magnitude_ms2 string | trigger_time_min string | exposure_points string | a8_ms2 string | band string | exposure_points_display string | a8_display string |
|---|---|---|---|---|---|---|
1 | 5 | 0.17 | 0.102 | below-action | 0 | 0.10 |
1 | 10 | 0.33 | 0.144 | below-action | 0 | 0.14 |
1 | 15 | 0.50 | 0.177 | below-action | 1 | 0.18 |
1 | 20 | 0.67 | 0.204 | below-action | 1 | 0.20 |
1 | 30 | 1.00 | 0.250 | below-action | 1 | 0.25 |
1 | 45 | 1.50 | 0.306 | below-action | 2 | 0.31 |
1 | 60 | 2.00 | 0.354 | below-action | 2 | 0.35 |
1 | 90 | 3.00 | 0.433 | below-action | 3 | 0.43 |
1 | 120 | 4.00 | 0.500 | below-action | 4 | 0.50 |
1 | 150 | 5.00 | 0.559 | below-action | 5 | 0.56 |
1 | 180 | 6.00 | 0.612 | below-action | 6 | 0.61 |
1 | 240 | 8.00 | 0.707 | below-action | 8 | 0.71 |
1 | 300 | 10.00 | 0.791 | below-action | 10 | 0.79 |
1 | 360 | 12.00 | 0.866 | below-action | 12 | 0.87 |
1 | 420 | 14.00 | 0.935 | below-action | 14 | 0.94 |
1 | 480 | 16.00 | 1.000 | below-action | 16 | 1.00 |
1.5 | 5 | 0.38 | 0.153 | below-action | 0 | 0.15 |
1.5 | 10 | 0.75 | 0.217 | below-action | 1 | 0.22 |
1.5 | 15 | 1.12 | 0.265 | below-action | 1 | 0.27 |
1.5 | 20 | 1.50 | 0.306 | below-action | 2 | 0.31 |
1.5 | 30 | 2.25 | 0.375 | below-action | 2 | 0.38 |
1.5 | 45 | 3.38 | 0.459 | below-action | 3 | 0.46 |
1.5 | 60 | 4.50 | 0.530 | below-action | 5 | 0.53 |
1.5 | 90 | 6.75 | 0.650 | below-action | 7 | 0.65 |
1.5 | 120 | 9.00 | 0.750 | below-action | 9 | 0.75 |
1.5 | 150 | 11.25 | 0.839 | below-action | 11 | 0.84 |
1.5 | 180 | 13.50 | 0.919 | below-action | 14 | 0.92 |
1.5 | 240 | 18.00 | 1.061 | below-action | 18 | 1.06 |
1.5 | 300 | 22.50 | 1.186 | below-action | 23 | 1.19 |
1.5 | 360 | 27.00 | 1.299 | below-action | 27 | 1.30 |
1.5 | 420 | 31.50 | 1.403 | below-action | 32 | 1.40 |
1.5 | 480 | 36.00 | 1.500 | below-action | 36 | 1.50 |
2 | 5 | 0.67 | 0.204 | below-action | 1 | 0.20 |
2 | 10 | 1.33 | 0.289 | below-action | 1 | 0.29 |
2 | 15 | 2.00 | 0.354 | below-action | 2 | 0.35 |
2 | 20 | 2.67 | 0.408 | below-action | 3 | 0.41 |
2 | 30 | 4.00 | 0.500 | below-action | 4 | 0.50 |
2 | 45 | 6.00 | 0.612 | below-action | 6 | 0.61 |
2 | 60 | 8.00 | 0.707 | below-action | 8 | 0.71 |
2 | 90 | 12.00 | 0.866 | below-action | 12 | 0.87 |
2 | 120 | 16.00 | 1.000 | below-action | 16 | 1.00 |
2 | 150 | 20.00 | 1.118 | below-action | 20 | 1.12 |
2 | 180 | 24.00 | 1.225 | below-action | 24 | 1.22 |
2 | 240 | 32.00 | 1.414 | below-action | 32 | 1.41 |
2 | 300 | 40.00 | 1.581 | below-action | 40 | 1.58 |
2 | 360 | 48.00 | 1.732 | below-action | 48 | 1.73 |
2 | 420 | 56.00 | 1.871 | below-action | 56 | 1.87 |
2 | 480 | 64.00 | 2.000 | below-action | 64 | 2.00 |
2.5 | 5 | 1.04 | 0.255 | below-action | 1 | 0.26 |
2.5 | 10 | 2.08 | 0.361 | below-action | 2 | 0.36 |
2.5 | 15 | 3.12 | 0.442 | below-action | 3 | 0.44 |
2.5 | 20 | 4.17 | 0.510 | below-action | 4 | 0.51 |
