Our reliance on data is increasing substantially. With so many aspects of our economic, commercial and social lives now online, data centres – or mission critical facilities – have become one of the most important classes of infrastructure.
Fundamentally, data centres must be designed and built for resilience, efficiency and security. To ensure security of power to the facility, a relatively small area of the site will be allocated to onsite emergency back-up power generation. Typically, these back-up generators will be fuelled by diesel and their installation will trigger the need for an air quality assessment to accompany a planning application or an environmental permit. The back-up generators need to be tested at regular intervals throughout the year and the generators emit potentially harmful pollutants into the air when tested or used in an emergency. The key pollutants of concern from diesel generators usually are:
- Nitrogen dioxide (NO2)
- Particles up to 10 µm and 2.5 µm in diameter, which remain suspended in the air for long periods and are small enough to be breathed in (known as PM10 and PM2.5 respectively)
- Sulphur dioxide (SO2)
- Carbon monoxide (CO)
- Benzene (C6H6)
In the UK, the Air Quality Standards Regulations and the Air Quality Strategy set limit values and objectives for the protection of human-health. The Environment Act 1995 established the system of Local Air Quality Management that requires local authorities to go through a process of review and assessment of air quality in their areas. The process requires local authorities to identify places where objectives are not likely to be met, then declare Air Quality Management Areas (AQMAs) and put in place Air Quality Action Plans to improve air quality. In 2021, there are more than 600 AQMAs designated across the UK in relation to high levels of NO2, PM10 and SO2. Most of the AQMAs are designated due to high levels of NO2. In London, every borough has designated at least one AQMA. The particular challenges faced by developments in London are considered later.
The existence of an AQMA indicates that air quality in the location is already a sensitive issue. With most AQMAs being designated due to high levels of NO2 and PM10, applications for data centres with diesel back-up generators, which emit both these pollutants, are likely to be scrutinised in detail. Air quality assessments need to demonstrate that emissions do not make air quality worse within an existing AQMA and do not raise levels elsewhere requiring the designation of a new AQMA.
Assessing the Impacts
Pollutant concentrations are primarily determined by the balance between pollutant emissions that increase concentrations, and the ability of the atmosphere to reduce and remove pollutants by dispersion, advection, reaction and deposition. An atmospheric dispersion model is used as a practical way to simulate these complex processes and is often used to assess the air quality impact of emissions from a data centre on the surrounding area. The model requires a range of input data, which can include emissions rates, meteorological data, building dimensions and local topographical information.
The atmospheric pollutant concentrations in an urban area depend not only on local sources at a street scale, but also on the ambient pollutant level made up of the local urban-wide background, together with regional pollution and pollution from more remote sources brought in on the incoming air mass. This ambient contribution needs to be added to the fraction from the modelled sources. It is usually obtained from local authority measurements, site specific measurements and/or Defra mapped concentration estimates.
One of the main ways to mitigate air quality impacts from a data centre is to make sure that the stack or flue from which emissions are released is high enough to provide adequate dispersion/dilution and to overcome local building wake effects. This is usually determined by running the dispersion model and predicting concentrations for a range of possible stack heights. The movement of air over and around buildings generates areas of flow circulation, which can lead to increased ground level concentrations in the building wakes. Where building heights are greater than about 30 – 40% of the stack height, downwash effects can be significant and the dominant structures (i.e. with the greatest dimensions likely to promote turbulence) are normally included within a dispersion model. The modelling needs to consider a wide range of potential meteorological conditions and it is good practice to model between three and five years of meteorological data collected at a nearby representative station. That’s between approximately 26,280 and 43,800 hours of different meteorological conditions.
The modeller analyses the variation in the predicted pollutant concentrations and determines the height at which no benefit is gained from further increases.
Analysing the Results
Once the optimum stack height has been determined, the model can be used to predict the air quality impacts of the development on people living nearby. The testing of the back-up generators requires each to be run for short periods, intermittently throughout the year. In an emergency power outage, it’s likely that all the back-up generators will be in use at the same time, but this is again only likely to be the case for a short period. The main challenge therefore is meeting short-term limit values and objectives, rather than long-term limit values and objectives. In the case of NO2, the limit value and objective allow the hourly-mean concentration to exceed 200 µg.m-3 in 18 hours over a calendar year. In other words, when the hourly-mean NO2 concentration predicted by the model is added to the background concentration, the total must be below 200 µg.m-3 99.79% of the time.
Clearly, the weather conditions when the generators are tested at some point in the future are not known. The common approach to dealing with this uncertainty is to initially model the generators as if they were each operating in every hour of the year. If the short-term limit value and objective are met for this highly conservative scenario then the applicant can be confident the air quality effects are not likely to be significant. This is largely the case for most of the pollutants emitted, except NO2. As set out above, the majority of the AQMAs in the UK are designated due to high levels of NO2, so changes in this pollutant are a key concern for decision makers. Local authorities that are trying to improve NO2 levels in their AQMAs by implementing Air Quality Action Plans will be particularly concerned about a new source of pollution.
If the short-term limit value and objective are not met for this highly conservative scenario, the next step requires some complex statistical analysis. Having run the model in every hour of the year, the air quality practitioner then needs to determine the probability of the operational hours coinciding with the meteorological conditions that give rise to the highest hourly concentrations. If the probability of this coincidence is low (less than 5%), then an exceedance can be considered unlikely.
Nature Conservation Sites
Air quality impacts are not just a problem for people living nearby. An increase in pollutant concentrations and/or deposition of pollutants onto surfaces of the plants or the ground surrounding plants can affect the biodiversity of habitats in nature conservation sites.
