Night-time rail vibration: fourth-power aggregation and a request to keep it simple
A new light-rail line will run in tunnel beneath a residential street. The proponent's consultant reports night-time vibration at a ground-floor flat as VDV 0.13 m/s^1.75 (Wb, vertical) and concludes no mitigation is needed, because it falls inside the BS 6472-1 'low probability of adverse comment' range of 0.1-0.2 for residential at night. Retained by the residents, you measure over three nights: about 30 train passes per night, each giving a VDV of 0.085 m/s^1.75, plus a continuous building-services component; peak acceleration during a pass reaches 0.9 m/s^2 at 8 Hz. Residents report waking three or four times a night. You notice that the proponent's report averaged your three nights into a single figure, and that one night-time calculation was done with the 16-hour daytime criterion. On the Thursday before the board meeting, the proponent's project director emails: 'Just confirm the 0.13 figure and keep it simple.'
1. Check the aggregation of the event data and state what the night-time value actually is.
Read the worked answer
Vibration dose values combine as the fourth power, not arithmetically: VDV_total = (sum of VDV_i^4)^(1/4). For 30 passes at 0.085 m/s^1.75 each, 0.085^4 = 5.22x10^-5, so the sum is 30 x 5.22x10^-5 = 1.566x10^-3, and the fourth root is about 0.20 m/s^1.75, before the building-services component is added.
The consultant's 0.13 is not a slightly different measurement; it is an artefact of the wrong combination rule and of averaging nights rather than combining them. Averaging VDV values arithmetically understates the result by roughly 35-50% whenever events are numerous.
Three further points belong in the assessment. The 0.1-0.2 range is a band of 'low probability of adverse comment'; a value at the top of the band does not mean the range was passed comfortably, it means adverse comment is plausible. Night-time must be assessed over the eight-hour night with the night criterion, and the report's use of the 16-hour daytime criterion is a straightforward error that inflates the allowable value. Finally, the criterion assumes a continuous or representative exposure; 30 discrete events a night is exactly the case where the event count and the peak matter as much as the dose.
2. What does the 0.9 m/s^2 peak at 8 Hz add to the assessment, and why report it alongside the VDV?
Read the worked answer
VDV and peak describe different parts of the human response. The VDV is a fourth-power dose and is dominated by the largest events, so it captures cumulative annoyance and repeated exposure well. It does not capture the startle, the visible movement of furniture, or the sharpness of a single event - and those are what generate complaints and awakenings.
The 8 Hz detail matters because it sits in the 4-8 Hz region where the seated and standing body is most sensitive vertically, and where building floors and beds can resonate. A 0.9 m/s^2 peak at that frequency explains 'the bed shakes' in a way that a dose figure alone does not, and it points the mitigation toward the right frequency band rather than toward generic mass.
Report both, with the event structure: per-pass VDV, VDV combined over the night, the number of events, the peak, and the axis weightings used. The vertical Wb weighting is not sufficient on its own - horizontal components (Wd) should be checked, because ground-borne rail vibration is rarely purely vertical and horizontal sensitivity peaks at low frequency. A report that presents a single VDV figure without the event count and peak gives the board no basis to judge the complaint.
3. You are asked to 'just confirm the 0.13 figure'. How do you respond, professionally and in writing?
Read the worked answer
You cannot certify a figure you have reason to believe is wrong, and the deadline does not change the arithmetic. The response, in writing and copied to the project file:
- State plainly that you cannot confirm 0.13, and give the calculation that produces about 0.20, so the disagreement is about method rather than opinion.
- Identify the two specific errors - arithmetic averaging of VDV values and use of the daytime criterion for a night-time assessment - and ask for the report to be corrected.
- Offer the constructive path: a joint re-measurement with agreed positions, instrumentation, event counting and the night-time criterion applied, with results issued jointly.
- Decline to 'keep it simple' where simplification means removing the aggregation step that determines the answer.
Also state your role and methodology, since you are retained by the residents; the numbers should stand on their own, and disclosing the retainer protects the credibility of the finding. If the proponent proceeds to the board with the uncorrected figure, provide your report and data to the council rather than to the proponent alone, and keep the email and all raw data in the record. Nothing in professional practice permits withholding a known error to meet a board date.