Traffic noise in LCA
- First Online:
- Cite this article as:
- Althaus, HJ., de Haan, P. & Scholz, R.W. Int J Life Cycle Assess (2009) 14: 560. doi:10.1007/s11367-009-0116-2
- 247 Downloads
Background, aim, and scope
According to some recent studies, noise from road transport is estimated to cause human health effects of the same order of magnitude as the sum of all other emissions from the transport life cycle. Thus, ISO 14′040 implies that traffic noise effects should be considered in life cycle assessment (LCA) studies where transports might play an important role. So far, five methods for the inclusion of noise in LCA have been proposed. However, at present, none of them is implemented in any of the major life cycle inventory (LCI) databases and commonly used in LCA studies. The goal of the present paper is to define a requirement profile for a method to include traffic noise in LCA and to assess the compliance of the five existing methods with this profile. It concludes by identifying necessary cornerstones for a model for noise effects of generic road transports that meets all requirements.
Materials and methods
Requirements for a methodological framework for inclusion of traffic noise effects in LCA are derived from an analysis of how transports are included in 66 case studies published in International Journal of Life Cycle Assessment in 2006 and 2007, in the sustainability reports of ten Swiss companies, as well as on the basis of theoretical considerations. Then, the general compliance of the five existing methods for inclusion of noise in LCA with the postulated requirement profile is assessed.
Six general requirements for a methodological framework for inclusion of traffic noise effects in LCA were identified. A method needs to be applicable for (1) both generic and specific transports, (2) different modes of transport, (3) different vehicles within one mode of transport, (4) transports in different geographic contexts, (5) different temporal contexts, and (6) last but not least, the method needs to be compatible with the ISO standards on LCA. One of the reviewed methods is not specific for transports at all and two are only applicable for specific transports. The other two allow generic and specific road transports to be assessed. The methods either deal with road traffic noise only or they compare noise from different sources, ignoring the fact that not only physical sound levels but also the source of sound determines the effect. Three methods only differentiate between vehicle classes (lorries and passenger cars) while one method differentiates between specific vehicles of the same class. Four of the methods consider the geographic context and three of them differentiate between day- and nighttime traffic.
None of the existing methods for traffic noise integration in LCA complies with the proposed requirement profile. They either lack the genericness for a wide application or they lack the specificity needed for differentiations in LCA studies. There is no method available that allows for appropriate inter- or intramodal comparison of traffic noise effects. Thus, the benefit of the existing methods is limited. They can, in the better cases, only demonstrate the relative importance of road or rail traffic noise effects compared to the nonnoise-related effects of transportation.
Currently, none of the major LCI databases includes traffic noise indicators. Thus, noise effects are usually not considered in LCA studies. We introduce a requirement profile for methods that allow the inclusion of noise in LCI. Due to the estimated significance of noise in transport LCA, this inclusion will change the overall results of many LCA studies. None of the existing methods fully complies with the requirement profile. Two of the methods can be modified and extended for inclusion in generic LCI databases. A third model allows for intermodal comparison. From an LCA perspective, all methods include weaknesses and need to be amended in order to make them widely usable.
Recommendations and perspectives
In part 2 of this paper, an in-depth analysis of the promising methods is provided, improvement potential is evaluated, and a new context-sensitive framework for the consistent LCI modeling of noise emissions from road transportation is presented. Appropriate methods for modeling rail and air traffic noise will have to be developed in the future in order to arrive at a methodological framework fully compliant with the requirement profile. Furthermore, future research is needed to identify appropriate methods for impact assessment.
KeywordsAdditional noise emission LCA LCI Traffic noise Transport
1 Background, aim, and scope
The health significance of noise pollution includes various effects: noise-induced hearing impairment; interference with speech communication; disturbance of rest and sleep; psychophysiological, mental health, and performance effects; effects on residential behavior and annoyance; and interference with intended activities (Berglund et al. 1999). Sound pressure levels of 55 and 65 dB(A) are generally regarded as disturbing to sleep and daytime activities, respectively (Berglund et al. 1999). At the end of the 1990s, about 40% of the population of the European Union was exposed to road traffic noise with an equivalent daytime sound pressure level exceeding 55 dB(A), and 20% were exposed to levels exceeding 65 dB(A). At night, more than 30% were exposed to equivalent sound pressure levels exceeding 55 dB(A). When noise from all transportation modes is considered, more than half of all European Union citizens are estimated to live in zones that do not ensure acoustical comfort to residents (Berglund et al. 1999).
