Reduced order models are popular and powerful techniques for circumventing the intensive computational burden in large complex numerical simulations in engineering and science, for example, ocean modelling, weather prediction, uncertainty quantification, sensitive analysis, data assimilation, sensor placement optimization, porous media, structural problems, convection diffusion reaction equations, molecular dynamics simulation and optimal control. The below video shows these capabilities in action, in this instance the model was used to examine air flow past two buildings.

Reduced order modelling techniques enable predictions to be made in seconds and can link with real time data. The techniques can be used for emergency response, real-time operational prediction and management. A rapid predictive modelling framework (including reduced order modelling techniques and data assimilation) is capable of:

1) Quantifying the effect of model uncertainties and performing sensitivity analysis;

2) Providing the optimal estimation of uncertainties by assimilating the field data into the models, thus improving the predictability of the models at the selected sites;

3) Optimal controls/designs and optimal sensor locations.

The below video shows the progress made in Reduced Order Modelling, demonstrating a NIROM (left) using a 12 POD basis vs a full model (right).

The below video shows how Reduced Order Modelling has been used as part of a collaborative effort by Imperial College London and their Chinese partners IAP, to indicate pollutant release from over 100 coal power plants across 55 densely populated cities, including Beijing. This type of work can help inform policy decisions and emergency health response teams.

Models demonstrating the flow past two buildings--top (full model), bottom (reduced order model). Credit: F. Fang et al., (Imperial College London, 2017)

Reduced order models allow for the simplification of large complex mathematical models, with the purpose of reducing intensive computer burdens and enabling predictions to be made in seconds, whilst linking to real time data.

NIROM using a 12 POD basis (left), full model (right). Credit: D. Xiao, (Imperial College London, 2017)

our mission

​01. Develop cities with no air pollution or heat island effect

The ultimate aim of this project is to find a cost-beneficial method in which to change the way our cities are developing. The Victorians improved health by covering sewage systems- let's see if we can do the same by improving air quality.

02. Work with others to improve our research

This project encompases a transdisciplinary research group from the Universities of Cambridge, Surrey and ICL, but we know that innovations take place beyond our reach, therefore we want to work with other academics and industry partners to further our work.

03. Share our research to improve global communities

We want to inform decision makers to ensure the results of this project can benefit cities across the globe, therefore we are excited to share all elements of our research to ensure the sustainable development of cities for the future.

REDUCED ORDER MODELLING

Models demonstrating air pollution in China (ROM is the top left). Credit: F. Fang et al., (Imperial College London, 2017)

what is REDUCED ORDER MODELLING?

Envisaging a world with greener cities.