2.5 | 30 | 6.25 | 0.625 | below-action | 6 | 0.63 |
2.5 | 45 | 9.38 | 0.765 | below-action | 9 | 0.77 |
2.5 | 60 | 12.50 | 0.884 | below-action | 13 | 0.88 |
2.5 | 90 | 18.75 | 1.083 | below-action | 19 | 1.08 |
2.5 | 120 | 25.00 | 1.250 | below-action | 25 | 1.25 |
2.5 | 150 | 31.25 | 1.398 | below-action | 31 | 1.40 |
2.5 | 180 | 37.50 | 1.531 | below-action | 38 | 1.53 |
2.5 | 240 | 50.00 | 1.768 | below-action | 50 | 1.77 |
2.5 | 300 | 62.50 | 1.976 | below-action | 63 | 1.98 |
2.5 | 360 | 75.00 | 2.165 | below-action | 75 | 2.17 |
2.5 | 420 | 87.50 | 2.339 | below-action | 88 | 2.34 |
2.5 | 480 | 100.00 | 2.500 | above-action | 100 | 2.50 |
3 | 5 | 1.50 | 0.306 | below-action | 2 | 0.31 |
3 | 10 | 3.00 | 0.433 | below-action | 3 | 0.43 |
3 | 15 | 4.50 | 0.530 | below-action | 5 | 0.53 |
3 | 20 | 6.00 | 0.612 | below-action | 6 | 0.61 |
3 | 30 | 9.00 | 0.750 | below-action | 9 | 0.75 |
3 | 45 | 13.50 | 0.919 | below-action | 14 | 0.92 |
3 | 60 | 18.00 | 1.061 | below-action | 18 | 1.06 |
3 | 90 | 27.00 | 1.299 | below-action | 27 | 1.30 |
3 | 120 | 36.00 | 1.500 | below-action | 36 | 1.50 |
3 | 150 | 45.00 | 1.677 | below-action | 45 | 1.68 |
3 | 180 | 54.00 | 1.837 | below-action | 54 | 1.84 |
3 | 240 | 72.00 | 2.121 | below-action | 72 | 2.12 |
3 | 300 | 90.00 | 2.372 | below-action | 90 | 2.37 |
3 | 360 | 108.00 | 2.598 | above-action | 108 | 2.60 |
3 | 420 | 126.00 | 2.806 | above-action | 126 | 2.81 |
3 | 480 | 144.00 | 3.000 | above-action | 144 | 3.00 |
3.5 | 5 | 2.04 | 0.357 | below-action | 2 | 0.36 |
3.5 | 10 | 4.08 | 0.505 | below-action | 4 | 0.51 |
3.5 | 15 | 6.12 | 0.619 | below-action | 6 | 0.62 |
3.5 | 20 | 8.17 | 0.714 | below-action | 8 | 0.71 |
3.5 | 30 | 12.25 | 0.875 | below-action | 12 | 0.88 |
3.5 | 45 | 18.38 | 1.072 | below-action | 18 | 1.07 |
3.5 | 60 | 24.50 | 1.237 | below-action | 25 | 1.24 |
3.5 | 90 | 36.75 | 1.516 | below-action | 37 | 1.52 |
3.5 | 120 | 49.00 | 1.750 | below-action | 49 | 1.75 |
3.5 | 150 | 61.25 | 1.957 | below-action | 61 | 1.96 |
3.5 | 180 | 73.50 | 2.143 | below-action | 74 | 2.14 |
3.5 | 240 | 98.00 | 2.475 | below-action | 98 | 2.47 |
3.5 | 300 | 122.50 | 2.767 | above-action | 123 | 2.77 |
3.5 | 360 | 147.00 | 3.031 | above-action | 147 | 3.03 |
3.5 | 420 | 171.50 | 3.274 | above-action | 172 | 3.27 |
3.5 | 480 | 196.00 | 3.500 | above-action | 196 | 3.50 |
4 | 5 | 2.67 | 0.408 | below-action | 3 | 0.41 |
4 | 10 | 5.33 | 0.577 | below-action | 5 | 0.58 |
4 | 15 | 8.00 | 0.707 | below-action | 8 | 0.71 |
4 | 20 | 10.67 | 0.816 | below-action | 11 | 0.82 |
HAVS and noise exposure tables by ComplyOnSite
Ready-made tables of daily hand-arm vibration (HAVS) and noise exposure for Great Britain, in CSV and JSON. Look up a tool's vibration magnitude or a task's noise level against how long someone is exposed, and read off the exposure points, the daily exposure and how long it takes to reach each action and limit value.