2 Materials and methods
The strengths and weaknesses of a given method depend on the scope of the LCA (e.g., a detailed model might be prerequisite for intercomparing different cars, but might hinder an application if, e.g., for a generic situation, the required data are not available). Therefore, an analysis of all 66 case studies (CS) published in International Journal of Life Cycle Assessment in the years 2006 and 2007 and of the environmental reports of ten Swiss companies (details are given in the ESM, Sections 1 and 2) was carried out. Even though LCA studies are also published in other journals then the International Journal of Life Cycle Assessment, the sample can be considered representative for the purposes of the present analysis. From this analysis, coupled with theoretical considerations on LCA and traffic noise, a requirement profile for methods to include traffic noise in LCA is proposed. We then assess the compliance of the five methods for the inclusion of (traffic) noise in LCA published so far (Doka 2003; Guinée 2001; Heijungs et al. 1992; Müller-Wenk 2002, 2004; Nielsen and Laursen 2005; Potting and Hauschild 2003) with this requirement profile.
3 Requirements for traffic noise methods in LCA
3.1 Analysis of transports in case studies
Inclusion and relevance of transports in 66 case studies published in International Journal of Life Cycle Assessment in 2006 and 2007
All case studies
Relevance unknown or n.a.
With transport in foreground
With no transport in foreground
Unclear if transport in foreground
With transport not an issue
Use of data and relevance of transports in 44 case studies including transports in foreground, published in International Journal of Life Cycle Assessment in 2006 and 2007
Case studies with transport in foreground
Relevance unknown or n.a.
Most of the LCA practitioners use generic transport datasets with a reference flow of ton kilometers (or person-kilometers) for road, rail, ship, and air transports. However, in some CS, additional information on, e.g., the fuel consumption, the emission standard, or the utilization factor of the vehicle is used for road traffic. On the other hand, for rail, ship, and air transport, usually only some geographical information on origin and destination seems to be available, but no specific vehicle characteristics. This implies that transport data need to be compatible for different means and modes of transport and that different degrees of genericness would make sense from a practitioner’s point of view.
3.2 Analysis of transports in environmental reports
Ten companies were chosen to analyze how transport is assessed in the environmental reports. Two of the companies do not report on environmental issues at all. Another company reports only on energy demand in production, water use, and waste management but not on transports related to its activities. Four out of the remaining seven companies report transport issues as rather relevant. In two companies, transports are of medium relevance, while in only one company that provides telephone and internet services transports are considered not important. All the reports give information on types and emission standards of the vehicles. One company reports only fuel consumption. Another one reports the fuel consumption and two specific emissions (PM10 and NOx). The other companies report indicators based on life cycle inventories (ecological scarcity points, CO2 equivalents, carbon footprint). Two of them explicitly use ecoinvent v1.x data while the others are not specific on the data source. In one case, the practitioners using the ecoinvent data needed to adapt them to account for environmental improvements achieved by changes in the fleet, since the ecoinvent v1.x database does not differentiate among various emission standards (Euro 1–5) and provides no data for liquefied petroleum gas or biofuel (biogas, biodiesel, bioethanol) fueled vehicles, and hybrid cars. How the company made this adaptation is not documented. Details can be found in the ESM in Section 2.
3.3 Theoretical considerations on requirements for transport data
Compliance with the ISO standards (ISO 14′040 2006; ISO 14′044 2006) is an important quality sign for LCA studies, LCI databases, and LCIA methods and should therefore be aimed at. The standards prescribe four phases in an LCA study: goal and scope definition, inventory analysis, impact assessment, and interpretation. These phases relate to each other but are separate. In the LCI phase, physical in- and outputs are compiled; the life cycle impact assessment (LCIA) phase is aimed at evaluating the significance of potential environmental impacts using the LCI results.
Consequently, LCI results need to carry sufficient information to assess the impacts according to the scope definition. Thus, if the LCIA should assess total annoyance or total human health effects due to the exposure to a noise emission, the minimum information needed in LCI data is the sound level, the time of the day (day vs. night), and the source of the sound (road, rail, or air traffic) (Schuemer et al. 2003). Sound is usually measured in decibels, a logarithmic unit for the sound pressure. The contributions of different frequency ranges to the overall level of a sound are usually weighted to account for the individual hearing thresholds for the frequencies. This is notated by adding the weighting index in brackets to the unit (“dB(A)”). Sound levels at a certain location are a composite of individual sound sources, e.g., single car engines, and therefore fluctuate over time. Ambient sound levels are usually integrated over time and then called “equivalent sound levels”. The time over which the sound is integrated is given as an index. Leq is usually calculated for a whole day (24 h) or for daytime (16 h) and nighttime (8 h) separately. The latter two indexes are then called LeqD and LeqN. These two indicators are sometimes combined to a “day–night level”, LeqDN, which is the addition of the two after 10 dB(A) is added to the night hours to account for larger effects of noise during nighttime. Even though weighted sound levels are not strictly physical flows, they are still potential LCI parameters according to the standard (ISO 14′040 2006), which foresees not only physical data but also data for “other environmental aspects” to be included in the inventory.
