Appendix 2 – Noise
2.1
HS2 Phase One Environmental Statement Consultation
Individual Response by Michael Woodhouse
Little Acre, Ingestre, Stafford ST18 0RE
e-mail:
Introduction
I am responding only to Volume 5, Appendix SV-001-000, Annex D2: Operational Assessment – Airborne Sound
I am a retired professional engineer. My wife and I are residential home owners, livingclose to (but not directly on) theproposed route of HS2 Phase Two. We both oppose HS2 on principle and believe it is an inappropriate use of tax-payers' money. HS2 should not be built. However, should it proceed, Parliament must ensure that the impact on the environment is as low as possible.
Our home isapproximately 490 metres from the proposed Birmingham-to-Manchester leg of the route, at a point where the track is raised on a 13 metre-high embankment
Because Ingestre is located in tranquil estate parkland and all the residential properties in the parish (except two) lie within 1km of the proposed route, noise has been a particular concern to me. As a result, I have taken a close interest in the assessment of noise from HS2; initially as outlined in the documentation for Phase Two but, following completion of the Phase Two Route consultation at the end of January 2014, also to the more detailed description contained in Annex D2 of Appendix SV-001-000 of the Environmental Statement (ES) for Phase One (Volume 5).
Although I have spent only a short time looking at the Phase One ES, my findings lead me to conclude that HS2 Ltd has made questionable and in some cases false assumptions about noise. This means that Parliament, and the public-at-large, have been presented with an overly optimistic view of the environmental impact of noise on affected communities.
Because problems exist at the fundamental level of noise source representation, the impacts are route wide. It is essential that these deficiencies are highlighted in the summary report to Parliament to be prepared by Dialogue by Design. This should be accompanied by a recommendation for Parliament to initiate an independent expert review of HS2 Ltd's assessment of the noise impacts of HS2.
The total response (including this one) is 10 pages long.
1. Preamble
Noise is a complex and highly technical subject. I have reservations about the noise limits used by HS2 Ltd to indicate the likely impact of HS2 but, in the time available to me, I have chosen to examine, in detail, only the sound source model and propagation characteristics that have been used for the route-wide studies.
As will be appreciated, discrepancy at the level of noise source representation will produce erroneous and misleading results for every affected property and community along the whole route.
Given the essentially non-technical audience to whom the Dialogue by Design report will be addressed, I will highlight only the most blatant examples of where I believe the methods described in Appendix SV-001-000, Annex D2, are questionable so that Parliament may be alerted to the more general issue of model validation and the need for a full, independent, review. Should such a review be initiated, I will be happy to make a more detailed submission to the reviewing authority.
2. Representation of sound sources
HS2 has to comply with the EU Technical Specification for Interoperability (TSI) – see ref [14] cited in Annex D2, 1.1.31. The TSI sets maximum values for overall noise (which are lower for high speed trains ordered after 2010) but makes no judgement on what the individual contributions are to the total. This is something that HS2 Ltd must do as a prelude to carrying out environmental noise studies.
There are three recognised main contributors to railway noise: traction noise, rolling noise and aerodynamic noise. Traction noise dominates at low speed, rolling noise dominates at medium speed and aerodynamic noise at high speed. Figure 1 illustrates the typical speed dependencies.
Figure 1: Typical relative dependence of railway noise sources
(taken from ref [27] of 1.1.40 of Annex D2)
In Annex D2, individual sound sources are developed that, for a TSI-compliant train, sum to the TSI noise limits (post 2010 values) at the corresponding speeds. The aerodynamic component has been divided (Annex D2, 1.1.27) between body aerodynamic noise which is represented as a noise source 0.5m above the rail-head and two sources higher up on the train: pantograph noise at 5.0 m above rail-head and pantograph recess noise at 4.0m above rail-head. I believe, the last item has been incorrectly described and has resulted in a false deduction being made (see 3.2 below).
