The D.S. N°14/2024 sets descriptors and measurement periods for wind farms, but the emission level does not come from averaging fourteen days of data. It comes from deciding, on explicit criteria, which data enter the analysis. That decision — the post-processing of the data — is spelled out in the «Propuesta técnica para la elaboración de los protocolos de medición y predicción de niveles de emisión de ruido para la Norma de Emisión de Ruido para Fuentes Fijas», the technical proposal for drawing up the measurement and prediction protocols for the Noise Emission Standard for Fixed Sources established by Supreme Decree No. 14/2024, prepared by Chile’s Ministry of the Environment (MMA). The Superintendence of the Environment (SMA) has yet to publish the final protocol, but the post-processing provisions are likely to be very close to the proposal.
Fourteen days of measurement are thousands of data points, not a result
A continuous monitoring campaign at a receptor point yields one L90 value every ten minutes for at least fourteen days: 2,016 records at a single point. Among them sit daytime and night-time hours, rain, birds, farm work, local traffic and periods with the turbines shut down.
The wind farm’s emission level is not the average of all that. It is what remains once the farm’s operation is isolated and correlated with the wind speed at hub height, aiming for a value of R2 ≥ 0.6. Everything else has to come out of the analysis, and you have to be able to explain why it came out.
The curve is built from a scatter plot of L90 against wind speed with a fourth-order polynomial trend line. If the full data set already reaches the required R2, there is nothing left to filter.
The base case and the four filtering criteria, applied sequentially
The technical proposal for drawing up the protocol sets out four data filtering criteria, in addition to the base case:
- Base case. The presentation of the full data set, with no filtering applied.
- Operating status criterion. Discard the records in which the turbines closest to the point were shut down or generating zero or negative output. With no source running, the data point does not describe emission. The minimum number of resulting data points may not fall below 200 values.
- Anomalous event exclusion criterion. Using the audio recorded and synchronised across the whole campaign, filter out the stretches contaminated by events unrelated to the farm, and discard the periods with precipitation. The minimum number of resulting data points may not fall below 200 values.
- Time-of-day criterion. Restrict the daytime analysis to between 12:00 and 15:00, and the night-time analysis to between 00:00 and 04:00. The minimum number of resulting data points may not fall below 200 values.
- Wind direction criterion. Narrow the analysis to the data grouped in wind direction ranges, prioritising the downwind condition (propagation with the wind from the turbine towards the receptor). The minimum number of resulting data points may not fall below 100 values.
The order matters: the criteria are applied sequentially, and only as far as they are needed. And if after the last criterion R2 is still below 0.6 and the curve sits above the limit, the proposal opens the door to assessing compliance through prediction methods.
How to prepare
- Synchronise the clock on the measurement stations with the wind farm’s wind and generation records from the moment they are installed. Without that synchronisation there is no correlation analysis to be had.
- Record audio throughout the campaign, not only in samples. It is what lets you justify each exclusion afterwards.
- Leave a record of every discarded data point and the reason for it, rather than deleting it from the spreadsheet without a trace. Each criterion applied has its own spreadsheet and its own scatter plot.
- Produce a wind rose for the measurement period. Without it there is no way to argue that the filtered direction genuinely corresponds to the downwind condition.
- Watch the number of data points. Each criterion improves the correlation but shrinks the sample, and there is a floor: if the filtering leaves fewer values than the minimum, the criterion cannot be applied.
How we help
At Acústica Austral we design the monitoring campaign, the data processing and the calculation report with the traceability required, so you know whether a wind farm in operation will comply with the new noise standard as things stand today.
If you have a wind farm in operation, we can go over with you how the regulatory change might affect it: book a 20-minute call. You can also see what our acoustic optimisation of wind farms service involves.
He leads Acústica Austral’s environmental noise and vibration studies for the SEIA across energy, port and infrastructure projects, and has contributed to reference work for Chile’s Ministry of the Environment.
Meet the team →Free resources for your project
Guide to the D.S. N°14/2024 and the acoustic checklist for entering the SEIA, in PDF. We email them to you.