3.4 Requirement profile for noise inclusion methods for LCA
Consideration of generic and specific transports in LCA
Separate treatment of different modes of transport
Separate treatment of different vehicles within one mode of transport
Accounting for transports in different geographic contexts
Accounting for different temporal contexts of transports
Compliance with ISO 14′040/44
The first requirement ensures that transport data are included in LCA studies as accurately as possible, i.e., that all relevant information which is available can be used. Often, information is not sufficient for specific modeling of transports in LCA studies. Other studies and environmental reports of companies need a specific consideration of transports and comparison of transport alternatives (e.g., between road and rail or Euro 3 and Euro 4 truck). This is reflected by requirements 2 and 3. Requirements 4 and 5 ensure that spatial and temporal aspects which are very important for noise emissions and effects but are usually neglected in LCA are heeded. Noise emissions of transports are highly dependent on the geographical context, mainly traffic speed and volume. Especially traffic volume depends on the time of the day and also on the day of the week. Noise propagation and damage caused by traffic noise depend on the geographical context (mainly shielding effects and local population density). Noise effects also depend on the intended activity of the exposed persons, which again depends on the time of the day. Together with requirement 6, these requirements implicitly determine that a method for the inclusion of traffic noise in LCA needs to be generic and thus suitable for inclusion in generic LCI databases. However, at the same time, the method needs to respect the temporal and spatial context of the transport and its contribution to the sound level and the effects.
4 Existing methodological approaches for the inclusion of traffic noise in LCA
Compliance of existing methods for inclusion of (traffic) noise in LCA with the postulated requirements (Section 2.1.3)
Danish LCA guided
Swiss EPA methodf
Swiss FEDRO methodg
Short characterization of the method
Unspecific physical addition of energy
Site specific threshold exeedance from road traffic
Context ignoring nuisance from site specific road and rail traffic
Generic additional DALYs from generic additional road traffic
Generic additional environmental scarcity points from specific additional road traffic
Suitable for generic and specific transports in LCA
Comparison of different modes of transports
Comparison of different vehicles within one mode of transport
Suitable for comparison of transports in different geographic contexts
Suitable for comparison of transports in different temporal contexts
Complies to ISO 14′040
4.1 CML guide for LCA
The CML guide for LCA (Guinée et al. 2001; Heijungs et al. 1992) proposes an unweighted aggregation of sound, in energy density per time units (Pascal-squared second or (joule per cubic meter)-squared second). This indicator is not limited to traffic noise but could be used for stationary sources of sound as well. This measure of sound energy has been criticized as being useless, since it depends on the measurement point (Lafleche and Sacchetto 1997). It should thus be integrated over the volume of the portion of space affected by noise to be suitable for LCA studies, which is claimed to be impossible due to lack of data (Lafleche and Sacchetto 1997). Another difficulty with this index is that it adds sound energy disregarding temporal and spatial aspects as well as all the situational and individual factors determining the perception and effect of the sound. Thus, even though the CML guide indicator could be calculated for any situation, it is rated not suitable for all the different comparisons in Table 3, since differences in the indicator values would not reflect differences in human health effects.
4.2 Ecobilan method
Lafleche and Sacchetto presented the first specific “methodological attempt” (Lafleche and Sacchetto 1997, p. 111) at traffic noise assessment in LCA. Their starting points are calculated or measured noise levels along traffic infrastructure and noise thresholds for day- and nighttime. From that, they calculate the surface of the area affected by noise above threshold and the perpendicular distance from the road encompassing disturbed people. This threshold distance is multiplied by the local population density and the result is integrated over the total length of the trip. Thus, the number of instantaneously disturbed persons along the infrastructure of the trip is derived. Finally, the effect is allocated to single vehicles and functional units. This procedure allows for an assessment of a specific road trip but provides no means for assessing generic transports other than averaging all specific road trips in a region of interest. Also, the number of people living in areas above threshold values is a very rough indicator for health effects of noise since effects start occurring below threshold and since they continuously increase with increasing noise level above the threshold.
