Annoyance and sleep disturbance due to vibrations from trains in The Netherlands: results from the follow-up study “Living in the vicinity of the railway”. Elise van Kempen 1 , Sendrick Simon, Harm van Wijnen, Arnaud Kok, Nick Mabjaia, Irene van Kamp RIVM, P.O.Box 1 3720BA, Bilthoven, The Netherlands

ABSTRACT In 2013, RIVM investigated how people, living in the vicinity of railways, experience vibrations due to trains. As a follow-up, and in response to questions from Parliament on the expansion of the Dutch rail network, the State Secretary of Infrastructure and Water management commissioned a second study and a repeated measurement of the persons who participated in 2013. This abstract focusses on the second study. The main objective of this study was to derive exposure-response relationships between exposure to vibrations caused by trains and annoyance and sleep disturbance. The results are important for the further development of policy, and possibly also regulations, on railway vibration in The Netherlands. For this follow-up study, 16.000 people of 16 years and older, living within 300 meters of a railway track were invited to participate. A second group of 1,189 persons who participated in two rounds of the previous study were also invited. By means of an online questionnaire, administered in September 2021, information was gathered about annoyance, sleep disturbance due vibrations and noise from trains and their co-determinants. The participants’ exposure to vibrations from trains was estimated by means of an improved version of the Dutch calculation model for railway vibrations (OURS).

1. INTRODUCTION

In 2013, RIVM investigated how people, living in the vicinity of railways, experience vibrations due to trains. To this end, a questionnaire survey was held among 4,927 people living with 300 meters from a railroad track in the Netherlands. It was estimated that in The Netherlands about 845,000 residential addresses, with some 1.347.400 residents of 16 years and older, are located within 300 meter distance to a railroad track. On the basis of the results of the questionnaire, it was estimated that about twenty percent of these residents experience severe annoyance from vibrations caused by trains. By far the largest part of annoyance is reported in relation to freight trains. Similar findings were found for sleep disturbance. As part of the study also exposure-response relationships were derived for the association between exposure to vibrations from trains and the percentage severe annoyance and sleep disturbance [1, 2]. However, these exposure-response relationships should be interpreted and applied with caution for several reasons: Firstly, exposure to vibrations was assessed by means of the Standard Calculation method Vibrations (SRM-T) [3]. This model used generic

1 Elise.van.Kempen@rivm.nl

model parameters. However, the situation may differ locally. Exposure misclassification cannot be ruled out. Secondly, the non-response study indicated possible selective non-response: it appeared that in particular severely annoyed people participated in the study. Thirdly, the study was carried out in a change situation: on a number of routes, more trains were running [1]. As a follow-up of the study of 2013, and in response to questions from Parliament on the expansion of the Dutch rail network, the State Secretary of Infrastructure and Water management commissioned a second study. The main objective was to derive exposure-response relationships between exposure to vibrations caused by trains and annoyance and sleep disturbance. To overcome the above-mention problems regarding exposure misclassification, we will make use of vibration exposure levels assessed by means of a new model: In 2016, the Secretary of State of the Ministry of Infrastructure and Water management commissioned RIVM to develop a uniform calculation method for vibrations from the railroad track [4]. In May 2020, this resulted in the first version of ‘Development Uniform Calculation model Railroad vibrations’ (OURS), that can amongst others be used for the calculation of expected vibration levels. In contrast with SRM-T, OURS uses better input data for soil, railway traffic, railway construction and dwellings [5].

2. METHODS

2.1. Sampling and recruitment

Two groups of participants entered the study: (i) participants obtained from a fresh sample, and (ii) participants who had already participated in two rounds of the previous questionnaire study. For the first group, we obtained the address data of all buildings in the Netherlands from the 2018 “Basisregistratie Adressen en Gebouwen”-file of the Dutch Land Registry (BAG, 2018), using a Geographical Information System (GIS). For all these buildings, we calculated the distance (in meter) to the closest railroad track. We selected all addresses with a residential function (n = 848,467). We excluded the addresses (i) of participants who were also invited for the study in 2013, (ii) located close to decommissioned railway tracks, or (iii) located close to a tunnel or bridge. This resulted in more than 532,000 addresses. Based on distance to the railway track and the year of construction of the building we created six strata (see also table 1). Subsequently we drew a random sample from the available residential addresses for each stratum. This resulted in a gross sample of 16.000 addresses.

