University College London Featured PhD Programmes
Imperial College London Featured PhD Programmes
Engineering and Physical Sciences Research Council Featured PhD Programmes
Sheffield Hallam University Featured PhD Programmes
European Molecular Biology Laboratory (Heidelberg) Featured PhD Programmes

Employing artificial intelligence & machine learning to optimise flood control for 21st Century cities

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
  • Full or part time
    Prof O Bokhove
  • Application Deadline
    Applications accepted all year round
  • Competition Funded PhD Project (European/UK Students Only)
    Competition Funded PhD Project (European/UK Students Only)

About This PhD Project

Project Description

Summary Rational and unique mathematical strategies for active flood control will be developed and tested in the design of flood alleviation schemes for UK rivers, such as Yorkshire’s Aire and Calder Rivers, based on increasing flood-plain storage capacity.

Flood-excess volume calculations reveal that enhancing flood-water storage on floodplains and in reservoirs is realistic and good value for money. It can minimise flood damage to our cities. A crucial component to optimise flood-water capacity, in advance of extreme precipitation predictions, involves using artificial intelligence and machine learning to find the approximate yet pseudo-optimal solution of the comprehensive operational system consisting of measured data assimilated into model predictions. Due to the nonlinearities involved, genetic algorithms have shown to be promising, but their accuracy and performance should be compared with more expensive and accurate model reduction techniques and two-point boundary-value techniques.

Applications to flood mitigation for Yorkshire rivers will be considered, in collaboration with the Environment Agency and hydraulic consulting firms. Background and motivation The Boxing Day Floods of 2015 in Leeds had an estimated flood-excess volume equivalent to a “lake” of 2150x2150x2m3 (~9.3Mm3) concerning river levels over 3.9m in Armley [1.5].

Similarly, the Calder River had an estimated flood-excess volume equivalent to a “lake” of 813x813x2m3 (~1.3Mm3) concerning river levels over 4.5m at Mytholmroyd. Flood-excess volume is the volume that caused the flood. If one can reduce it to zero, then there would be no flood [1,6]. The first hypothesis that can be investigated is whether multiple strategically placed dynamic weirs, normally open but actively controlled on average every 1:50 to 1:100 years, say, could have prevented the worst or all flooding in the middle Aire River catchment around Leeds.

Furthermore, the Environment Agency (EA) has explored reservoirs in the Calder River Valley as flood-storage sites, hitherto by a static control in drawing down reservoir levels before heavy rainfall. The EA has found that 0.88M to 1.3Mm3 of flood-water storage can be reached. The second hypothesis that can be investigated is whether one can optimally adjust these reservoir levels to both minimise flood damage and maintain drinking water supplies. Both nature-based solutions and active control are applied because weirs and reservoirs are placed at naturally suitable locations, creating “space for water” in the Aire and Calder River catchments, and weir and reservoir levels are actively controlled to minimise flood damage and maintain drinking water capacity.
Objectives
We propose an exciting PhD opportunity to develop the mathematical research foundations of flood control. The objectives and tasks are the following:
The final objective is to create a professional guidance document (as a proof of principle) on flood control for the industrial partners, Mott MacDonald and the Environment Agency, regarding both weir and reservoir control, potentially exemplified for the Aire River concerning Leeds and/or the Calder River concerning Mytholmroyd and Hebden Bridge.
The study and further development of genetic algorithms (Artificial Intelligence –AI) as well as Model Predictive Control (MPC, Breckpot 2013 [2]) for control of floods and drinking water supply, through computational and mathematical modelling.
The translation of the real-world demands into a conceptual, idealised mathematical model, for which the associated computational realisations can be tested with data from an actual laboratory environment, e.g. by using the table-top flood demonstrator Wetropolis [4].
The transfer of the AI and MPC techniques for flood control from the idealised, conceptual testing environment into the real-world cases, leading to the professional guidance document for the industrial partners.
Training
The student will be given all the necessary training (via the Schools involved, the CDT, [email protected] and training courses/placements at the respective industrial partners) to perform all required tasks and exploit capabilities of the Hydraulic Laboratory in the School of Civil Engineering for demonstrations.

References

References:
[1] Bokhove 2016/2017: Facebook page "Resurging Flows''
[2] Breckpot 2013: Flood control of river system with model predictive control. PhD Thesis KU Leuven, Belgium.
[3] Ding, Wang 2012: Adv. Water Res. 44, 30.
[4] Bokhove 2017: Wetropolis Flood demonstrator.
[5] Bokhove, Kelmanson, Kent 2018: On using flood-excess volume for flood mitigation, exemplified for the Aire River Boxing Day flood in 2015. Under review at EA.
[6] Bokhove, Kelmanson, Kent 2018: On using flood-excess volume to assess natural flood management, exemplified for extreme 2007 and 2015 floods in Yorkshire. In preparation.

Related Subjects



FindAPhD. Copyright 2005-2019
All rights reserved.