I am a member of the Molecular Systems Engineering (MSE) research group and the Clean Energy Processes (CEP) Laboratory in the Chemical Engineering Department. The activity of the MSE group involves the intelligent molecular design of materials for engineering applications. The research combines a fundamental physical understanding, mathematical models and numerical methods into new techniques and tools for the design of better products and processes. The emphasis is on integration of models across different scales so that molecular-level models can be used at the larger scale of products and processes. My role in the CEP Laboratory follows the same philosophy, applied in the context of the working fluids utilised in heat engines and refrigerators. In this context, my principal research interests reflect my background in statistical mechanics and thermodynamics, and can be summed up as investigating how fundamental understanding of physics at the molecular level can be applied to tackle real-life engineering problems.
INDUSTRIALLY DRIVEN / APPLIED RESEARCH
Qatar Carbonates and Carbon Storage Research Centre (QCCSRC)
From the autumn of 2009 until its conclusion in December 2018 I was a member of the QCCSRC, a multifaceted research centre aimed principally at furthering understanding of all aspects of carbon sequestration in saline aquifers, as well as issues relating to Qatari oil and gas. An aquifer is an underground layer of water-bearing permeable rock; the name of the centre reflects the carbonate nature of the rock comprising the Qatari aquifers. My role was to direct and coordinate the research in the MSE group relating to the thermodynamics of all the relevant fluids – principally hydrocarbons, carbon dioxide, water and brines. In association with other members of the MSE group, in particular, PhD students Simon Dufal and Daniel Eriksen, and, latterly, fellow researcher Srikanth Ravipati, this has involved the development of new models for use with the latest in-house version of the statistical associating fluid theory, SAFT-VR Mie, as well as a refinement of our electrolyte-modelling capability, SAFT-VRE Mie. Most recently, the focus of the research was directed at interfaces. We developed a tractable implementation of the square-gradient theory to calculate interfacial tensions, in conjunction with our SAFT approaches. Our work on coarse-grained fluid-molecule–solid surface interactions represents a significant advance in the study of fluid–solid interfaces.
Organic Rankine cycles – conversion of waste heat to useful energy
Following the development in the Molecular Systems Engineering research group of the state-of-the-art, molecular-based equation of state (EoS) SAFT-VR Mie, using a single model we are now able to routinely capture a wider variety of thermodynamic properties of a fluid with quantitative accuracy; among these are calorific quantities, and second-derivative properties such as heat capacities. This has provided the opportunity to embark on research into the design of working fluids for heat engines based on Rankine and related cycles. These organic Rankine cycle (ORC) devices can be used to convert different sources of heat, including waste heat, to useful energy. Their analysis requires quantitative accuracy in the prediction of the calorific quantities that are used to compute efficiencies; hitherto it had not been possible to supply this information (at the required quality) cheaply using an EoS. In this context, I have collaborated for several years with Dr Christos Markides, head of the Clean Energy Processes Laboratory in the Chemical Engineering Department, who is an expert in the design and analysis of these types of engine; cementing this collaboration I joined the CEP Laboratory in January 2019.
Flow assurance: gas (clathrate) hydrates
The plugging of pipelines by this unfamiliar form of “ice” remains a topic of enormous importance in the recovery of natural gas from deep-sea reserves. This research centres on the thermodynamic modelling of gas hydrates and its incorporation in flow simulations for gas risers. An important aspect of this work is the thermodynamic modelling of aqueous mixtures involving natural gases and hydrate inhibitors, such as methanol.
In a recent collaboration with Borealis, an international plastics-manufacturing company, I carried out a thermodynamic study of gas absorption in polyethylene (PE) with a view to improving the efficiency of PE manufacture in the gas-phase polymerisation process. The result of this study was a 30% improvement in yield of PE in bench reactor experiments. This work is featured as a Research Highlight (below)
The thermodynamic modelling of asphaltenes constitutes an increasingly topical problem in the oil industry. Since it remains an open question exactly what constitutes an asphaltene this represents a fascinating area for study. Our current emphasis is to treat asphaltene deposition as a phase-equilibrium problem. At the most fundamental level asphaltenes can be regarded as “plate-like” molecules and treatment of their phase equilibria may be likened to that of discotic liquid crystals.
At the heart of modelling fluids, whether by molecular simulation or statistical thermodynamic theory, is the potential model which underlies the representation of intermolecular forces. In mixtures of two or more different components, the potential for the interaction of unlike molecules is generally assumed to be given by averages of those of the like-like interactions: an arithmetic mean for the size parameter (the Lorentz rule); a geometric mean for the energy parameter (the Berthelot rule). When using these rules, failure to adequately model a mixture is frequently regarded as a failure of the theory; the potential is implicitly assumed to be (at least approximately) correct. This belief stems from the successful use of the Lorentz-Berthelot (LB) rules over many years, albeit for the treatment of systems of simple molecules. The theoretical origins of the LB rules can be traced back to the seminal works of London in the early 20th century, which represent the dispersion interaction of simple molecules. By pursuing an analogous derivation, but starting with more-complicated interactions (e.g., incorporating polar in addition to dispersion interactions) I have shown that the LB rules often provide a particularly poor representation of polar and, in particular, hydrogen bonding fluids, and demonstrated how to obtain more-realistic representations of the unlike intermolecular potential.