Table 1: Overview of the strata and the required number of addresses per stratum. Stratum Distance to railroad

Year of construction

Required number of

track (m)

of building

addresses 1 < 50 Before 1950 4,000 2 < 50 From 1950 4,000 3 50 – 100 Before 1950 2,400 4 50 – 100 From 1950 2,400 5 100 - 300 Before 1950 1,600 6 100 - 300 From 1950 1,600 Total 16,000 In addition to the 16.000 addresses, we invited a second group of 1.189 persons who had already participated in the first study in 2013 [1, 2], and again later in 2019, and who indicated that we could approach them again. In Simon et al [6] more details can be found. In September 2021 we sent an invitation letter to the 16,000 selected addresses. From the residents at an address, we asked the person who was sixteen years or older and whose birthday was the next to

occur, to fill in an online questionnaire. We also sent a letter to the 1.189 participants of 2013 and 2019 (the second group). In this case we asked the person who had already participated in the two previous measurements to fill in the questionnaire. After the first invitation, we sent two reminders at intervals of about two weeks. To further increase the response, we raffled a prize among the participants. In order to determine to what extent participants differed from non-participants, a non-response study was performed among a selected sample of non-responders.

2.2. Questionnaire

By means of an online questionnaire, we gathered information about the following aspects: (i) perception and satisfaction with direct living environment, (ii) perception and experience of vibrations from the railway track and other sources (incl. annoyance and sleep disturbance), (iii) perception and experience of noise situation (incl. annoyance and sleep disturbance due to noise from the railway track), (iv) how people think about living alongside a railway track (including attitudes, worries and acceptance) (v) health (incl. general health and sleep quality), (vi) characteristics and construction of the participants house, and (vii) demographics of the participant and his family. In order to measure constructs such as annoyance, general health, sleep quality, demographics, we made as much as possible use of national and international standards.

2.3. Exposure to vibrations

While writing this paper, the calculations of the exposure to vibrations will be completed. We will estimate the participants’ exposure to vibrations from trains by means of the latest version of the Dutch calculation model for railway vibrations (OURS, version 2.0, January 2022). Based on the vibration source (emission of trains and railway track) and the transfer in the soil towards the foundation of the building, the OURS model calculated vibration levels caused by the trains. In different countries, different measures of exposure are being used. The European project ‘Attenuation of ground-borne vibration affecting residents near freight railway lines’ (CARGOVIBES) found that measures that not only depend on the momentary level of vibration but also on the duration of the vibration (the number of train events) might be better predictors for annoyance [7]. An analysis of Janssen et al [8] showed that the root-mean-square (RMS) has an added value in predicting annoyance. Consequently, we will not only express exposure to vibrations from trains by means of V dir,max , but also by means of the RMS and V per . 3. RESULTS

3.1. Response

The fieldwork was completed in November 2021. In total 5,611 (33.2%) persons completed the online questionnaire. The response per group differed: in group 1, the response was 30.2%, while in group 2 the response was almost 70%.

Table 2: Response overview

Group 1 Group 2 Total

Invited 16,000 1,189 17,189

Invalid address 160 - 160

Died or moved - 2 2

No permission 112 10 122

Usable address 15,728 1,177 16,905

Not completed 10,937 357 11,294

Completed 4,791 820 5,611

Response (%) 30.2 69.7 33.2

3.2. Characteristics of the study population Table 3 illustrates some general characteristics of the study population. It shows that: (i) group 2 contains relatively more men compared to group 1; (ii) group 2 is relatively old: it contains only a few persons below 45 years compared to group 1; (iii) the participants are highly educated: the highest educational level of almost 50% of the participants includes higher professional education or science education; (iv) more than 75% of the participants is homeowner and lives at least 5 years in the current dwelling; (v) almost 60% of the participants lives within 50 meters of a railway track

Table 3: General characteristics of the participants

Group 1

Group 2

Total % (n)

% (n)

% (n)

Sample (n = 4,791) (n = 820) (n = 5,611)

Sexe Male Female

100.0 (4,791)

95.6 (784) 60.8 (477) 39.2 (307)

99.4 (5,575) 53.8 (3,001) 45.9 (2,558)

52.7 (2,524) 47.0 (2,251)

Other

0.3 (16)

0

0.3 (16)

Age (yrs.)

99.9 (4,790) 29.4 (1,408) 41.0 (1,963) 29.6 (1,419)

95.1 (780)

99.3 (5,570) 26.1 (1,456) 41.9 (2,331) 32.0 (1,783)

16-44 45-64

6.2 (48) 47.2 (368) 46.7 (364)

65+

Highest educational level *

100 (4,791)

100 (820)

100 (5,611)

Lower education Middle education 1 Middle education 2

2.0 (94) 18.1 (866) 31.8 (1,524) 48.2 (2,307)

0.7 (6) 14.3 (117) 31.6 (259) 53.4 (438)

1.8 (100) 17.5 (983) 31.8 (1,783) 48.9 (2.745)

Higher education

100 (4,791) 75.9 (3,635)