They are for supervisors, health and safety advisers, trainers and developers who need HSE's arithmetic in a spreadsheet, a toolbox talk or their own software. Every figure comes from the formulas in the regulations and HSE's guidance, and equals what the ComplyOnSite HAVS calculator and noise exposure calculator show for the same inputs.
| File | Rows | What it holds |
|---|---|---|
| havs-exposure-points.csv / .json | 656 | 41 vibration magnitudes (1 to 40 m/s²) × 16 daily trigger times (5 min to 8 h): exposure points and A(8) |
| havs-time-to-action-values.csv / .json | 41 | Points per hour and time to reach the EAV (2.5 m/s² A(8), 100 points) and ELV (5 m/s² A(8), 400 points) for each magnitude |
| noise-daily-exposure.csv / .json | 828 | 46 noise levels (75 to 120 dB(A) LAeq) × 18 durations (2 min to 12 h): daily exposure LEP,d and exposure points |
| noise-time-to-action-values.csv / .json | 46 | Points per hour and time to reach 80, 85 and 87 dB(A) LEP,d for each level |
- Method: formulas, sources with page numbers, how the tables are built and checked
- Validation: the latest output of
python3 validate.py - Licence and citation
- Checksums: SHA-256 for the eight data files
- sources/: HSE's printed ready-reckoners as transcribed, and the calculator-code output the tables are built from
Made by ComplyOnSite. See all the free tools at complyonsite.com/tools.
The rules in one place
Hand-arm vibration (Control of Vibration at Work Regulations 2005, regulation 4(1) and Schedule 1; HSE L140 Appendix 1):
- A(8) = a × √(T / 8 h), where a is the vibration magnitude in m/s² and T the daily trigger time.
- Exposure points = 2 × a² × T (T in hours). 100 points is the exposure action value (EAV, 2.5 m/s² A(8)); 400 points is the exposure limit value (ELV, 5 m/s² A(8)).
- Time to reach the EAV = 8 h × (2.5 / a)²; to reach the ELV = 8 h × (5 / a)².
- Points from several tools add up; A(8) values do not (combine those as the square root of the sum of squares).
Noise (Control of Noise at Work Regulations 2005, regulation 4 and Schedule 1; HSE L108 Appendix 3):
- LEP,d = LAeq + 10 log10(T / 8 h).
- Exposure points = 100 × (T / 8 h) × 10^((LAeq − 85) / 10). 100 points is 85 dB(A); about 32 points is 80 dB(A); about 158 points is 87 dB(A). HSE's daily ready-reckoner uses these points.
- Time to reach a target level = 8 h × 10^((target − LAeq) / 10).
- The 87 dB(A) exposure limit value takes account of hearing protection (regulation 4(5)); the 80 and 85 dB(A) action values do not. So
minutes_to_87andreaches_87only tell you about the limit when the level you look up is the level at the ear with the protector worn. Looked up with the unprotected workplace level, they show when the unprotected day reaches 87 dB(A), not whether the limit is breached.
Fields
havs-exposure-points
| Field | Meaning |
|---|---|
magnitude_ms2 |
Vibration magnitude a_hv, m/s² (frequency-weighted, vector sum) |
trigger_time_min |
Daily trigger time: minutes the tool is actually running in the hands |
exposure_points |
2 × a² × hours, to 2 decimal places |
a8_ms2 |
Daily exposure A(8) in m/s², to 3 decimal places |
band |
As the calculator classes it: below-action under 100 points; above-action at or above 100 and under 400; above-limit at or above 400. Decided on the unrounded value |
exposure_points_display, a8_display |
The figures exactly as the ComplyOnSite HAVS calculator prints them: whole points and 2-decimal A(8), with an extra place only where plain rounding would land on the other side of 100/400 points or 2.5/5 m/s² |
havs-time-to-action-values
| Field | Meaning |
|---|---|
magnitude_ms2 |
Vibration magnitude, m/s² |
points_per_hour |
2 × a² |
minutes_to_eav, minutes_to_elv |
Trigger time on this tool alone to reach 100 and 400 points, to 2 decimal places |
eav_display, elv_display |
As the calculator words it: rounded to the nearest minute, or Over 8 hours |
noise-daily-exposure
| Field | Meaning |
|---|---|
laeq_dba |
Noise level over the task, LAeq in dB(A) |
duration_min |
Minutes exposed at that level in the day |
lepd_dba |
Daily personal noise exposure LEP,d in dB(A), to 2 decimal places |
exposure_points |
L108 exposure points, to 2 decimal places |
band |
As the calculator classes it: below under 80 dB(A); lower from 80 to under 85; upper at 85 or above. Decided on the unrounded value |
reaches_87 |
LEP,d is 87 dB(A) or more. See the hearing-protection note above |
lepd_display, exposure_points_display |
The figures exactly as the ComplyOnSite noise calculator prints them: whole numbers, with a decimal only where rounding would cross 80, 85 or 87 dB(A) or the matching points |
noise-time-to-action-values
| Field | Meaning |
|---|---|
laeq_dba |
Noise level, LAeq in dB(A) |
points_per_hour |
Exposure points per hour at that level |
minutes_to_80, minutes_to_85, minutes_to_87 |
Minutes at this level alone before LEP,d reaches 80, 85 and 87 dB(A), to 2 decimal places |
display_80, display_85, display_87 |
As the noise calculator words it: rounded down to the whole minute, under a minute, or 24 hours or more |
The JSON files hold the same records, with numbers as numbers, after a block naming the sources and the calculator code they came from.