Aerodynamic noise is strongly dependent on the train speed, the peak value rising as the 7th power of the speed. Additionally, the split assumed between the aerodynamic noise sources at 0.5m and at 4m/5m is critical, as the former can be shielded by noise barriers whereas the latter will not be (a maximum noise barrier height of 3m has been assumed by HS2 Ltd in all noise calculations).
Trains ordered for HS2 are assumed will be specified to have a lower noise level than the TSI requirements. The greatest reduction (5dB) is made to the high-up noise sources, assuming the adoption of low-noise pantograph technology from east Asia (Annex D2, 1.1.53). I have doubts about the validity of this assumption as discussed in section 3 below.
The train speed above which aerodynamic noise becomes dominant is widely accepted to be around 300kph. Because HS2 will operate at up 360kph and potentially to the design maximum speed of 400kph, it is important that the dominance of aerodynamic noise above 300kph is taken into account, something that was not required when HS1 was being designed (HS1 has a maximum speed of 300kph).
Much is said in Annex D2 about how the models have been developed to do this but no attempt has apparently been made to validate the result against the basic premise.
As a simple test of consistency, I sought to check if the noise source models developed by HS2 Ltd gave a transition speed of 300kph for the TSI-compliant train. It did not. From 1.1.29 and 1.1.52 of Annex D2 the following values for contribution to the maximum sound level, LpAF,max, relevant to this issue, are:
Rolling noise assumed by HS2 Ltd is: (16.6 + 30logV)dB . . . . [1]
Body aerodynamic noise assumed by HS2 Ltd is: (–85.5 + 70logV)dB . . . . [2]
Pantograph aerodynamic noise assumed by HS2 Ltd is: (–92.3 + 70logV)dB . . . . [3]
…. where V is velocity in kph and log is base 10.
From the above, it can be shown that the value of V at which the logarithmic sum of [2] and [3] exceeds [1] is just under 345kph. This is well above the accepted value of 300kph.
The model therefore appears inconsistent and immediately raises the suspicion that, even before the benefits of east Asian low noise pantographs are assumed, the split of contributing noise sources for TSI-compliant trains is biased towards rolling noise, thus understating the aerodynamic component.
Furthermore, the aerodynamic component of noise has apparently been split so as to understate the component from the pantograph, which cannot be shielded by noise barriers, in favour of the body aerodynamic component, which can be shielded. The work cited in 1.1.39 of Annex D2 indicates measured values for the pantograph noise at 320kph of 85 – 87dBA, yet HS2 Ltd has used 83dBA (1.1.49). On examination, it appears that this lower value is a self-fulfilling outcome of using HS2 Ltd's own (optimistic?) model for pantograph noise!! It thus appears that HS2 Ltd has already squeezed 3dB out of the pantograph noise level even before applying further reduction on the back of the work carried out in east Asia (Japan).
3) Claimed reductions in pantograph noise for HS2 trains.
The noise source levels for HS2 trains have been set lower than the limits applicable to TSI-compliant trains. The reductions used are listed in Tables 2 and 3 of 1.1.54 of Annex D2, together with an indication of the assumed design improvements believed capable of giving rise to these reductions. All the stated assumptions can be challenged but I want to focus on two:
3.1 Pantograph Noise
A reduction of 5dB is assumed relative to European pantographs (on top of the apparent 3dB understatement as noted in section 2) above.
The Japanese have spent considerable time and money in investigating noise reduction measures for their high-speed trains, particularly pantograph noise, as this is the dominant source at high speeds. Annex D2 does not cite any work that compares a current European design of pantograph with that of a current Japanese design of pantograph to see what is potentially achievable. It is therefore not clear that the improvements achieved in Japan are necessarily attainable for a HS2 train.
The situation is complicated by the fact that the improvements made in Japan result from the combination of two principle measures: the streamlining of the pantograph itself and the addition of train-mounted noise insulation barriers. Figure 2 shows the current state-of-the-art configuration as used on the Shinkansen E5 bullet train (maximum speed 320kph).
Figure 2: Shinkansen E5 Pantograph with noise barrier
The Shinkansen E5 is the result of development carried out for the FASTECH360S high speed train[1], later proven by tests at full-scale[2]. Fig 3 shows the FASTECH360S pantograph with shield.
