A recent paper by Salim et al. assessed the performance of three different numerical techniques – Reynolds-averaged Navier-Stokes (RANS), Unsteady Reynolds-averaged Navier-Stokes (URANS) and Large Eddy Simulations (LES), to identify their suitability to predicting urban airflow and pollutant dispersion processes. This article reviews the performance of RANS v LES with regards to the prediction of airflow and pollutant dispersion.RANS v LES In The Fight Against Pollution
Pollution: A Global Health Risk
The quality of air in both urban and industrial environments has a number of implications for human health, and as such, is of great importance. The World Health Organization (WHO) has named air pollution as the world's largest single environmental health risk, and has recently released country-by-country estimates on air pollution exposure and the impact it has on health in a report.
The importance of developing new simulation tools, as well as improving those already in existence, cannot be underestimated. Such improvements enable regulators and urban planners to address problems with pollution in cities, as well as assisting emergency services and authorities with their evacuation plans in the event of natural disasters, accidents or hazardous airborne matter release.
It is widely accepted that computational fluid dynamics (CFD) can provide significant insight into the urban canopy microclimate, and the growth of computing resources over recent decades correlates to an increased ability to tackle these challenging simulations, and also increased focus on development of turbulence models.
RANS v LES and For Pollutant Airflow Prediction
While many turbulence models may be used in predicting airflows, their accuracy and robustness must be considered, to ensure the most reliable results, and allocation of resources.
Previous urban airflow investigations used RANS turbulence closure schemes, but over-predict pollutant levels compared to wind tunnel results. The assumption of steady-state solution in these numerical analyses is widely regarded a major cause of discrepancy.
In an attempt to address these apparent shortcomings, the study by Salim et al. compared the performance of steady-state RANS against transient URANS and LES results in predicting airflow and pollutant dispersion within an urban street canyon.
In addition to the obvious reliability and accuracy considerations, the question as to whether URANS could be a more financially viable alternative to LES, was also investigated.
The results of this study are relevant to many flow problems where large scale eddies dominate, and where accurate results can be acquired by resolving transience.
RANS v LES and URANS v LES: The Results For Pollutant Airflow Prediction
Numerical results were evaluated against wind tunnel experimental data acquired from an online database.
Steady-state vs Transient solution (RANS vs LES)
It was found that LES reproduces the pollutant concentration distribution as predicted by wind tunnel experiments, more accurately than RANS turbulence models, and its accuracy is less sensitive to location within the simulated domain. LES agrees with experience for all locations along the leeward wall, only slightly over predicting along the windward wall. Steady-state RANS models varied according to location, over predicting in some places and under predicting in others.
The flow variables vary significantly over time and LES is able to capture pockets of intertwining bubbles of opposing velocities.
Fig. 1 Mean normalized concentration on Wall A and Wall B for a) WT, b) Standard k-e, c) RSM and d) LES
Fig. 2 Time-evolution of the normalized concentration along Wall A and Wall B at different times, obtained by LES
Transient Solutions (URANS vs LES)
Although solving for transient solution, URANS’ results did not vary with time when compared against LES. The reason for this is URANS’ dissipativeness and its inability to resolve smaller fluctuations of the flow field, on which the transport of pollutants is so dependent. For this reason, URANS cannot be a direct replacement for LES when needing turbulent mixing is an important factor in the study.
In fact, URANS are applicable only to non-stationary flows, such as periodic or quasi-periodic flows that involve deterministic structures, often falling short of capturing the energy-containing scales.
LES resolves the fluctuations of the flow variables, capturing the transient mixing that proves so important in accurately predicting pollutant dispersion.
Fig. 3 URANS against LES for unsteady simulations at different time instances showing the wall concentration levels
RANS V LES: Which Is Best?
The study employed the use of RANS, URANS and LES within an urban street canyon, comparing and validating against wind tunnel experimental data.
For an accurate prediction of the flow and concentration fields to be made within urban street canyon, the unsteadiness must be accounted for by resolving both internally and externally induced fluctuations.
During the study, RANS simulations prodiced high spatial variability in the results, and generally predicted pollutant concentrations poorly. The study identifies the reason for this as a ‘failure to capture the turbulent mixing of the flow field’.
Although URANS is an appealing choice for a transient solution (due to lower computational costs than LES), it was not found to be a good replacement for LES in the case of air pollution prediction.
Furthermore, the study concludes that URANS cannot be relied upon in any generic situation where small scale eddies are an integral part of the flow field development.
Overall, LES demonstrated the highest level of accuracy and consistency, when compared to both RANS approaches. This is because LES resolves the inherent fluctuations, so captures the turbulent mixing process in the canyon’s flow field.
EXN/Aero: CFD Simulation In The Cloud
- Our general purpose, cost-effective CFD, cloud solver (EXN/Aero) reduces simulation times by up to 20x, and the platform includes a meshing tool, the solver, and a post-processing tool, as well as plenty of storage for your files. Hosted in the cloud, the pay-as-you-go model fits seamlessly into the workflow of engineering freelancers and consultancies alike.
There's no excuse to not being able to take advantage of simulation. Envenio has created the Onboarding Program to further assist those without CFD experience. With the Onboarding Program, Envenio works alongside the client to kick start their simulation project while training them fully on the platform. Alternatively, the Discovery Project allows teams to outsource their projects to Envenio's experienced team.