Checks
python3 validate.py (Python 3, standard library only) recomputes every row from the published formulas, independently of ComplyOnSite's code, and compares:
- all 1,571 rows against the regulations' Schedule 1 formulas and L140/L108 equations;
- all 229 cells of HSE's HAVS ready-reckoner (L140 Table 4) and all 173 cells of HSE's daily noise ready-reckoner (L108 Table 10);
- ten worked-example figures from L140 and L108, and the formulas in HSE's own spreadsheet calculators;
- the two figures the live calculators are known to show: 327 points / 4.52 m/s² A(8) for a breaker (14 m/s², 20 minutes) plus a chainsaw (7 m/s², 2 hours), and 86.6 dB(A) for L108's Box 4 day;
- with
COMPLYONSITE_SOURCE_ROOTset to a ComplyOnSite checkout, it also re-runs the calculator code and confirms every number and display string is unchanged.
What HSE's own tables get wrong
HSE's printed HAVS ready-reckoner (L140, page 55) disagrees with HSE's own formula in 8 of its 229 cells. The first three below are out by 23 to 200 points; the rest by 21 points or less:
| Magnitude × time | L140 Table 4 prints | Formula (and this dataset) |
|---|---|---|
| 25 m/s² × 30 min | 825 | 625 |
| 7 m/s² × 6 h | 500 | 588 |
| 6 m/s² × 6 h | 455 | 432 |
| 12 m/s² × 3 h | 885 | 864 |
| 9 m/s² × 4 h | 660 | 648 |
| 3.5, 4.5, 5.5 m/s² × 30 min | 13, 21, 31 | 12.25, 20.25, 30.25 |
None of them moves a cell across 100 or 400 points, or into a different colour band, so the table's conclusions stay right; the numbers do not. If you add points from that table, use this dataset's figures instead. This dataset follows the formula, which is what the regulations define and what HSE's spreadsheet calculator uses.
HSE's noise ready-reckoner (L108 Table 10) rounds points to preferred numbers rather than to the nearest whole point, so it differs from the formula by up to 0.27 dB. That is by design, not an error.
Limits
These tables do the arithmetic only. They do not measure anything or tell you what a tool or task really produces. Use a magnitude or noise level that represents the work as it is done, from measurement, the manufacturer or HSE's typical values, and treat the result as an estimate. Being below an action value does not end the duty to reduce exposure as far as reasonably practicable.
Each row covers one tool or one noise level for the day. For several, add the exposure points across rows and convert back, or use the ComplyOnSite calculators. Noise peak sound pressure (135, 137 and 140 dB(C)) and weekly exposure are not covered.
The tables reflect the regulations and guidance as read on 3 October 2026. Check the sources before relying on them for a legal decision.
Hugging Face packaging
This upload is a byte-preserving copy of the data and source evidence at GitHub. The upstream README is retained unchanged in UPSTREAM_README.md.
Each CSV table has an explicit configuration and all columns load as strings. This preserves leading zeroes, CAS and industry codes, provisional year labels, printed decimal precision and calculator display values. Parse numeric or Boolean columns explicitly in downstream analysis; empty CSV cells remain missing. JSON retains the upstream numeric, Boolean, list and null types.
train is the Hub table-viewer split name; these are reference tables, not machine-learning training data. The OGL v3 licence is named explicitly through license: other; no Creative Commons relicensing is implied.
- Downloads last month
- 139