Project Highlights:

  • Novelty: Development of a novel and efficient tool to assess air quality in a complex built environment.
  • Feasibility: A successful CASE partnership, an experienced interdisciplinary team, and available data for case studies.
  • Impact: Translation of science into practical guidance to enable regulators to design efficient modelling and monitoring.

A key task for environmental regulators is to protect human and environmental health by ensuring that exposure to harmful pollutants is controlled. Air pollution is a particular challenge at present: estimates of deaths due to exposure to air pollution in the UK are in tens of thousands, and WHO estimated that in 2012, nearly seven million deaths (1 in 8 of deaths globally) were attributable to air pollution.

In relatively simple situations, current “routine” assessment tools based around Gaussian dispersion models provide valuable insights into what exposures are likely to be received. Many current regulatory situations, however, involve complex flow fields, e.g. in built environments. Sources can be at ground or elevated levels and buildings have a strong impact on airflow and turbulence characteristics, and hence on pollution transport and dispersion. The use of models (e.g. CFD) which are able to resolve flows in complex topography (Fig. 1) is time and labour-intensive, and beyond the present resources of regulators in any but the highest risk situations.

Complex flow fields yield two related challenges:

  1. How can we better estimate probable peak and time-averaged pollution concentrations across an area of complex topography in order to apply appropriate limits to emissions at source?
  2. How can we interpret measurements taken within complex flow fields to get the best estimate of emission rate (and in some situations, such as those involving diffuse or multiple sources, source locations)? Conversely, where should we place monitors in order to get the best possible information on pollutant concentration and how long must we monitor for to get representative concentration data?

In this project, we aim to address the above two challenges using CFD-based sensitivity analyses combined with multiscale modelling to derive heuristics and a tool which may be applied in real-world situations to optimise sensor placement, and to gain the best understanding of pollutant concentrations to support regulatory decision-making. The outcome will be that regulators will be better equipped to estimate likely exposures resulting from a polluting activity. Designers will also benefit, gaining insights into “problem” emissions which might be avoided by alternative plant design.

An example of CFD output from a pilot study of pollutants emitted from an array of diesel generators on a regulated site: (a) air flow represented by streamlines, and (b) its arrival at ground level (red) for a specific meteorological scenario.


We will generate data from CFD for (1) regular array of cuboids for a targeted set of dimensions and spacing, and (2) real-world building configurations under various background wind conditions for which on-site measurements of pollution concentration were made. The output from (1) will be used to construct systematic knowledge of persistent circulation regions and dividing streamlines, and subsequently to build a modified plume model. The output from (2), together with the new model’s estimates, will be used to analyse the impact of monitors’ setting (number and placement) on the uncertainty of concentration estimates using Bayesian methods. A similar approach will be used to test the new model’s performance with increasing complexity of built environments.

The research outcome will be used to derive heuristics, prediction and guidance for monitoring design and modelling strategy needed for appropriate estimates of pollution concentration field to a given degree of statistical confidence.

Training and Skills

Student will take a training course on CFD. High-Performance Computing training will be provided by IT Services of University of Birmingham, e.g. UNIX and parallel computation. Student may receive training on (whichever is needed depending on student’s educational background) numerical modelling, non-Gaussian modelling, and Bayesian statistics, by supervisors and research fellows, Drs Zhong and Mazzeo, of the modelling group.

Student will also be trained by the CASE partner, Environment Agency, on air quality assessment, communication, teamwork, managerial, and leadership skills, and gain the knowledge of how science is used in the public sector and how Environment Agency operates.


Year 1:

  • Conduct literature review
  • CENTA generic skills training
  • Specialist modelling training (CFD & other codes)
  • Subject training according to needs (eg numerical modelling, non-Gaussian modelling, and Bayesian statistics);
  • Define and set-up model configurations;
  • Design the strategy of model evaluation;
  • Conduct initial analysis of case study data

Year 2:

  • Write a literature review paper
  • Advanced CENTA training
  • Design strategy of scenario simulations
  • Conduct simulations
  • Complete model evaluation
  • Analyse model output
  • Develop Modified plume Model
  • Write a jounral paper

Year 3+:

  • Continue simulations and analysis
  • Write further journal paper(s)
  • Compile thesis


Partners and collaboration (including CASE)

Environment Agency (EA), the CASE partner, depends upon good knowledge of pollutant exposure risks to set proportionate and targeted regulation. They are keen to engage with this project for better design monitoring approaches and data interpretation. Together with a CASE financial contribution and co-supervision, the EA will provide support on data for case studies against which to develop and test methodologies.

Kinnersley and Cai have successfully supervised one CASE student previously on air pollution studies. This collaboration yielded three published journal papers and a PhD graduate currently employed by Cambridge Environmental Research Consultants, working on developing air quality models.

Further Details

This project has been selected as a CENTA Flagship project. This is based on the projects fulfilment of specific characteristics e.g., NERC CASE support, collaboration with our CENTA high-level end-users, diversity of the supervisory team, career development of the supervisory team, collaboration with one of our Research Centre Partners (BGS, CEH, NCEO, NCAS), or a potential applicant co-development of the project.

Any further details of the project can be obtained from:

Dr Xiaoming Cai, School of Geography, Earth and Env. Sci., University of Birmingham, email: x.cai@bham.ac.uk

website: https://www.birmingham.ac.uk/schools/gees/people/profile.aspx?ReferenceId=10025&Name=dr-xiaoming-cai