Six years of pain and suffering.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
bibliography
chapters
docs
resources
submission
.gitignore
README.md

README.md

PhD Thesis

Six years of part-time pain and suffering in LaTeX form. Awarded July 2018.

The generated thesis is available as a PDF. The code for the software produced is in the HiPIMS repository.

Table of contents

1     Introduction                                                                     1
1.1   Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2   Aim and objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3   Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2     Literature review                                                                7
2.1   Flood simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2   Physically-based modelling of shallow flows  . . . . . . . . . . . . . . . . . . 9
2.3   HPC- and GPU-based flow modelling  . . . . . . . . . . . . . . . . . . . . . . . 12
2.4   High-resolution modelling in practice  . . . . . . . . . . . . . . . . . . . . . 13
2.5   Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3     First- and second-order shallow flow modelling for flood simulation on 
      heterogeneous computing devices                                                  17
3.1   Numerical methods for shallow flows  . . . . . . . . . . . . . . . . . . . . . . 17
3.1.1 Shallow water equations  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.2 HLLC solution to the Riemann problem . . . . . . . . . . . . . . . . . . . . . . 20
3.1.3 Boundary conditions  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.4 Stability criterion  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.5 Source terms for energy loss and topographic slope . . . . . . . . . . . . . . . 22
3.2   Extension to second-order solution . . . . . . . . . . . . . . . . . . . . . . . 22
3.3   Solution using heterogeneous architectures . . . . . . . . . . . . . . . . . . . 24
3.3.1 Control flow and software structure  . . . . . . . . . . . . . . . . . . . . . . 25
3.3.2 Identification of CFL-constrained timestep . . . . . . . . . . . . . . . . . . . 28
3.3.3 Storage and memory access patterns in a heterogeneous environment  . . . . . . . 29
3.3.4 Decomposition to stencil operations  . . . . . . . . . . . . . . . . . . . . . . 30
3.4   User interface and platform availability . . . . . . . . . . . . . . . . . . . . 34
3.5   Conclusions  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4     Validation and performance assessment for single-domain simulations              37
4.1   Ensuring the well-balanced property  . . . . . . . . . . . . . . . . . . . . . . 37
4.2   Moving wet-dry fronts  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.3   Dam-break simulation over an emerging bed  . . . . . . . . . . . . . . . . . . . 41
4.4   Dam-break against a fixed obstacle . . . . . . . . . . . . . . . . . . . . . . . 42
4.5   Hypothetical pluvial flood event . . . . . . . . . . . . . . . . . . . . . . . . 45
4.6   Real-life dam failure at Malpasset (1959)  . . . . . . . . . . . . . . . . . . . 50
4.7   Computational performance  . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.7.1 First-order Godunov-type scheme  . . . . . . . . . . . . . . . . . . . . . . . . 53
4.7.2 Second-order MUSCL-Hancock scheme  . . . . . . . . . . . . . . . . . . . . . . . 54
4.8   Effects of floating-point precision  . . . . . . . . . . . . . . . . . . . . . . 55
4.9   Conclusions  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5     Application to fluvial flooding and analysis of the impacts of grid resolution
      and parameterisation                                                             59
5.1   Flooding in Carlisle, January 2005 . . . . . . . . . . . . . . . . . . . . . . . 59
5.1.1 Model generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.1.2 Effect of grid resolution and parameterisation . . . . . . . . . . . . . . . . . 62
5.1.3 Computational performance  . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2   Hypothetical flooding in Thamesmead caused by defence failure  . . . . . . . . . 69
5.2.1 Model generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2.2 Effect of grid resolution and parameterisation . . . . . . . . . . . . . . . . . 71
5.2.3 Computational performance  . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.3   Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

6     Domain decomposition and multi-device computation                                79
6.1   Principles of domain decomposition in explicit time-marching schemes . . . . . . 80
6.2   A multi-device multi-nodal implementation with decomposition . . . . . . . . . . 82
6.3   A temporally-decoupled implementation for decomposition  . . . . . . . . . . . . 84
6.4   Software structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.5   Validation and testing of the domain decomposition algorithms  . . . . . . . . . 89
6.5.1 Lake at rest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.5.2 Moving wet-dry fronts  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.6   Performance of forecasted and coupled implementations  . . . . . . . . . . . . . 95
6.7   Conclusions  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

7     Application of multi-device computation to a detailed city-scale surface-water
      flood simulation                                                                 99
7.1   Flooding in Newcastle during June and August 2012  . . . . . . . . . . . . . . . 99
7.2   Data sources and model generation  . . . . . . . . . . . . . . . . . . . . . . . 100
7.3   Results  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
7.3.1 Computational performance  . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.3.2 Ability to reproduce large-scale flood features  . . . . . . . . . . . . . . . . 104
7.3.3 Ability to reproduce small-scale flood features, and sensitivity to
      grid resolution  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.4   Conclusions  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

8     An assessment of the potential presented by soft data sources for pluvial flood
      simulation                                                                       113
8.1   Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
8.2   Modelling framework  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.2.1 Social media harvesting and analysis . . . . . . . . . . . . . . . . . . . . . . 116
8.2.2 Real-time modelling  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
8.2.3 Flood model result analysis  . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.3   Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
8.3.1 Geolocating Tweets and identifying binary assertions against which
      to assess model performance  . . . . . . . . . . . . . . . . . . . . . . . . . . 123
8.3.2 Correlation of models against criteria from social media and known
      data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.4   Conclusions  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

9     Conclusions and recommendations                                                  129
9.1   Leveraging advances in computing power for expedited flood simulation  . . . . . 129
9.2   Ability to produce accurate results comparable to existing software  . . . . . . 130
9.3   Limitations of data sources available  . . . . . . . . . . . . . . . . . . . . . 132
9.4   Recommendations for professional practice  . . . . . . . . . . . . . . . . . . . 133
9.5   Potential for real-time high-resolution flood forecasting and nowcasting . . . . 134

References                                                                             136