Journal of Sound and Vibration (1996) 193(1), 215–225

THE CALCULATION AND ASSESSMENT OF

GROUND-BORNE NOISE AND PERCEPTIBLE
VIBRATION FROM TRAINS IN TUNNELS
R. A. H, R. J. G, M. B  P. R. W
Ashdown Environmental Limited, The Oast House, Upper Hartfield, East Sussex,
TN7 4AR, England

(Received in final form 20 November 1995)

The assessment and calculation of perceptible vibration and ground-borne noise from
the operation of railway tunnels are discussed. The data and methodologies reported
include material obtained by the authors in studies of the operation and construction of
the proposed 108 km long high speed railway link between London and the Channel Tunnel
which include an 18 km long tunnel beneath residential areas in London.
7 1996 Academic Press Limited

1. INTRODUCTION
Ashdown Environmental Limited (AEL) carried out an independent assessment of the
noise and vibration effects associated with the operation and construction of the Channel
Tunnel Rail Link (CTRL). The assessment covered the construction and operation of the
new 108 km long high speed railway which will run between London and the Channel
Tunnel. In London the assessment included the construction and operation of an 18 km
long railway tunnel beneath residential areas.
This paper sets out our findings for both the assessment and calculation of perceptible
vibration and ground-borne noise from the operation of railway tunnels. In section 2 we
describe the assessment criteria that have been developed from extensive literature reviews
and AEL’s professional experience. Section 3 gives a brief description of the calculation
methodologies that were used for perceptible vibration and ground-borne noise from trains
in bored tunnels. The work in this paper is more fully reported in the CTRL Environmental
Statement [1] and other associated reports [2–4]. The calculation procedures continue to
be refined and further developments will be reported separately.†

2. EVALUATIVE CRITERIA
Environmental vibration, from sources such as rail traffic, is associated with two
principal manifestations: (i) ground-borne noise and (ii) perceptible vibration.
Ground-borne noise occurs when ground vibration gives rise to a vibration of building
surfaces which is transmitted in turn to the air inside the building in the form of audible
sound. Ground-borne noise is generally associated with ground vibration with appreciable
energy between 50 and 160 Hz. The assessment of this phenomenon was based on the
LAmax,S noise level evaluated near the centre of dwelling rooms as the result of the passage
of a single train.
† Several papers have been published since this presentation to the Fifth International Workshop on Railway
Noise.
215
0022–460X/96/210215 + 11 $18.00/0 7 1996 Academic Press Limited
216 . .   .
When the occupants of buildings experience vibration from environmental sources, their
comfort can be adversely affected. This phenomenon is termed perceptible vibration.
However, vibration may also affect the use of resources by interfering with the operation
of equipment or other activities. The unit used for the basis of assessment was the vertical
Vibration Dose Value (VDV), evaluated at the centre of the floor of interest.
Where vibration from the passage of underground trains is perceptible, either as
ground-borne noise or vibration, people may worry about damage to buildings. However,
vibration from railways is highly unlikely to result in building damage [5].

2.1.   - 

There are no appropriate British Standards or International Standards for the
assessment of ground-borne noise. The ground-borne noise criteria therefore were
principally derived from (i) the American Public Transit Association Guidelines for Design
of Rapid Transit Facilities (the APTA Guidelines) [6]; (ii) the Draft Guidance Manual
for Transit Noise and Vibration Impact Assessment prepared for the Urban Mass
Transportation Administration (UMTA), U.S. Department of Transportation [7];
(iii) London Transport complaints history data [8]; and (iv) professional experience.
The APTA Guidelines were published in 1981 and have been widely used in the United
States. The UMTA Draft Guidance, however, has not been published yet, but we
understand that it is scheduled to be published shortly.† It should be noted that the UMTA
Guidance is not intended to be a revision of the APTA Guidelines. However, it is
understood that the UMTA Guidance, which specifies impact criteria for ground-borne
noise, was developed in the United States upon the basis of experience of the application
of the APTA Guidelines, which specify ‘‘design goals’’.
From the above, assessment criteria for a wide range of resources have been developed,
and these are shown in Table 1.
The severity of impacts on residential resources was classified by the application of the
semantic scale shown in Table 2. If the levels predicted for a non-residential resource
exceeded the relevant criterion in Table 1, a more detailed assessment was undertaken to
determine the likelihood of a significant effect.

2.2.    

