Publicly available data and open-source software for environmental noise and vibration calculations 

by Arnaud Kok & Rob van Loon, Dutch National Institute for Public Health and the Environment 

Introduction 

Determination of noise or vibration levels by calculations requires three ingredients: a set of calculation instructions and specifications (a model), software that implements these instructions, and input data. With the expert center on noise, the Dutch national institute for public health and the environment (RIVM) works on all three ingredients. We maintain the Dutch national calculation methods; we collaborate with partners to make data suitable for noise calculations and finally,y we produce software with which these calculations can be done. This article is about the power of using open source/publicly available data in conjunction with open-source software.  

Data 

There is a strong push in the Netherlands to make data gathered by order of the government publicly available [1]. This includes data from the land register and topography maps in high detail. There is a 2D digital copy of basically every structure in the Netherlands (including bike sheds). Aerial photographs are made twice a year and every few years point cloud laser (lidar) measurements are made of the surface and objects with high resolution. Noise calculations require a 3D model of the environment. RIVM collaborates with several partners [2] to combine building information with height data to provide a 3D model of all the buildings in the Netherlands. The lidar dataset is also transformed into a surface elevation model that is needed for noise calculations. Ground properties (reflecting or absorbing) are obtained from the topography maps. Anyone performing noise calculations in the Netherlands can use these datasets to jumpstart the 3D noise model. An example is shown in figure 1: 

Figure 1: Example of buildings and height profile of the surface of the town Naarden. 

Other datasets are used to calculate vibrations from trains, which is further elaborated on in the section about the vibration model.  

Noise calculation software 

The current Dutch noise legislation for road, rail and industrial noise originate from the 1980s. Those regulations also included detailed instructions on noise calculations, formulas and parameters: an approach that later resulted in the ISO 9613 standard. Commercial software solutions for noise calculations also started in the 80s. In the following 40 years the calculation instructions evolved and got updated. The last major revision was one and a half years ago when new noise legislation went into effect [3]. 

With this major update on calculation specifications our institute acquired one of the existing software solutions on noise calculations for road, rail, and industrial noise. We have updated the software to comply with the new legislation. This update [3] involved smaller adjustments like revised noise emission parameters for cars and pavements and an update on meteorological corrections based on average Dutch weather and wind conditions. Reflections of sound now take the height of the reflective surface relative to the wavelength of the sound into account. That means that lower frequencies (longer wavelengths) will show a different reflection factor compared to higher frequencies (short wavelengths). The biggest change overall was the inclusion of tilted barriers: the calculation specification requires a full 3D analysis when considering reflections from the barriers. Unfortunately, the 3D analysis creates a performance hit that considerably slows down the calculation. Therefore, the analysis is bypassed when there are no tilted barriers in the model. However, the tradeoff is that barrier configurations can be optimized. In figure 2 we show an example of increasing noise levels due to reflections when there are two barriers. Clearly, there is one configuration that has severe negative effects. 

Figure 2: Increase in noise level (in steps op 0.5 dB going from 0 to 10.5 dB) due to reflections for three different barrier configurations. 

Executable binaries are available for free via the website (in Dutch): https://www.rivm.nl/geluid/rekenmodellen-en-tools/rekenhart-geluid. We are currently preparing to publish the source code of the calculation software as an open-source project. This code will be accessible through the same webpage. By providing the source code, we offer transparency and extra information on the interpretation of noise calculations. It also enables a cost-effective solution for noise calculations by professionals. We hope to work together with the community to improve and enhance noise calculations. 

Having access to source code of noise calculation software enables us to run detailed trial runs on future adjustments or updates on the calculation specification. It also allows us to make alternate versions of the software for specific research questions. Although the main audience is focused on the Dutch situation, it might be of interest to other places as well. The source code could be a source of inspiration to implement noise calculation specifications by other countries. A new round of noise mapping with respect to the European Environmental Noise Directive (END) is in sight. The code may provide a basis for a calculation core that can be used to do calculations for rail, road and industry in accordance to the calculation method stated in Annex II of this directive (commonly known as CNOSSOS-EU). 

Railway vibrations 

Apart from noise, vibrations from trains are a growing concern. Approximately 11% of people (aged 16 years or older) living within 300 meters from a railroad track report severe annoyance due to vibrations and 13% are severely sleep disturbed [4]. However, due to housing crises many cities are looking to build new homes near railroad tracks. Up until recently not much thought was given to potential issues with vibrations. This now has changed. To see if any issue can occur, it is possible to measure vibrations from trains passing by at the building site. To acquire enough data (train passers-by) it is often necessary to measure continuously for a few weeks, which can be expensive. Often it is concluded that there are no issues with vibrations. It would be helpful if screening calculations could be done first to see if there is any risk of problems with vibrations. We have developed an open-source model that can do just this.  

One of the main difficulties with vibrations is that transmission through the ground is very dependent on (sub)soil type. As a nation that is largely below sea level and being a delta, soil is often layered and will vary significantly from location to location in the Netherlands. Without information on soil type no calculations can be made that have any value. Fortunately, this is where open data comes in. All Cone Penetration Tests (CPT) that have been performed by order of a public body must be registered in a central database that is publicly available. It has an average of 7 CPT’s per square km (18 per square mile). Although not evenly spaced, there are usually enough CPT’s available to model the subsoil at any given location. For this also a geomorphological map and a map of groundwater heights are used (both of which are also publicly available).  

The software itself consists of several modules. The first module uses the CPT’s and maps to model the subsoil in layers that each have characteristics needed to calculate vibration propagation in the next module. This next module is a Finite Element Method (FEM) module that calculates the transmission through the subsoil. Normally FEM calculations take up a lot of computer time, but here it was optimized as a fast 2D-model and calculations are possible up to 300 meters from the railroad track. A final module then calculates transmission from soil to foundation and finally to the floor inside a building. This is based on some common characteristics of buildings like size and expected type of floor (wood or concrete). All these modules are written in python and are open source with a public license. Next to these python modules there is a basic shell that connects to (SQLite) databases and can control all the modules. One of those databases is an emission database of different train types. These emissions were determined based on measurement campaigns (see figure 3). The shell takes care of all communication between the modules and transforms an input model (polylines with train intensities) to input that can be used by the modules. It is open source but written in a propriety language (the commercial Delphi Pascal compiler is necessary). Source code is available at [5]. Be aware that it is in Dutch and for the program to work it needs a subsoil database in a specific format. 

Figure 3: Measurement to determine emissions from trains with 9 measurement points in a grid of 25×25 meters (82×82 feet).  

Open and beyond 

Open-source software and open data help the noise and vibration professionals in the Netherlands. Consulting and engineering companies that perform studies on noise and vibrations for either new buildings or new infrastructure already benefit from these advancements. They can automate the modelling of the 3D environment needed for noise calculations, making their studies more cost effective and in the end cheaper for society. It also helps to uniform the studies, making it less dependent on who or which company performs the study. It unifies the research methods throughout the country. 

Already, several companies are building further on the open-source software and open data, thereby sparking new innovations and new ideas in the field. It also shows that by making data and methods available, new initiatives can start that maybe were not foreseen before. 

References 

[1] www.pdok.nl 

[2] https://innovation.3dbag.nl/en/ 

[3] Kok, A.Y. Enhancing the Dutch engineering calculation method, Internoise proc. 2022 

[4] Kempen, E. van et al, “Annoyance and sleep disturbance caused by vibrations from passing trains”, RIVM- report 2024-0052 

[5] https://github.com/rivm-syso/ours