time losses,
economic losses and serious doubts were cast over the assumed environmental
and safety benefits. Unfortunately most of the discussion was on the basis of
ideology and prejudice. Scientific analysis was either ignored or was unavailable
for use in the discussion at that time (De Vlieger et. al, 2005).
We suggest that more sophisticated methods are needed to estimate the impact
of speed management policies on the emissions of PM.
Methods
Description of the VeTESS model
VeTESS (Vehicle Transient Emissions Simulation Software) was developed
within the European project Decade as a vehicle level tool for the simulation of
fuel consumption and emissions for real traffic transient vehicle operation
(Pelkmans et al, 2004). It is specifically designed to calculate dynamic emissions,
and thereby reaching higher accuracy than traditional emission simulation
models including those using steady state engine maps. We used this model to
calculate emissions and fuel consumption on a second-by-second basis for
specific vehicles on a given speed profile. The calculations in this vehicle
simulation tool are based on a detailed calculation of the engine power required
3
to drive a given vehicle over any particular route. This includes the rapidly
changing (transient) demands placed on the engine.
Description of the driving cycles
Driving cycles were recorded during on-the-road emission measurements in the
cities of Mol (32 474 inhabitants, Belgium) and Barcelona (4.2 million inhabitants,
Spain), using three different vehicles: VW Polo (Euro 4, petrol), Skoda Octavia
(Euro 3, diesel) and a Citroen Jumper (Euro 3, diesel) light commercial vehicle.
We believe these vehicles are representative for an important fraction of current
car sales in Belgium. We refer to Pelkmans et al. (2004) for a detailed technical
description of the vehicles and set-up of the test cycles.
From each of the 6 different driving cycles we derived a modified version in which
the top speed was limited to 30 km/h without changing the acceleration or
deceleration. The length of time driven at the new top speed was elongated
where appropriate to preserve the original cycle distance. Figure 1 shows an
example comparison between one of the original driving cycles and the derived
cycle.
Table 1 shows a summary of statistics describing the cycles and the
modifications that were made. It is clear from the average speeds and the
number of stops that these cycles represent urban trips in heavy traffic.
Table 1 Summarized descriptive statistics for the urban driving cycles used in this study
(Cycles 1-3: Barcelona, Cycles 4-6: Mol). Data for modified cycles in last two columns.
Cycle
N°
Length Stops Max a Max -a Avg v Additional
length
New
avg v
(s) (km) m/s.s -m/s.s km/h (s) km/h
1 1615 6.6 22 7.8 10.2 14.8 107 13.9
2 1765 7.1 27 7.8 10.5 14.5 72 13.9
3 1475 7.3 22 9.4 15.4 17.8 173 15.9
4 1497 10.5 16 8.3 11.3 25.2 163 22.7
5 2003 10.5 22 6.7 8.8 18.9 68 18.3
6 1735 10.5 22 9 11.3 21.8 125 20.3
4
0
10
20
30
40
50
60
1 600 1199 1798
Time (seconds)
Speed (km/h)
Original cycle Derived cycle
Derived cycle
Figure 1: Example showing the conversion of cycle 6 to a cycle with limited top speed
Results
Zone 30 km/h introduction in urban areas
The emissions of each of the three vehicles were modeled with each of the 6
available urban driving cycles, resulting in 18 emission estimates for a reduction
of the top speed from 50 km/h to 30 km/h. Overall results are summarized in
Figure 2. Positive values indicate that emissions go up when the new speed limit
is implemented. Negative values indicate that pollutant emissions decrease.
Results for CO and HC differ widely between vehicles and cycles. Because
emissions of these pollutants are very low in modern cars, we believe that they
are not modeled with sufficient accuracy to lend credibility to the relative changes
shown in the graph. (Even a 100% increase represents only a tiny amount of
pollutants emitted, close to the smallest amount that can be measured;
Pelkmans, pers. comm., 2005.) For the emissions of CO2 and hence fuel
consumption it was found that the change to the driving cycle only had a limited
impact, either positive or negative, on the emission. Emissions decreased for
both cars, but increased for the LGV. For the emissions of NOx the LGV mostly
showed a small increase whereas the results for the cars indicate moderate to
important decreases of the emission. Both diesel vehicles (Octavia and Jumper)
showed a moderate or large decrease in the modeled emissions of PM in each of
the cycles. No PM emissions can be modeled with VeTESS for petrol fueled
vehicles (i.e the VW Polo).