British Standard 6472 [9] offers guidance with regard to the assessment of perceptible
vibration. The approach adopted for the assessment of CTRL was to build upon the
principles in BS 6472. In particular, in accordance with the standard, the assessment was
based upon the use of Vibration Dose Value (VDV), a measure of the cumulative exposure
to vibration during the measurement period. However, the assessment went further than
the standard by (i) using a scale based upon the concept of a doubling of VDV giving rise
to a doubling of sensation, to evaluate changes in vibration level where residential
resources were identified as being subject to appreciable existing levels of vibration, and
(ii) providing a more detailed classification of resources.
The classification of resources developed for the assessment of perceptible vibration is
shown in Table 3. A semantic scale was developed to facilitate the description and ranking
of impacts on residential resources not subject to appreciable vibration prior to the new
railway, as shown in Table 4. Where the predicted magnitude of vibration from the new
railway for a non-residential resource exeeded the relevant criterion in Table 3, a more
detailed assessment was carried out.

† Final version published Summer 1995.

        217
T 1
Classifications for ground-borne noise impacts
Criterion
Classification LAmax,S (dB) Indicative examples of resources
Potentially highly sensitive 25 Radio studios
Television studios
Professional recording studios
Film studios
Concert halls
Audiological testing facilities
Potentially sensitive 35 Residential resources
Medical facilities
Research establishments
Laboratories
Editing suites and post-production facilities
Libraries
Places of worship
Law courts
Theatres
Office, retail, general community 40 Offices
or institutional Shops
Public halls
Exhibition halls
Dental surgeries
Museums
Art galleries
Community centres
Educational establishments
Cinemas
Rehearsal rooms
General industrial, manufacturing 50 Factories
or distribution Warehouses
Workshops
General industrial installations

For residential resources subject to appreciable existing vibration, defined as where the
baseline VDV from an existing external source is more than the relevant criterion in
Table 3, impacts due to changes in the predicted overall levels of vibration associated with
the opening of the new railway were classified according to the resultant percentage
increase in VDV. The categories used are shown in Table 5.

T 2
Classification of ground-borne noise impacts upon residential dwellings
LAmax,S (dB) Classification
35–39 Low
40–44 Medium
45–49 High
q49 Very high
218 . .   .
T 3
Resource classifications for vibration impacts
Criterion VDV (m s−1·75 )
ZXXXXXCXXXXXV Indicative examples
Classification Day Night of resources
Potentially highly sensitive 0·11 0·09 Medical facilities
Dental surgeries
Optician’s facilities
Hazardous installations
Laboratories
Research establishments
Telephone exchanges
Manufacturers using particularly
vibration-sensitive equipment
Services which require particularly
vibration-sensitive equipment
or are otherwise particularly
vibration-sensitive
Potentially sensitive 0·22 0·13 Residential resources
0·18 Radio studios
Television studios
Professional recording studios
Film studios
Editing suites and post-
production facilities
Rehabilitation centres
Libraries
Law courts
Theatres
Places of worship
Concert Halls
Office, retail, general 0·43 0·36 Offices
community or Shops
institutional Public halls
Exhibition halls
Museums
Art galleries
Community centres
Educational establishments
Cinemas
Rehearsal rooms
General industrial, 0·87 0·73 Factories
manufacturing or Warehouses
distribution Workshops
General industrial installations

3. CALCULATION METHODOLOGIES
Calculation procedures have been developed for the prediction of ground-borne noise
and perceptible vibration levels from trains on the surface, in bored tunnels, and in cut
and cover tunnels. In this paper only the predictors for ground-borne noise and perceptible
vibrations from trains in bored tunnels used for the assessment of the CTRL are described.
An empirical approach, based on the analysis of over 3000 measurements, was adopted.
This was considered preferable, as the theoretical models available at the time had not been
        219
T 4
Classification of perceptible vibration impacts upon residential resources not subject to
appreciable baseline vibration
Daytime Night-time
(07:00–23:00) (23:00–07:00)
16 hr VDV (m s−1·75 ) 8 hr VDV (m s−1·75 ) Classification
q0·22–0·31 q0·13–0·18 Slight
q0·31–0·44 q0·18–0·26 Moderate
q0·44–0·62 q0·26–0·37 Substantial
q0·62 q0·37 Severe
Note: use the higher of the two classifications where daytime and night-time impacts differ.

developed or validated for a high speed railway, and the engineering, lithological and other
necessary information are not available.
The two calculation procedures are summarized as flowcharts in Figures 1 and 2.
The calculation procedures consist of three stages; (i) source terms, (ii) propagation, and
(iii) building response.†

3.1.   

