Research Overview

Introduction

• This presentation is written using http://orgmode.org/ in Emacs and the reveal.js framework.
• Type ? for help at any time.
• Use left and right arrows to move between topics.
• Use up and down arrows to move within a topic.

Table of contents

1. µ-grids for distributed energy systems.
2. Large scale off-grid mining operations.
3. Sustainable production for transparent conducting oxides
4. Shale gas water processing.
5. Beer fermentation
6. Surrogate modelling

µ-grids

An optimisation framework for the design of integrated distributed energy systems for residential use.

Research by Ms Carmen Wouters, PhD student, UCL Australia & UCL Chemical Engineering.

C Wouters, E S Fraga, A M James (2015), Energy 85:30-44

Modelling a µ-grid

• Superstructure for power, heating & cooling.
• Distributed generation: µ-CHP, solar PV, wind turbines.
• Storage: hot water tanks, absorption chillers, battery banks.
• Dimensions: time + 2 × space.
• Simple discretisation in time → mixed integer linear programming (MILP) model.

Superstructure: power

Superstructure: heating and cooling

Multi-objective Optimisation

1. cost (capital, operating, grid energy, value of excess production),
2. unavailability,
3. environmental, and
4. impact of ownership options.

Optimal design: 5 houses

Key: grey: µ-CHP; B: boiler; H: hot water storage; →: heat transfer; sun: PV.

20 houses

All houses have PV and heat storage.

Unavailability

Regulatory effects

Large scale off-grid mining operations

An optimisation based framework for the design of integrated energy systems, including generation, storage and demand, for large scale off-grid mining operations.

Research by Mr Oluwamayowa Amusat, PhD student, Chemical Engineering, UCL.

O Amusat, P Shearing & E S Fraga (2016), Computers & Chemical Engineering 95:21-37

Characterised by

• Large energy use, both thermal and power,
• continuous (24 hours per day) operation over years,
• beneficial to use local renewable energy sources, and
• need thermal and electric energy storage to address variability of energy generation.

Integrated energy system

A superstructure including alternative power generation sources and alternative storage technologies.

Case study: Collahuasi mine

Specifications and model

• North-east Chile.
• Thermal demands are 10% of electrical demands.
• Hourly solar radiation using clear sky models.
• Daily and seasonal behaviours.
• Resulting mixed integer non-linear programming (MINLP) model solved using BARON in GAMS.

Resulting Design

The design chosen incorporates solar PV for electricity generation, coupled with pumped hydro storage, and solar thermal for heat generation, coupled with molten salt storage.

Variability and reliability

Generation variability affects performance reliability and costs.

Sustainable production for transparent conducting oxides

Process scale-up for aerosol assisted chemical vapour deposition using sustainable materials for transparent conducting oxides.

Research by Mr Pedro Ivo de Oliveira Filho with UCL Chemistry, EPSRC EP/L017709/1.

Droplet size distribution

Shale gas water processing

EC funded project, Maximising the EU shale gas potential by minimising its environmental footprint, (ShaleXenvironmenT) includes a work package on the life cycle for water processing.

This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 650979.

Design alternatives

1. technologies for water purification, and
2. destinations for water processing.

Design Superstructure

Optimisation problem with multiple, possibly conflicting, objectives.

Multi-effect evaporation for desalination

Beer fermentation

• Minimise batch times while maximising fermentation with minimal impurities.
• Multi-criteria decision making with dynamic (time depedent) models.
• In collaboration with University of Edinburgh.

A Rodman & D Gerogiorgis (2016), Food & Bioproducts Processing 100:255–274.

Strawberry algorithm

• Nature inspired plant propagation algorithm.
• Stochastic population based method.
• Novel fitness ranking method based on Hadamard product of solution ranking.

E S Fraga & O Amusat (2016), in Advances in Stochastic and Deterministic Global Optimization, 243–254.

The algorithm

Three criteria

2d view

Surrogate modelling

For computationally expensive optimisation problems.

Motivation

Noisy objective function

Kriging

A statistical interpolating approach used for approximating deterministic models.

Surrogate Based Optimisation

(continued)

Dual piston pressure swing adsorption

Pareto front: n=64

Pareto front: n=96

Pareto front: n=176

Pareto front: n=256

Further information

My personal pages have more information including more detailed descriptions of my research interests.