5
-100%
-50%
0%
50%
100%
150%
200%
CO2 (g/km) CO (g/km) NOx (g/km) HC (g/km) PM (g/km)
Figure 2 Estimated relative change in emission for 5 pollutants. Average and range for 18
estimates.
In Figure 3 we present the detailed results for the Skoda Octavia for one
representative cycle in each city. The emissions estimates were made with the
relevant MEET functions based on average trip speed and with VeTESS on the
full speed profile respectively. Results for most other vehicle/cycle combinations
yield similar results. Not surprisingly, the MEET methodology results in a slightly
higher estimate for the emissions. The small difference can be attributed to the
fact that although the derived driving cycle may seem quite extreme (e.g. Figure
1) the resulting change in average speed is quite limited (Table 1). The results
from the VeTESS model runs are less straightforward to interpret or explain
because a large number of factors contribute and interact. Nevertheless it is clear
that emissions of CO2, NOx and PM decrease in each situation for this specific
vehicle. This is the combined result of lower top speeds, longer driving periods at
30 km/h and extended driving to reach the end of the cycle (i.e additional length
in Table 1). Emissions of CO2 are marginally smaller and NOx emission factors
are also lower. The largest reduction however is found for emissions of PM which
decrease in most cases by approximately one third.
In Figure 4 we present some detailed results for the light delivery van. In this
case the result of detailed emission modeling agrees well with the simpler MEET
calculation for CO2 emissions. Both fuel consumption and CO2 emissions are
projected to increase slightly (~3-5%). Results for NOx emissions are mixed
because the small increase evident from the MEET functions is not reproduced
by VeTESS which indicates insignificant changes. For the PM emissions, this
vehicle would show an important decrease (although smaller than for the
passenger cars) under the speed-limited driving cycle.
6
-40%
-30%
-20%
-10%
0%
10%
Mol Barcelona Mol Barcelona
Meet Vetess
CO2 (g/km)
NOx (g/km)
PM (g/km)
Figure 3 Relative change between two normal urban drive cycles (up to 50km/h) and drive
cycles limited at 30 km/h (Skoda Octavia; Cycle 4: 25.2->22.7 km/h in Mol, Cycle 1: 14.8-
>13.9 km/h in Barcelona
-15%
-10%
-5%
0%
5%
10%
Mol Barcelona Mol Barcelona
Meet Vetess
CO2 (g/km)
NOx (g/km)
PM (g/km)
Figure 4 Relative change in emissions between two urban driving cycles and a derived
cycle limited at 30 km/h (Citroen Jumper Van; Cycle 5: 18.9 -> 18.3 km/h in Mol, Cycle 2:
14.5 ->13.9 km/h in Barcelona)
7
80 km/h speed limit for haulage trucks on highways
In this section we discuss the relative emissions when speed limits for lorries are
decreased from 90 to 80 km/h. The total CO2 emission is projected to decrease
by 5 to 10 % and this trend is consistent for all trucks studied (Table 2).
We compare these results with the estimates from TEMAT (the Belgian standard,
Copert-based emission model). TEMAT confirms the results of the detailed
vehicle based modelling (VeTESS) which makes the results more credible.
In absolute numbers the CO2 emission factors would on average drop by
approximately 100 g/km if the policy resulted in a decrease from 90 to 80 km/h.
Using more realistic estimates of the impact of the policy on real traffic speeds
yields a reduction of only 50 to 70 grams/km.
For exhaust emissions of PM, TEMAT predicts an increase in the fleet average
emissions with 3-4%. The detailed results for PM are however very confusing
(Figure 5). PM emission factors decrease for the 3.5-7.5 and 16-32 tonnes weight
classes and increase for the 7.5-16 and 32-40 tonnes weight classes. All
changes (increases and decreases) become smaller in the future. Because of the
dominance of the largest trucks the fleet average emission factor also increases.