The source terms for the two procedures differ in their form (vibration dose value for
vibration calculation procedures and one-third octave root mean square (r.m.s.) particle
velocity for ground-borne noise) and in the corrections which have to be made (in terms
of train length and speed, for instance).
The source terms have been derived from measurements of vibration due to the passage
of TGVs on surface sections of ballasted track carried out in France. The source terms
relate to levels of vibration, 10 m from the nearest rail, for each of the four generic
classifications of lithology (sands, mixed sands and clays, and clays and chalks) to be found
along the alignment of the CTRL.
The surface source terms derived from measurements carried out on TGVs were adjusted
to reflect the source levels of other trains, such as the Eurostar and Networker Class 465.
The most important parameters of train design, with regard to levels of vibration, are
generally unsprung mass, speed and length.

† It should be noted that, following the completion of this paper, further research identified that the procedure
described for calculating perceptible vibration from bored tunnels is likely to result in significant over-estimation
of the vibration levels where non-ballasted track systems are used inside the tunnel. A revised, more accurate
model has since been developed [10].

T 5
Classification of perceptible vibration impacts upon resources experiencing appreciable
baseline vibration
Increase in VDV (%) Classification
25–40 Slight
q40–100 Moderate
q100–185 Substantial
q185 Severe
Note: calculate the percentage increase in VDV for both day and night for residential resources and use the
higher of the two classifications where daytime and night-time impacts differ.
220 . .   .

Figure 1. The calculation procedure for ground-borne noise from bored tunnelled sections of railway.
        221

Figure 2. The calculation procedure for perceptible vibration from bored tunnelled sections of railways.

Adjustments were made to the surface source terms for bored tunnelled sections on the
basis of a transfer between ballast surface track and PACT (continuously laid concrete
base with continuously supported track) in tunnels, derived from measurements carried
out at Stansted Tunnel in the U.K. Adjustments were made to the surface source terms
for cut and cover tunnelled sections which feature a slab base on the basis of an analysis
222 . .   .
of measurements carried out by Deutsche Bahn. Adjustments required to reflect the
different levels of mitigation were also made.

3.2. 
3.2.1. Ground-borne noise
The propagation model for ground-borne noise from bored tunnelled sections of railway
was derived from a statistical analysis of the results of measurements of ground-borne
noise and vibration from a variety of train types in tunnels, both in the U.K. and in France.
Separate aspects of the transmission path are addressed by three specific terms to
account for (i) absorption and geometric dispersion of the bulk waves from tunnel to
surface (a function of the depth from the rail head to the surface); (ii) absorption and
dispersion of the surface wave (a function of horizontal distance from the tunnel centre);
and (iii) the effect of tunnel width.
Analysis of the available data for ground-borne noise from trains in bored tunnels
indicated that differences in lithology do not have a major influence upon the propagation
characteristics. The procedure used for the calculation of propagation losses is accordingly
lithology independent.
3.2.2. Perceptible vibration
The propagation model for vibration from bored tunnelled sections of railway has been
derived from a statistical analysis of the results of measurements of vibration from a variety
of train types in tunnels, both in the U.K. and France.
Propagation losses are modelled as being governed by radial distance from the rail head.
Analysis of the available data for vibration from trains in tunnels indicated that differences
in lithology do not have a major influence upon the propagation characteristics. The
procedure for the calculation of propagation losses is accordingly lithology independent.
3.3.  
3.3.1. Ground-borne noise
Ground-borne noise levels (LAmax,S ) near the centre of ground floor rooms are calculated
from r.m.s. one-third octave band vertical particle velocities (evaluated for the period
during a train is passing) outside the building of interest. The conversion to internal
ground-borne noise levels is based on the equation proposed by Kurzweil [11], validated
and adjusted through an analysis of measurements carried out as part of a collaborative
study with British Rail and London Underground Limited. As a cautious (i.e., worst case),
estimate, it was assumed for high rise buildings that ground-borne noise levels decrease
by 1 dB(A) per floor above ground floor regardless of building construction.
3.3.2. Perceptible vibration
Analysis of measurements of vibration from trains carried out during collaborative work
with British Rail and London Underground Limited suggested that vertical VDVs
measured near the centre of wooden floors of ground floor rooms in brick-built residential
properties can be twice those measured on the ground immediately outside the property
of interest. The vertical VDV values for the first floor were found to be around four times
the level measured on the ground immediately outside.
The analysis of building response in the collaborative study was limited to brick-built
residential properties with wooden floors and strip foundations. It is considered that the
application of an approach, based upon analysis of data from brick-built residential
properties, to concrete high rise buildings gives rise to a cautious (i.e., worst case) estimate
of vibration levels.
        223

Figure 3. Comparisons with pseudo-measurements of predictions given by the calculation procedure of

Figure 1. Q, Sceaux tunnel, SNCF; *, Potters Bar tunnels, BR; r, Stansted tunnel, BR; q, Central Line, LUL;
——, mean fit regression line. Statistics: gradient 0·80; correlation coefficient 0·87; number of observations 205;
standard error 4·09.

