Cost-benefit analysis of low-noise pavements: dust into the calculations

Abstract

This paper presents a cost-benefit analysis (CBA) of roads noise measures in Norway. Low-noise pavement alternatives were compared to stone mastic asphalt with a maximum aggregate size of 11 mm. The low-noise alternatives were expected to reduce the noise levels by 1–4.5 dB over their lifetime, compared to the reference, but had shorter lifetime and, mostly, higher investment cost. A new element included into our CBA of low-noise asphalts is their property in terms of asphalt-wearing and dust production. Given a relationship between asphalt-wearing and airborne particulate matter, there is a potential health impact of pavement choice. Official valuations of both noise changes and PM₁₀ changes were applied for the benefit estimations. Thin-layer asphalts obtained higher benefit-cost ratios than porous asphalts, mainly due to small changes in unit costs and technical lifetime compared to the reference. Alterations in dust production had considerable weight in the benefits, but did not considerably alter the ranking of asphalts compared to analyses not taking dust into account.

Keywords:

• asphalt-wearing properties
• economics
• health impact
• particulate matter
• porous asphalt
• risk analysis
• thin-layer asphalt

Notes

^1. Road noise control can be regarded as a local public good. Noise reduction measures can be portioned out to certain localities (e.g. giving a preference to measures in areas with the wealthier segment of the population), but all those living (or working or performing leisure activities) in these areas cannot be denied the benefit of a less noisy environment – the noise reduction is a non-exclusive good to all those within the locality. The noise reduction will also be non-rival in the sense that any person's ‘consumption’ of the reduced noise will not reduce other persons’ ‘consumption’ of the same good within the locality (Hanley et al. 1997).

^2. Elvik and Greibe (2005) concluded, from a meta-analysis of 18 studies, that there were no statistically significant effects on road safety (and thus, crash costs) of changing to noise-reducing pavements (porous asphalts). Furthermore, regarding vehicle operation costs, Bendtsen (2004) did not find evidence for differences in rolling resistance and fuel consumption between standard dense asphalts and porous asphalts. Also, additional external (environmental) effects are probably limited and/or going in both directions. Berbee et al. (1999) and Pagotto et al. (2000) argue that noise-reducing porous asphalt has an adsorption property allowing a more gradual run-off of water (limited peak flows and slower discharge) and a filtering effect. This could yield an additional benefit that our analysis omits. Yet, the same porous asphalts may be recycled to a slightly lesser degree than dense asphalt types (Litzka et al. 1999, Pucher et al. 2004); and the gradual run-off may also imply more pollutants in the worn asphalts, although Descornet et al. (1998) found that the quantities of pollutants in worn porous asphalt were generally relatively low.

^4. The average double-layer porous asphalt (DPAC 11/16) represents different pavement types in terms of aggregate size and void content in surface layers, but all with similar stone quality. The ‘best possible’ version (DPAC 11/16x) has a larger share of aggregate larger than 4 mm and slightly higher void content than the other pavement types (Lerfald 2008, Evensen 2009).

^3. Porous asphalt has a high stone content and a grading that provides a high void content (>20%) – a feature that increases sound absorption. Double-layer porous asphalt consists of two layers of porous asphalt: a coarse open-graded bottom layer and a finer textured top layer. Thin-layer asphalts are very thin and relatively open layers (∼14%) that are laid on a thick layer of polymer-modified bitumen emulsion (Morgan 2006).

^5. Compared to other measures to control road noise, the great advantage of noise-reducing pavements is the reduction of the (tyre/road) noise at the source (Sandberg 2001). The potential reduction of noise levels will depend on speed levels, average annual daily traffic (ADT) and this traffic's composition of light (and medium) and heavy vehicles (Morgan 2006, Ch 11). The tyre/road noise (vs. propulsion noise) will be more dominant at higher speeds. Still, a pavement producing less tyre/road noise may yield a noticeable effect on total vehicle noise at speed levels as low as 40 km/h for light vehicles. Expected noise reduction from low-noise pavements, at different speeds, is assessed either at test sites or at roadside, and the reference pavement is some standard dense asphalt type (Morgan 2006, Ch 4, Table 4.2). The CBA should principally also compare the profitability of noise-reducing asphalt to other possible noise-reducing measures, e.g. noise barriers or facade insulation.

^6. Test values from the Nordic ball mill test (NB) enter the estimation of a pavement wearing parameter: wearing parameter =  (NB / aggregate size) ×  100. The Nordic ball mill test is a wet abrasion test that determines resistance to wear by abrasion from studded tyres (CEN 1998). The test has shown high correlation with studded tyre wear, applying different pavement stone materials (Snilsberg 2008). An alternative test considered was the Tröger test, where measures of the resistance to studded tyres are based on asphalt dust formed from hammering asphalt cores by steel threads, yielding a Tröger value (D _k ) defined as: where Δm is the total particulate matter produced (lost) over a given period, and ρ _d is the density (Raitanen 2005, Lerfald 2008). Notwithstanding the test then type, yielding estimates of dust from different pavements, it should be remarked that there is not necessarily very high correlation between the amount of (airborne) dust and the amount of PM₁₀ (Snilsberg 2008) and it is the latter that enters the cost-benefit analysis.

^7. The Norwegian Institute of Public Health has applied the following equation for estimating the change in health effects for a given area: PM = D ₀ × ΔRR × ΔE _x/10 μg/m³, where PM is the number of incidents caused by PM₁₀ increase, D₀ is the number of incidents, per year in the given area, ΔRR is the increase in relative risk, and ΔE _x is the increase (μg/m³) in PM₁₀ (NIPH 2005, Snilsberg 2008).

^8. The simulations will yield estimates of the expected values – the more the iterations, the closer we get to these expected values; i.e. the simulation error (ϵ) is proportional to the number of iterations (N), where σ is the standard deviation. Thus, increasing the number of iterations above 1000 will yield an error lower than 10%.

^9. Earlier versions of the spreadsheet model were applied under the EU-project SILVIA – ‘Sustainable Road Surfaces for Traffic Noise Control’ (Morgan 2006, Veisten et al. 2007a) and the Norwegian projects ‘Societal consequences and cost-benefit analyses of low-noise pavements’ (Veisten 2006) and “‘TORNADO” – a tool for strategic cost-benefit analyses of noise control measures and other environmental measures’ (Veisten et al. 2007b).