Table 2. Relative emissions for different speed reductions (VeTESS results)
CO2 NOX PM
Scenario: 90 km/h to 80 km/h
IVECO Eurocargo 7500 kg 84% 71% 84%
IVECO Eurocargo 12,000 kg 86% 72% 100%
MAN 30,000 kg 91% 89% 103%
Scania 30,000 kg 90% 85% n.a.
Scenario 2: 100 km h-1→ 90 km h-1
IVECO Eurocargo 7500 kg 73% 85% 71%
IVECO Eurocargo 12,000 kg 80% 88% 67%
8
-0,015
-0,01
-0,005
0
0,005
0,01
0,015
0,02
2005 2010 2020 2005 2010 2020
Theoretical Real
g PM/km
32-40 tonne 16-32 tonne 7,5-16 tonne 3,5t-7,5 tonne Fleet average
Figure 5 Absolute difference in the PM fleet emission factors for the 3.5-32
tonne trucks, 90 km h-1 compared to 80 km h-1 (TEMAT)
A high R2 was reported in MEET for this emission function, indicating it was
based on a small sample. The lack of consistency between the effects for the
different classes indicates a large amount of uncertainty.
Detailed modelling of the engine and vehicle characteristics with VeTESS
provides us with an additional set of results. Emissions of NOx are expected to
decrease for the selected vehicles while only one vehicle showed a very small
increase in PM emissions.
Trucks below 12 tonnes are currently the only class that can drive faster than 90
km/h. Policy options for these vehicles include the installation of mandatory
speed limiters like those used on all heavier trucks at this moment or reducing
their speed even further to 80 km/h similar to the proposed policy for heavier
trucks. In the results presented in Table 3 we see that there is a difference
between both weight classes. The PM emissions for the smallest category (3.5 -
7.5 tonnes) would decrease by 3 to 9 % but trucks in the 7.5-16 tonnes class
could see their PM emissions increase by up to 5%.
Table 3 Relative PM-emissions for lighter truck categories (TEMAT results)
Maximum speeds Real speeds
Scenario 100 to 80 km/h 2005 2010 2020 2005 2010 2020
3.5-7.5 tonnes 91% 91% 91% 92% 93% 93%
7.5-16 tonnes 103% 103% 102% 104% 104% 105%
9
Discussion
The emission modeling and results presented in this paper demonstrate that
estimating emissions, even of classical pollutants, from vehicles is a complex
endeavor. Estimating the impact of policies on emissions proves to be even more
difficult (e.g. Cornelis et al, 2005, Int Panis et al, 2005). In the case of a severe
decrease of the urban speed limit, neither the naïve assumption that emissions
will decrease nor the straightforward (but methodologically unjustified) application
of the MEET methodology seem to be correct. We have tried to shed some light
on this problem by applying a very detailed model that can take changes in the
speed pattern into account because it models the entire drive train including
transient effects in the engine. The obvious disadvantage is that the necessary
engine and vehicle data for the model is only available for a limited number of
vehicles and it is not feasible to apply this model to look for changes in emissions
at the macroscopic emission inventory level. Nevertheless the detailed analysis
of the behaviour of these vehicles emissions’ is relevant for two reasons. First the
available data used for this study are from quite popular vehicles that represent
analogues models from other brands as well as other cars with similar engines.
Secondly the engines and after treatment technology of these modern cars is a
fair proxy to what may become the average fleet in the near future. This is clearly
more relevant to the study of policies than the ability to accurately model older
model years.