For the purposes of the CTRL assessment therefore, vibration levels from the passage
of trains on the first floor and above were considered to be four times the free field level
outside the property.

3.4. 
An indication of the accuracy of the individual procedures can be judged from
Figures 3 and 4, which show predicted and measured values.
These figures show the results of the calculation procedures plotted against measured—
or in the case of ground-borne noise pseudo-measured—values (pseudo-measured
ground-borne noise values are calculated by applying the ground-borne noise building
response function to measured vibration values). A perfect model would result in all points
on these graphs lying upon a diagonal line (i.e., predicted = measured). However, it can
be seen that there is considerable inter-site and inter-train variability, so that the measured
results alone exhibit a degree of scatter.
In addition to presenting the scatter plots in Figures 3 and 4, a straight line regression
has been carried out to enable inferences to be drawn with regard to trends in prediction
errors. The slopes of these regressions are Q1, implying that the calculation procedures
tend to overestimate low levels (which may occur relatively distant from the track) while
underestimating high levels (which may occur relatively close to the track).

4. SUMMARY
Assessment criteria and calculation procedures have been developed for ground-borne
noise and perceptible vibration from trains in tunnels. The assessment methodology has
been successfully applied to a major rail project in the U.K. The assessment criteria are
based upon the best information presently available, and are robust and practicably
224 . .   .

Figure 4. Comparisons with measurements of predictions given by the calculation procedure of Figure 2.
w, Saltwood, BR; Q, Sceaux, SNCF; *, Potters Bar, BR; R, Camden Road, BR; r, Stansted, BR; q, Central
Line (three sites), LUL; ——, mean fit regression. Note that measured VDV is defined for one train pass-by.
Statistics (defined for log/log fit): gradient 0·70; correlation coefficient 0·77; number of observations 208; standard
error 0·23.

applicable for the purposes of environmental assessment. However, it is recognized that

there is a need for further work to be done with regard to evaluating community/subjective
response to ground-borne noise and vibration from underground railways. Likewise,
although we believe that the calculation procedures represent the best practicable means
for the prediction of ground-borne noise and vibration from railway tunnels, they are
presently undergoing further development to improve their accuracy and to ensure that
they can be applied to a wide range of projects.

ACKNOWLEDGMENTS
The development and application of the calculation and assessment methodologies were
carried out by AEL for Union Railways Limited. AEL gratefully acknowledge the help
and support provided by the staff of URL.

REFERENCES
1. U R L 1994 (November) Channel Tunnel Rail Link: Environmental
Statement.
2. A E L 1994 (November) Channel Tunnel Rail Link: Assessment of
Vibration Effects. Final Report.
3. A E L 1995 (February) Channel Tunnel Rail Link: Criteria for the
Assessment of Vibration for Railways. Final Report.
4. A E L 1995 (January and February) Channel Tunnel Rail Link:
The Groundborne Noise and Vibration Calculation Procedures for the Channel Tunnel Rail Link,
Volumes 1 and 2.
        225
5. B S I 1993 BS 7385: Part 2. Evaluation and Measurement for
Vibration in Buildings. Part 2: Guide to Damage Levels from Groundborne Vibration.
6. A P T A 1981 Guidelines for Design of Rapid Transit Facilities.
7. U S D  T F T A. Guidance
Manual for Transit Noise and Vibration Impact Assessment (unpublished).
8. O   S A, L T E 1979 (11 December)
Complaints of Railway Noise Received by London Transport 1969–1979.
9. B S I 1992 BS 6472. Guide to Evaluation of Human Exposure to
Vibration in Buildings (1 to 80 Hz).
10. A E L 1995 (June) Channel Tunnel Rail Link. Addendum to the
Groundborne Noise and Receptible Vibration Calculation Procedures for the Channel Tunnel
Rail Link.
11. L. G. K 1979 Journal of Sound and Vibration 66, 363–370. Ground-borne noise and
vibration from underground rail systems.