4.3 Danish LCA guide method
Another method for including specific traffic noise in LCA (Nielsen and Laursen 2005) was proposed as part of the Danish LCA guide (Potting and Hauschild 2003), which emphasizes spatial differentiation in LCA. The method calculates the “noise nuisance impact potential” in person seconds. It is based on a “noise nuisance factor” (NNFLp), i.e., the number of persons affected by the peak noise of a single vehicle and the duration of the noise. The NNFLp represents the inconvenience caused by the noise to humans as a function of the part of the noise that exceeds the background noise. It is a subjective parameter which was determined for traffic noise in interviews. Sound pressure levels are calculated by emission and propagation modeling according to the Nordic prediction method (Nordic Council of Ministers 1996); for exposure modeling, average population densities are applied. The method provides calculation routines for three different sizes of lorries on three different types of road in five different areas and for a train in two different areas. Thus, this method distinguishes different traffic situations and can be applied to freight transports by road and rail. It could easily be expanded to include passenger transport in cars, buses, and trains and the principle could also be applied to air traffic noise, and even though temporal aspects are not taken into account, the possibility of differentiating daytime and nighttime noise is proposed in the outlook (Nielsen and Laursen 2005). Unfortunately, the accessible information does not allow for a reconstruction of the calculation1. This method directly calculates noise impacts for a unit process, but does not provide LCI parameters for the noise emission of this unit process and is thus not suitable for inclusion in generic LCI databases. On the other hand, the method is designed to satisfy the need for specific consideration of geographic context. However, it overestimates road traffic effects due to the use of maximal sound levels instead of equivalent sound levels in the calculation of the impact and does not consider the source specificity of impacts when comparing road to rail traffic. For a detailed discussion of the method, see part 2 (Althaus et al. 2009).
4.4 Swiss EPA method
The first method for generic road transports was proposed by Ruedi Müller-Wenk (2002, 2004). Based on the Swiss traffic noise model from 1991 (SAEFL 1991), it calculates marginal noise levels for additional passenger cars and lorries in various real traffic situations in Switzerland. Müller-Wenk presumes that if the available information on a road transport does not indicate the routing but is restricted to the amount of vehicle kilometers driven on a country’s road network, the best guess is to assume that the transport is distributed fractionally over the whole road network, in proportion to the preexisting traffic on each of the road segments. Müller-Wenk further suggests that under this assumption, it can be shown that the additional transport increases the noise level in dB, on each of the road segments, by roughly the same amount. As a consequence, the increase of the road noise exposure of the country’s whole population, due to the additional transport, can be calculated easily: It is a constant shift of the whole population’s preexisting road noise distribution toward a marginally higher level. If dose–effect relationships are available for noise-related health effects, it is then possible to derive the numbers of additional health cases from this constant shift toward marginally higher road noise levels. Thus, generic additional noise levels (ΔL) for 1,000 km additional transports on the Swiss road network are calculated for passenger cars and lorries for day- and nighttime. The additional effects are calculated based on marginally increased data of the actual exposure in the Canton of Zurich extrapolated to the whole of Switzerland. The effects considered are communication and sleep disturbance. Based on a Swiss survey made in 1990 (Oliva 1998), a linear increase in sleep and communication disturbance with increasing sound level and threshold levels where sound starts to become disturbing was approximated. Thus, the effect analysis results in additional cases of communication and sleep disturbance due to the additional transports. These effects could be used as midpoint results in LCA. However, the Swiss EPA method went one step further by assessing the damage caused by these effects. Therefore, disability weights for communication and sleep disturbance were established and used. Thus, the final result is presented in disability adjusted life years lost (DALYs) and can be directly compared with damage to human health through toxic emissions, calculated, e.g., according to the EcoIndicator 99 (Goedkoop and Spriensma 2000), or through accidents. This method fully complies with the relevant standards (ISO 14′040, 2006; ISO 14′044 2006) and is suitable for inclusion in background LCI databases. Its major drawback is that it only provides data for two generic types of road vehicles and for an average additional journey. Thus, comparisons of different means of transport (e.g., train versus road), of different vehicles of the same type (e.g., lorry A versus lorry B), and of different routes taken (e.g., national highway vs. country road) are not possible. However, the general framework can also be applied to specific vehicles and to specific journeys if the information on vehicle specific emission, local traffic volumes, and on population densities along the roads is known. The adaptation of the method to rail and air traffic noise is possible if for those modes of transport, the assumption on additional traffic being proportional to existing traffic holds true and if additional noise for these modes is also therefore independent of the location of the emission. Due to assumptions and simplifications, especially due to the lack of differentiation of heavy vehicles from motorcycles, the method, however, might overestimate noise effects, especially for heavy vehicles. For a detailed discussion of the method, see part 2 (Althaus et al. 2009).