Figure 3: FASTECH360S showing Pantograph with noise barrier
When considering the pantograph noise component, the key finding of the Japanese work is that the noise insulation barriers are the most effective element, achieving around 4dB reduction compared with around 2dB for the streamlined pantograph arm.
While the pantograph arm will almost certainly be transferable to a European situation, it is not at all clear that noise insulation panels of the appropriate size and position on the train body can be accommodated within the European loading gauge UIC GC limits, to which all trains operating over the HS2 track have to comply (see Figure 4 below).
Unless noise insulation panels can be accommodated, within GC limits, in a manner that is functionally equivalent to those on the Japanese trains, the assumption of a 5dB reduction in pantograph noise for HS2 trains relative to TSI-compatible trains looks optimistic.
Figure 4: UIC Railway loading gauge profiles (GC applies to HS2)
3.2 Recess Noise
Each train unit (200m long for HS2) has two pantographs, one raised and one lowered. Which one is raised depends on the direction of travel and is usually the trailing one.
In addition to a noise source for the raised pantograph, P(pantograph), at 5m height above rail-head, HS2 Ltd has used an aerodynamic noise source, at 4m above rail-head, called “recess noise”.The term “recess” is misleading since the noise source to which it refers corresponds, in reality, to the second, lowered, pantograph. Because European TGV-type trains retract the lowered pantograph into a recess within the roof of the train, this has led to the term “recess noise” being used synonymously with “lowered pantograph noise”.
The first point to make is that, for TSI-compliant trains, HS2 Ltd has assumed that the noise from a raised pantograph, P(pantograph), is the same as for a lowered pantograph, P(recess) – see Table 1 in 1.1.52. This is at variance with the observation, made in 1.1.39 of Annex D2, that Gautier et al (footnote [25] of 1.1.39) showed that the noise from the lowered pantograph (recess noise), at 320kph, was 2dB greater than that from the raised pantograph.
The modern Japanese Shinkansen trains use roof-mounted pantographs, without recess, in combination with noise barriers, as discussed in 3.1 above. HS2 Ltd has assumed that a similar arrangement can be used for HS2 trains (Table 2 of 1.1.54 refers) which will result in an arrangement generally as illustrated in Figure 5.
Figure 5: FASTECH360S showing pantograph positions (from Yamada et al[3])
When pantographs are directly roof mounted, as suggested in 1.1.54, Table 1, there will still be a minimum of two pantographs per 200m train unit, one of which will be raised and one lowered. There will thus always be a noise contribution from both pantographs, whether or not there is a “recess”. It is therefore wrong for HS2 to eliminate “recess” noise from the source model for HS2 trains.
All the values for HS2 trains assumed in Table 3 of 1.1.54 are suspect but the “N/A” for P(recess) is technically wrong and will result in all noise predictions for HS2 trains being incorrect.
NB 1: A lowered roof-mounted pantograph will be at a greater height above the rail-head than one that retracts to within the train body. As a result, the noise source height for a lowered pantograph on a HS2 train, with roof-mounted pantographs, is likely to higher than the one assumed by HS2 Ltd for “recess” noise on a TSI-compliant train. In addition to adding a noise source to the HS2 train model, to represent the lowered pantograph, it will be necessary to assign a different noise source height to this term.
NB 2: The work by Ido et al[4] shows that, with the Japanese arrangement, the shielding effect of the noise insulation barriers is greater for the lowered pantograph (as might be expected), making the raised pantograph the larger noise source. This is the opposite to that observed by Gautier et al for TGV (see reference earlier in this section).
NB 3: The text header within Table 3 of 1.1.54, incorrectly states “Values for TSI-Compliant Trains at 25m” whereas it should read “Values for HS2 Trains at 25m”.
4) Missing Sound Source
Conspicuous by its absence is any treatment of noise impacts from structures, especially viaducts.
Figure 6: Structure radiated noise
Figure 6 (taken from HS2 Ltd Phase Two Sustainability Statement, Appendix E6) illustrates the phenomenon.