This being said, there are some important aspects which we have not taken into
account and that could potentially invalidate our results and conclusions. Firstly
we have not made any changes to the acceleration and deceleration of the
driving cycles. This is an implicit assumption that needs to be validated because
changes in driving style (e.g. between individual drivers) have a major impact on
emissions (De Vlieger et al, 2000). It is generally assumed that reducing the
speed limit will also lead to a less dynamic driving style and a more fluent traffic
flow. On the other hand it is not unlikely that very low speed limits such as those
discussed here irritate people who then try to make up for the time lost by
accelerating faster (although they may also simply not obey the speed limit). In
cases where the speed limit is imposed by a device in the cars (e.g. ISA) is was
shown that some drivers accelerated faster up to the speed limit (Vlassenroot et
al., 2006). Unfortunately we cannot take this into account in this study because
detailed (i.e. measured) data are currently lacking. A large scale monitoring
programme will start later in 2006 (Broekx, pers. comm., 2006). Theoretically this
problem can be circumvented by using microscopic traffic simulation models that
generate instantaneous speed estimates (and hence also acceleration) for
individual vehicles. Unfortunately detailed as the models may seem at first glance
the acceleration estimates are largely based on very rough estimates of vehicle
performance and driver behaviour. More importantly, the results of such model
studies are, if conducted properly, validated against counted vehicle flows and
measured speeds. The results for acceleration however are never validated. It
would therefore be questionable to use them as a basis for any emission
estimates (Joumard, pers. comm., 2005). In addition several authors have found
10
it very difficult or impossible to include acceleration (as the most straightforward
variable that describes dynamics) as an input variable for Copert-like emission
functions. For example the results of multiple non-linear regression techniques
(e.g. Int Panis et al., 2006) are rather disappointing. From this point of view the
methods used in this study are certainly justified.
Other types of models may be used to study some other consequences of the
zone 30 introduction such as the avoidance of the area by transit traffic or the
shift to slow modes by local residents (e.g. the class of “Activity-Based” models;
e.g. Beckx et al, 2005, 2006a). But these considerations are far beyond the
scope of the study presented in this paper.
One of the most conspicuous differences in driving behaviour, because it is not a
continuous variable, is gear shifting behaviour. The decision to shift up or down
depends on a combination of technical factors specific to the vehicle (gear ratio,
torque, …) and personal preferences. In this study we have used the default
values for each car provided within the VeTESS-tool. It is however possible that
imposing a speed limit influences the gear shifting behaviour. Unfortunately,
again no data are available to account for this. In a follow-up study different gear
shifting strategies (e.g. gentle, aggressive,…) will be used in a sensitivity
analysis. In cases where the speed limit is very close to the point where most
people shift e.g. between second and third gear, this may have a significant
effect on the emissions (Beckx et al, 2006b).
The finding that PM emissions of trucks could increase following speed
reductions on motorways is consistent with the results from other studies
(HBEFA 2004, IEA 2005). The COST 346 working group decided that a further
speed reduction (below 80km/h) would not improve fuel consumption but would
increase PM emissions. The large uncertainties are blamed on the fleet
composition and on difficulties to derive a typical driving pattern. The choice of
gear features among the most prominent changes to the driving pattern and is
likely to be influenced by changes to the speed limit. Although this may be more
important in urban locations than on highways. The VeTESS model was
therefore used to study the effect of different gear shifting strategies in
connection with different speed limits. Our conclusions were confirmed for any
gear shifting strategy for any speed reduction down to 80 km/h although there
seems to be some variation in the magnitude of the effect. Further speed
reductions below 80 km/h resulted in much higher emissions for some (but not
all) trucks although the fuel consumption remained fairly stable.
Although the measures discussed are deemed important to reduce accidents, it
is unlikely that they will have a significant effect on emissions at a regional or
national level. The number of vehicle kilometers affected is likely to be very small.
In addition the sign of the changes for most pollutants is not clear from our
modeling. Nevertheless it cannot be ruled out that the effect on exposure to PM
is important. Our (very limited) set of estimates point to a consistent decrease in
PM emissions. Increased concentrations of PM and especially those emitted by
11
road transport have often been blamed to be the cause of adverse health effects.
Some epidemiological studies have found significant relationships between
health effects and peoples proximity to important sources of traffic related air
pollution. There is a growing consensus that the link between health effects and
PM concentrations may be causal. Any decrease of PM emissions, concentrated
in urban areas with poor mixing (e.g. street canyons) and high densities of people
should therefore be considered an important benefit.
Finally we would like to draw the readers attention to the fact that this paper only
refers to exhaust emissions of PM. PM emissions from the wearing of tyres,
brakes and road surfaces or the re-suspension of road dust were not considered.