4.5 Swiss FEDRO method
The Swiss EPA method (Müller-Wenk 2002, 2004) was used as a basis for the method developed for the Swiss Federal Roads Office (FEDRO) (Doka 2003). It proposes ecological scarcity factors for road traffic noise in accordance with the Swiss Eco-Scarcity method (Brand et al. 1998), which calculates burden factors by relating actual flows to the square of critical flows. The Swiss FEDRO method chose the number of highly annoyed people as actual flow and 20% of the Swiss population as critical flow. This critical flow is derived from the immission threshold in Switzerland, which is set to a level where 20% of the population is highly annoyed by the noise. The actual flow is derived from the exposure data from the Swiss EPA method (Müller-Wenk 2002) and from effect curves based on the Swiss survey from 1990 (Oliva 1998). Thus, burden factors per highly annoyed person for day- and nighttime noise are calculated. The number of highly annoyed persons is calculated based on Müller-Wenk’s (2002) method but the emission model was changed to a more recent one (Heutschi 2004a, b). Effect analysis ends at the chosen midpoint of highly annoyed persons. The Swiss FEDRO method also introduced vehicle specific noise emissions, based on type approval noise tests, and showed that noise effects vary by approximately a factor of four for vehicles with type approval values between 69 and 75 dB(A). This method is thus generic in terms of time and place of the emissions and specific in terms of the vehicle and additionally allows for intramodal comparison of vehicles. However, there is even evidence from measurements that vehicles with lower noise values from test approval measurement do not necessarily cause lower sound emissions in real traffic situations (Steven 2005). Thus, type approval test results are not suitable for use in combination with the empiric noise emission calculation procedure proposed in SonRoad (Heutschi 2004b). For a detailed discussion, see part 2 (Althaus et al. 2009).
A requirement profile for methods to include traffic noise effects in LCA was developed based on a sample of 66 LCA case studies, eight environmental reports, and theoretical considerations. This requirement profile is very difficult to meet since it demands genericness and consideration of spatial and temporal aspects at the same time. Existing methods either lack the genericness for a wide application or lack specificity for the differentiations needed in LCA studies. The Swiss EPA method (Müller-Wenk 2002, 2004) introduces a highly promising way of dealing with this problem for road traffic noise. Based on that, the Swiss FEDRO method (Doka 2003) additionally allows an intramodal comparison beyond that of lorries and passenger cars. Meeting the demand for intermodal comparisons is attempted by the Danish LCA guide method (Nielsen and Laursen 2005) but the method directly calculates noise impacts for a unit process, but does not provide LCI parameters for the noise emission of this unit process and is thus not suitable for inclusion in generic LCI databases.
6 Conclusions and recommendations
A method complying with the requirement profile is needed since an inclusion of traffic noise effects could considerably change the overall results of many LCA studies. However, since none of the existing methods fully complies with the requirement profile, none of the major LCI databases includes traffic noise indicators. Moreover, noise effects are usually not considered in LCA including transports. Two of the methods proposed so far are in principle suitable for inclusion in generic LCI databases and a third allows for intermodal comparison. However, the methods need to be thoroughly analyzed and weaknesses need to be amended in order to make them widely usable.
An in-depth analysis of the promising methods is made and improvement potential is evaluated in part 2 (Althaus et al. 2009), where also a new framework for context-sensitive inclusion of relevant human health effects from road transportation noise is proposed. Appropriate methods for rail and air traffic noise will have to be developed in the future since noise effects are also relevant in these contexts (e.g., Brons et al. 2003; Clark et al. 2006; Griefahn et al. 2006; Lam et al. 2008; Lu and Morrell 2006; Miedema 2004; Raschke 2004; Sandrock et al. 2008; Schuemer et al. 2003; Wirth 2004). Thus, a methodological framework fully compliant with the requirement profile can be achieved.
We thank Gabor Doka, Ruedi Müller-Wenk, and the anonymous reviewers for their valuable comments, Kurt Heutschi and Kurt Eggenschwiler from the Empa acoustic laboratory for fruitful discussions, and the Swiss Federal Office for the Environment FOEN for financial support.