Many people live near viaducts yet nothing is said in Annex D2 about airborne noise from them.
The elevated nature of the track, the fact that slab track will be used and not ballasted track, that only low-height noise barriers can be used and, of course, the fact that the structure itself will be a noise source in its own right all appear to be ignored. The sound source models developed in Annex D2 will not be appropriate and will give misleading results if applied to any section of the above-ground route that involves a viaduct.
As a minimum, the rolling noise component must be increased (the rolling noise from slab track is typically 4dB greater than from ballasted track) and a new noise source added to represent the direct sound emissions of the structure itself.
5) Inconsistencies and misleading statements
Annex D2 contains many inconsistencies and misleading statements. Four that I single out to mention are as follows:
5.1 Justification for using HS1 modelling method: Clause 1.1.18 makes the claim that: “measurements have shown that [the HS1 method] provided an overestimate of actual in-service sound levels” thus conveying the impression that the HS1 method is conservative. This claim is repeated in 1.3.3 but is not supported by the measured vs predicted results depicted in figure 7. For both SEL and LAmax, the median value of the measured vs calculated plot shows that, on average, there is an overestimation only at the lower noise values, with the trend reversing to an underestimation at the higher values. If the validity of the claim hinges on the words “actual in-service sound levels” (for HS1) this is doubly misleading. This is because, even if HS1 in-service noise levels are generally in the lower part of the range (where an overestimation is more likely), HS2, because of its much higher speed, will be towards the upper end where the HS1 model is more likely to underestimate the real effect.
5.2 Claimed accuracy of HS1 modelling method: The unnumbered first paragraph of section 1.3 (presumably intended to be 1.3.1) makes the claim, for the HS1 method, that: “the difference between predicted and measured levels is typically within ±3dBA”, thus conveying the impression that it conforms with conventionally accepted tolerance for such studies. As with 5.1 above, this is not supported by figure 7. Figure 7 shows a spread between the 5th percentile and the 95th percentile lines of 9dB. This gives a nominal accuracy of ±4.5dB and, even so, 10% of all results will be outside this range. The claim for ± 3dB accuracy is highly misleading and can only be true if an extremely liberal interpretation has been placed on the word “typically”.
5.3 Sound propagation model: Clause 1.3.14, through to the end of Annex D2, deals with outdoor sound propagation. The described empirically-derived terms (1.3.16) are, presumably those developed by Hood et al, as referenced in 1.1.8, based on TGV pass-by measurements made in 1989/90 (1.3.23). I do not have the knowledge to challenge the basic equations given in 1.3.16 except to note that the final term, representing ground absorption, is shown as a train speed dependent term (in mph!) which certainly does not reflect any known physical effect that I am aware of. From other reading, I understand ground attenuation to be strongly dependent on the ground surface condition: from very soft (eg freshly fallen snow = high sound absorption) through to very hard (eg paved surfaces = low sound absorption) and to the frequency content of the sound source, with high frequencies being attenuated much more rapidly with distance than low frequencies. Neither the type of ground cover nor the frequency content of the source are accounted for in the stated expression. More generally, I would be much happier if HS2 were using a modern physics-based sound propagation model rather than some empirical rules derived from field measurements of a first generation TGV, undertaken a quarter of a century ago.
5.4 Inconsistency re ground attenuation: As noted in 5.3 above, the stated sound propagation model is empirically derived and contains a ground attenuation term that is independent of ground condition. On the other hand, the statement is made, in 1.1.10, that the sound propagation model includes (amongst other things) “ground cover types”. If the underlying equations do not include a “ground cover type” term, how can the predictions take it into account?
6) Noise Mitigation
While much of Annex D2 is devoted to describing the development of the models for representing the noise sources, there is no section describing the development of the models representing noise mitigation measures (barriers and earth bunds etc).
Almost as an afterthought, there is a graph (Figure 11) slipped into section 1.3 (Limitations and Sensitivity Tests) showing the speed-dependence of noise attenuation for a 3m high barrier for both TSI-compliant and HS2 trains (1.3.12 refers).