It is likely that speed is a factor that influences these emissions, but today no
functions exist that are accurate enough to complement the exhaust modeling
discussed in this paper.
Conclusions
It is unlikely that imposing strict speed limits in urban areas has a significant
influence on emissions of NOx or CO2. Concerning the impact on emissions of
PM VeTESS results indicate that the exhaust from the diesel vehicles may show
a significant decrease, whereas MEET functions assume a moderate increase.
The effect on emissions of PM should be confirmed by further research, also
focusing on the impact of acceleration or gear shifting behaviour.
All results for trucks consistently indicate that a lower maximum speed on
motorways result in lower emissions of CO2. Results PM are not consistent and
uncertain but probably too small to offset the clear benefits of the CO2 reduction.
Acknowledgement
The authors wish to thank Luc Pelkmans for making the VeTESS model and the
driving cycles available for this study.
References
Beckx C., Broekx S. and Janssens D., 2005. Activity-based policies to reduce
human exposure to traffic air pollution. In proceedings of the 32nd international
“Transportation Research Colloquium”, Antwerp, Belgium, November 24-25,
2005.
Beckx C., Int Panis L., Vanhulsel M., Wets G. and Torfs R. 2006a. Gender-linked
disparity in vehicle exhaust emissions? Results from an activity-based survey. In:
Proceedings of the 8th International Symposium on Highways and the Urban
Environment. Rauch S. & Morrisson G. (eds.), Chalmers University. (in press)
Beckx C., Int Panis L., Debal P., Wets G. 2006b. Influence of gear changing
behaviour on fuel-use and vehicular exhaust emissions. In: Proceedings of the
12
8th International Symposium on Highways and the Urban Environment. Rauch S.
& Morrisson G. (eds.), Chalmers University. (in press)
De Vlieger I., De Keukeleere D. and Kretzschmar J., 2000. Environmental effects
of driving behaviour and congestion related to passenger cars. Atmospheric
Environment, Vol. 34, ppp. 4649-4655.
De Vlieger I., Schrooten L., Pelkmans L., Int Panis L., 2005. 80 km/h maatregel
voor vrachtwagens - Wetenschappelijke screening van het effect op de uitstoot
van CO2 en schadelijke emissies. 2005/IMS/R/252. Confidential report for the
Flemish minister of transport (in Dutch). Available on-line at:
http://www.mobielvlaanderen.be/pdf/persberichten/80-studie03.pdf
HBEFA (2004) Infras Handbook emission factors for road transport. Version 2.1,
Vienna 2004.
IEA (2005) Saving oil in a hurry, OECD/IEA, ISBN 92-64-109414.
Int Panis L., Cosemans G., Torfs R., Liekens I., De Nocker L., Broekx S. and
Beckx C., 2005. Mobilee: effect of a local traffic plan on exposure to local air
pollution. VKM-THD Mitteilungen, Heft/Volume 85/II, pp.13-21.
Int Panis, L., De Vlieger, I., Pelkmans, L., Schrooten, L. 2006. Effect of speed
reduction on fuel consumption and emissions of heavy duty lorries in Belgium. In:
Proceedings of the 8th International Symposium on Highways and the Urban
Environment. Rauch S. & Morrisson G. (eds.), Chalmers University. (in press)
Int Panis, L., Beckx, C., Broekx, S., 2006. Impact of zone 30 introduction on
vehicle exhaust emissions in urban areas. Proceedings of the European
Transport Conference, Straatsburg. (in press).
Int Panis, L., Broekx, S., Liu, R., 2006. Modelling instanteneous traffic emission
and the influence of traffic speed limits. Science of the total env., Vol 372, N°1-3,
270-285.
Pelkmans L., Debal P., Hood T., Hauser G. and Delgado M., 2004. Development
of a simulation tool to calculate fuel consumption and emissions of vehicles
operating in dynamic conditions. SAE.
Vlassenroot S., Broekx S., De Mol J., Int Panis L., Brijs T., Wets G., 2007.
Driving with intelligent speed adaptation: Final results of the Belgian ISA-trial,
Transportation Research Part A, Vol 41, N° 3, 267-279
