Waking reactions predicted by SEL or LAeq: a big difference

Elly H. Waterman,

NS Technisch Onderzoek, P.O.Box 8125, NL-3503 RC Utrecht, The Netherlands

e-mail: e.h.waterman@geluid.nsto.ns.nl


To calculate the environmental impact of a railway line, the Equivalent Sound Pressure Level (LAeq) outside the building is generally used. Research by the Dutch Health Council [1] has shown, however, that the LAeq is not a good indicator for the number of waking reactions. In 1997 a new dose-effect relation for waking reactions became. This relation is based on the Sound Exposure Level (SEL) in the bedroom. The waking reactions depend on the number of individual train passages during the night exceeding a threshold SEL level.

In this paper it is shown that this dose-effect relation gives unexpected results. A lower LAeq might in certain cases result in a higher number of waking reactions.


Waking reactions caused by (railway) traffic during the night are a great nuisance for people living along railway lines and may interfere with their wellbeing. Therefore the general increase of nightly traffic might cause a future public health issue. Correct prediction of the expected waking reactions is of importance to planners of existing and new railway lines, in particular to be able to take sufficient noise reduction measures.

To determine the environmental effects of traffic noise the Equivalent Sound Pressure Level (LAeq) has been in long term use. Recent studies [2] show that noise measures like the Lden, which is based on LAeq, are a reliable indicator of the general annoyance. To take the annoyance during the night into account, a penalty of 10 dB(A) is usually added to the equivalent noise level during the night. In a recent publication of the Dutch Health Council [1] it was shown, however, that the LAeq is not a good indicator of the number of waking reactions. A relation based on the Sound Exposure Level (SEL) in the bedroom should be preferred. The number of waking reactions appear to depend on the number of individual noise events (train passages) during the night exceeding a threshold SEL level. This dose-effect relation is considered as a given relation in this paper.

Dose-effect relation for waking reactions

The new dose-effect relation for determining the expected number of waking reactions is valid for isolated noise events during the night. The probability of awakening, Pw is related to the indoor SEL level of a single noise event as follows:



This relation is valid only for SEL levels higher than 55 dB(A). Below this threshold level no awakenings are expected to occur. Formula (1) seems to indicate that the probability might reach a value higher than 1. In practice this will not occur. To reach the value of 1, the SEL value should be as high as 600 dB(A), which is impossible. If during the nighttime a number of n noise events is occurring, the probability of awakening should be multiplied by a factor n. In the following sections, the expected number of noise events is calculated for the period of a year. This results in numbers, which are easy to understand.

The probability to wake up during the night is related in a linear fashion to the SEL level. An increase of the SEL with 10 dB will only increase the probability to wake up with 1.8 %. However, if the number of noise events becomes 10 times as high, the probability to awake will also be 10 times higher.

In figure 1 the relation between LAeq and the number of awakenings is shown for various numbers of noise events per night. In this figure it is assumed that all noise events are identical. In situations with the same LAeq, for instance 50 dB(A), an enormous range in the number of waking reactions may occur, depending on the number of noise events per night. At 50 dB(A) equivalent noise level this ranges from about 25 awakening per year up to 350 per year. It also appears that lowering the indoor noise level whilst increasing the number of noise events, might increase the number of awakenings. Such situations may occur if a home is better isolated in order to reduce annoyance due to increased traffic.

Figure 1. Relation between the number of noise events and the number of awakenings, for various values of the indoor equivalent noise level, according to equation (1).

Calculation of the SEL level caused by a train pass-by

The Dutch noise regulations for railway noise are based on the equivalent noise level during the night, LAeq,night. Calculation schemes have been developed to calculated this noise level on the outside of a dwelling, based on the amount of traffic, the distance to the track etc. Determination of the SEL from the calculated LAeq can be performed if the number of passages is known. The simplest case is the situation that during an 8 hour night time period n identical trains are passing by. Identical means in this case that the trains are of the same type, are having the same length, and are running with the same speed.

The indoor SEL level of each train passage is in this case given by:


The term NR is the noise reduction of the outer wall of the bedroom. For goods trains this is assumed being 20 dB (for a typical home in the Netherlands). For passenger trains NR amounts to 23 dB. If various types of trains are passing, their SEL can be calculated on basis of a calculated equivalent noise level per train type. In table 1 indications are given of outdoor SEL levels. SEL levels may vary due to source and receiver height, as well as various noise propagation parameters.

Type of train

Number of wagons per train




50 m


100 m


200 m


500 m

Disk braked passenger train







High speed train







Goods train







Table 1. Indication of outdoor SEL levels for various trains at several distances from the track

Application of dose-effect relation for waking reactions

The dose-effect relation of equation (1) has been applied in several environmental impact studies for new railway lines in the Netherlands [3], [4]. These studies show unexpected results. At a lower equivalent noise level, a significantly higher number of waking reactions may occur. To illustrate this result, the following example is included.


This example is taken from an environmental impact study for a goods line between Rotterdam and Antwerp. There are several alternatives studied for this goods line. Some alternatives make use of an existing railway track through a densely populated town. Other alternatives will avoid this town, but are much more costly. If the future goods line will cross the town, the number of goods trains during the night will increase substantially. This is not allowed without acoustical measures, which will be taken on basis of the Dutch noise regulations. Accordingly, the indoor equivalent noise levels will be reduced to a maximum of 35 dB(A) during the night. In the current situation the indoor night time noise levels may be higher than 40 dB(A). If the goods lines will lead through the town, the acoustic situation will improve significantly, at least, this appears to be the case on basis of the equivalent indoor noise levels.

However, for the number of waking reactions the situation is different! It appears that waking reactions will might increase. In table 2 one of the situations that might occur is presented.


Indoor noise level during the night LAeq,8hours

Number of trains per night (total of 3 types)

Predicted number of waking reactions per year

Current situation




Future situation with goods traffic




With extra measures




Table 2. Example of expected number of waking reactions for 3 situations along a railway line with future goods transport.

It appears from table 2 that, although the indoor noise level is reduced with 5 dB(A), the expected number of waking reactions will increase with a factor 2.5. Even if additional noise measures are taken, achieving a total noise reduction of 10 dB(A), the number of waking reactions show a 38% increase with respect to the current situation. Based on this large reduction of 10 dB(A), one would initially expect that the number of waking reactions should be substantially less.


The number of noise events during the night is the most important factor for the occurrence of waking reactions. The noise level of the individual events is much less relevant. It is found that noise reduction measures based on the LAeq model do not provide sufficient protection against waking reactions. In some cases, lowering the LAeq, whilst increasing the number of noise events, might increase the number of waking reactions. To ensure that the situation for awakening does not deteriorate, substantial additional noise measures will be required in such cases.



  1. Dutch Health Council, Calculating environmental noise, proposal for a uniform system of noise measures for the determination of annoyance and sleep disturbance by noise, Dutch Health Council, 1997, publication nr. 1997/23 (in Dutch)

  2. H.E. Miedema, H. Vos, Exposure-response relationships for transportation noise, J. Acoust.Soc.Am. 104(6)1998

  3. NS Railinfrabeheer, Environmental impact study for the Hanzelijn passenger railway, to be published (in Dutch)

  4. NS Railinfrabeheer, Environmental impact study for the railway goods line between Roosendaal and Antwerp, to be published (in Dutch)