100 (820) 74.9 (614) 20.7 (170)

100 (5,611) 75.7 (4,249) 20.1 (1,128)

Dwelling ǂ

1 2 3

20.0 (958)

4.1 (198)

4.4 (36)

4.2 (234)

Homeowner

100 (4,791) 77.7 (3,722) 22.3 (1,069)

100 (820) 85.4 (700) 14.6 (120)

100 (5,611) 78.8 (4,422) 21.2 (1,189)

Yes

No

Distance to railway track

100 (4,791) 58.3 (2,793) 27.5 (1,318)

99.6 (817) 64.0 (523) 25.0 (204)

99.9 (5,608) 59.1 (3,316) 27.1 (1,522)

< 50 m 50 – 100 m 100 – 300 m

14.2 (680)

11.0 (90)

13.7 (770)

Lived in current dwelling

99.9 (4,790) 28.3 (1,355) 71.2 (3,435)

95.1 (780)

99.3 (5,570) 24.3 (1,355) 75.7 (4,215)

Less than 5 yrs.

0 100 (780)

At least 5 yrs.

*Educational level: Lower education includes no education or primary school, Middle education 1 includes primary and pre-vocational education and general secondary education, Middle education 2 includes secondary vocational education and higher general and pre-university education, Higher education includes higher professional education and science education; ǂ Type of dwelling: 1 = Terraced/corner/(semi)detached house, 2 = (part of) flat, 3 = Other

4. CONCLUSIONS

At the moment, in The Netherlands there is limited legislation to prevent annoyance and other health effects or damage to the dwelling from vibrations caused by rail traffic. This is problematic, because now and in the near future an increase of both passenger and freight trains is foreseen. This is a result

of the goal to improve accessibility and to achieve climate targets. Furthermore, during recent years an increase in the number of complaints from people living close to railroad tracks has been observed. In response to questions from Parliament on the expansion of the Dutch rail network, the State Secretary of Infrastructure and Water management commissioned a study, with the aim to derive exposure-response relationships between exposure to vibrations caused by trains and annoyance and sleep disturbance. The results of this study are important for the further development of policy, and possibly also regulations, on railway vibration in The Netherlands. Unfortunately, we are only able to present some pre-liminary results in this paper. While writing this paper, the calculations of the exposure to vibrations will be completed. In addition, we are calculating noise exposure levels for all participants (expressed as L den and L night in dB). Both exposure data will be linked to the questionnaire data. Furthermore, we will enrich the questionnaire data at address level with information from different registries about situational and contextual factors such as the number and types of trains, speed, level of urbanization. During the conference we expect to present more results. 5. ACKNOWLEDGEMENTS

This project was funded by the Dutch Ministry of Infrastructure and Water management. 6. REFERENCES

1. Van Kamp, I., E.E.M.M. Van Kempen, H.J. Van Wijnen, E. Verheijen, T. Istamto, O.R.P. Breugelmans, Health effects of vibrations due to trains (in Dutch) . 2015, RIVM: Bilthoven. 2. Van Kamp, I., E. Van Kempen, H. Van Wijnen, E. Verheijen, T. Istamto, Ferreira J., O. Breugelmans, L. Dirven, A. Koopman, Health effects of vibrations due to trains , in The tenth European Congress and Exposition on Noise Control Engineering (EURONOISE) 2015, Nederlands Akoestisch Genootschap, Belgische Akoestische Vereniging, European Acoustics Association: Maastricht. p. 599-633. 3. Steenbergen, R.D.J.M., S.S.K. Lentzen, A. Koopman, Betrouwbare trillingsmaatregel ontwikkeling ontwikkeling Standaard Rekenmethode voor Trillingen (SRM-T) . 2009, TNO: Delft. 4. De Gruijter, D., A. Koopman, E. Verheijen, S. Lentzen, H. Boshuizen, J. Bolte, A uniform calculation method for railway vibration 2016 (in Dutch) . 2016, RIVM: Bilthoven. 5. Kok, A. OURS: Dutch calculation method for railway vibrations . 2021 February 9, 2022 [cited 2022 April 14, 2022]. 6. Simon, S., E. Van Kempen, H. Van Wijnen, A. Kok, N. Mabjaia, I. Van Kamp, A longitudinal study into response to vibrations due to rail traffic: design, questionnaire and response , in INTERNOISE 2022 . 2022: Glasgow. 7. Janssen, S.A., H. Vos, A. Koopman, Deliverable D1.2 - Exposure-response relationships and factors influencing these relationships . 2013, CargoVibes: Leiden. 8. Janssen, S.A., A. Koopman, Hinderspecificatie ten behoeve van de Beleidsregel Trillingshinder Spoor (Bts) . 2014, TNO.