The UK Air Quality Regulations set out critical levels which are air pollutant concentrations above which adverse effects on ecosystems may occur based on present knowledge. The relevant critical levels in this case are for total nitrogen oxides (known as NOX) and SO2. The United Nations Economic Commission for Europe sets out critical loads which refer to the quantity of pollutant deposited, below which significant harmful effects on sensitive elements of the environment do not occur, according to present knowledge. The relevant critical loads in this case are for nutrient nitrogen deposition and acid deposition. These critical loads have been adopted by regulators (including the Environmental Agency in the UK).
Complying with the critical levels and critical loads is usually an easier task for data centres than meeting limit values and objectives for human exposure. This is because most critical levels and loads are measured over the period of a year so impacts from short, intermittent usage are less likely to be severe. The one pollutant which can be problematic is NOX which has a critical level expressed as a maximum daily-mean of 75 µg.m-3. This can be an issue if there is a sensitive habitat immediately downwind of the site. In exceptional cases, it may be necessary to use statistical tools to establish the probability that the back-up generators will ever be required to operate for a full day of use and the probability that if a day is randomly selected that it coincides with meteorological conditions that give rise to the highest daily concentrations.
London’s air quality problems are well known: The Low Emission Zone was implemented in 2008, followed by the Ultra-Low Emission Zone (ULEZ) in 2019 and an extension to the ULEZ in 2021. For major developments, the current London Plan also requires developers to design developments to be at least Air Quality Neutral which adds further challenges for data centres.
Air quality assessments consider the impacts of a development on ambient air quality at sensitive receptor locations (i.e. people or nature conservation sites) by comparing predicted levels with limit values and objectives. Air Quality Neutral assessment requires the development’s building (and transport) emissions to be compared with benchmarks. Where emissions are above the benchmark, developers are required to identify proportionate mitigation. The benchmark for building emissions depends on the land-use class and the development’s floor area.
The land-use class commonly adopted for data centres is B8 (Storage or Distribution) and, in the absence of any specific benchmark, it is commonly adopted in the calculation for the data centre. However, using this approach, the development’s emissions often exceed the benchmark by some margin. As the emissions are related to the necessary, regular testing of the generators, it is rarely possible to reduce the emissions. If the developer is not able to demonstrate that the development is at least Air Quality Neutral, it may be difficult to secure planning permission for the development.
B8 land-use would have very different building emissions than a data centre and so the benchmark commonly adopted may be inappropriate. The current London Plan explains that the ‘Air Quality Neutral’ policy is designed to address the problem of multiple new developments that individually add only a small amount to pollution, but together lead to baseline air pollution levels creeping up. This tends to suggest that the policy relates to residential and mixed-use developments which each only generate small emissions. A data centre is expected to have substantial emissions and therefore falls outside this description.
“the WHO published new guidelines in September 2021. The new WHO guideline for PM2.5 is 5 µg.m-3”
On behalf of our clients, we have questioned whether the Air Quality Neutral calculation is designed to be used for data centres. The more critical issue is surely whether emissions from the development are likely to have an adverse impact on surrounding land-users, after they have dispersed away from the site. As set out above a key mitigation measure is ensuring that the stack or flue is sufficiently high to ensure pollutants are adequately dispersed.
What does the Future Hold?
The Environment Bill will require the government to set new targets on air quality, including for fine particulate matter. The Bill introduces a legally binding duty on the government to bring forward at least two air quality targets by October 2022. The first is to reduce the annual average level of PM2.5. The second air quality target must be a long-term target (set a minimum of 15 years in the future).
The World Health Organisation (WHO) publish Air Quality Guidelines to inform the setting of air quality standards. The guidelines are not standards as no consideration is given to whether the levels can be reached in any country or region, or indeed how levels could be achieved and at what cost. The WHO published new guidelines in September 2021. The new WHO guideline for PM2.5 is 5 µg.m-3. Placing this in context, the current limit value and objective for PM2.5 is 25 µg.m-3 and a review of measurements across the UK in 2019 (i.e. in pre-pandemic times), the annual average concentration was above 5 µg.m-3 at almost all monitoring stations. The adoption of a considerably more stringent PM2.5 limit value under the Environment Bill could add to the challenges for back-up generators using diesel. A public consultation on proposed targets is expected in early 2022.
“more stringent PM2.5 limit value under the Environment Bill could add to the challenges for back-up generators using diesel”
The London Plan published in March 2021 acts as an integrated framework for a set of strategies, including improvements to air quality. Policy SI 1: Improving Air Quality in the London Plan states that developments should “consider how local air quality can be improved across the area of the proposal as part of an air quality positive approach”. A pre-consultation draft of London Plan Guidance – Air Quality Positive was published in March 2021. It remains to be seen whether data centres would be required to be Air Quality Positive; however, the draft includes an example matrix of air quality positive measures including one that states that alternatives to diesel generators are now available, and back-up power should be provided by zero emission gensets from either battery or hydrogen technology.
“future legislation is expected to place more stringent limits on ambient air pollutant concentrations”
Data centres must be designed and built for resilience, efficiency and security. To ensure security of power, onsite emergency back-up power generators are employed. Typically, these back-up generators will be fuelled by diesel which emit potentially harmful pollutants into the air when tested or used in an emergency. Future legislation is expected to place more stringent limits on ambient air pollutant concentrations and create additional challenges. Early consideration of the likely air quality impacts allows appropriate mitigation to be built into the design. This can be a key factor in a successful planning application or an environmental permit.