An ongoing area of my research has been the statistical-thermodynamic modelling of nematic liquid crystals with a view to obtaining an engineering eq uation of state capable of predicting the phase equilibria of this interesting and important class of compounds. This work has been the focus of the Ph.D. studies o f Mario Franco-Melgar, who recently defended his thesis. Together with Prof. George Jackson we are publishing this research in a series of papers, the first of which was recently accepted by Molecular Physics.
Solids: metallic polycrystals and the role of defects
All of the above research areas may loosely be cast under the heading fluids; half of my research output relates to this broad area. The remaining half is concerned with my interest in the role played by defects, particularly grain boundaries, in the physical characteristics of metals. This work was the focus of my research at the Argonne National Laboratory in the USA, both as a Research Associate and later, during a leave-of-absence from Imperial College, as a Visiting Scientist. Under the leadership of Dr. Dieter Wolf, I worked as part of a team studying grain growth and deformation in metals using a hierarchical approach to multiscale simulation, similar to that described earlier in the context of poly mer dynamics. One of the studies I carried out is featured as a Research Highlight.
Research highlight - THEORY
In the industrial manufacture of polyethylene (PE) in gas-phase polymerisation reactors (GPRs), reacting alkenes are introduced in gaseou s form, while the polymer produced forms a liquid phase, so that there is a vapour-liquid equilibrium (VLE) between th e gases and (liquid) PE. The catalyst, at which the reaction takes place, resides in the PE-rich phase. Consequently, in order to maximise yield of polymer, it is essential to maximise the absorption of react ing alkenes in the PE (liquid) phase. Pressure is maintained in the reactor by the introduction of non-reacting diluent gases; nitrogen, the most common, is typically present in even larger quantities than the reacting alkenes.
Using SAFT-VR, a state-of-the-art version of the Statistical Associating Fluid Theory, we modelled VLE of gas + PE mixtures. We successfully reproduced available experimental binary-mixture data; it was evident from these calculations that absorption of gas in PE rises increasingly steeply at pressures approaching the saturation (vapour) pressure of the gas. Calculations for ternary mixtures (two gases + PE), for which experimental data are very scarce, suggested that less-volatile gases enhance absorption of more-volatile gases, while more-volatile gases inhibit absorption of less-volatile gases.
If a similar effect is present in multi-component mixtures of many gases + PE, su ch as those used in the GPR process, then nitrogen, in particular, would act as an absorption inhibitor to reacting alkenes and its presence restrict the yield of polymer. Our calculations for such mixtures suggested that indeed such an effect should be expected – and that by tailoring the mixture appropriately using, for example, pentane instead of nitrogen to pressurise the reactor, absorption of reacting alkenes may be approximately doubled, which would imply a large increase in yield.
Bench-reactor experiments, based on these calculations, were carried out recently by Borealis, our industrial partner. In these experiments diluent nitrogen was partially replaced by pentane. As predicted, this change in the non-reacting gases in the mixture resulted in an enormous increase in yield of polymer; the increase in polymerisation activity was 30%. This result could have a substantial impact in industrial polymerisation of polythene using the GPR process. It also casts light on the so-called co-monomer effect, in which increase in yield of polymer is observed in the GPR polymerisation of ethene in the presence of hexene co-monomer, since hexene is a less volatile gas than ethene.
This, and related work has been published in Fluid Phase Equilibria and forms a chapter of the book Multiscale Modelling of Polymer Properties, M. Laso and E. Perpète (Eds), Elsevier, 2006 (see Publications).
Research Highlight - Simulation
In the solid state metals are polycrystalline; rather than individual, large pieces of perfect crystal, metals consist of many tiny pieces of crystal, called grains, joined together in a sort of microscopic 3D “jigsaw puzzle.” Grains are distinguished from one another by the different orientations of their crystal axes; the regions in which they meet constitute surface defects known as grain boundaries. In nature grains are not static, but grow, in so doing affecting the physical properties of the metal, making the process of grain growth one of great importance.
Using molecular dynamics (MD) I simulated a thin ï¬ÂÂÂlm of nanocrystalline palladium, a metal in which grain growth is rapid. Performing the calculations on a parallel computer enabled the simulation of a physically realisable, nanocrystalline microstructure of 25 grains with an average grain size of 15nm, containing almost 400,000 atoms for O (10) ns. I developed a suite of programs to visualise the evolving grain topology and to analyse features such as grain orientation and size.
The conventional picture of grain growth is that the process is driven by curvature-driven grain-boundary (GB) migration, reducing the total area of GBs in the material and so lowering the total excess energy. The simulations showed that in nanocrystalline materials grain rotations sometimes lead to the coalescence of neighbouring grains by eliminating the GB separating them, forming new, highly-elongated grains. Detailed analysis revealed that in grains neighbouring coalescence events, GB migration is accelerated, further enhancing the grain-growth process. These were invaluable insights, which have since been veriï¬ÂÂÂed experimentally. These and other results from the MD simulations were used to parameterise our kinetic Monte Carlo (KMC) mesoscale simulation of grain growth, a type of finite-element simulation wherein the individual atoms are “thrown away” and the microstructure is modelled instead as linked sets of “nodes” mapping out the GBs and thus the grain topology.
The KMC technique was developed to study much larger systems at far greater time scales, yet the simulations accurately reproduced the results of the MD simulations, confirming both our understanding of the underlying physics and the validity of the technique. More importantly, the simulations revealed that the competition between grain-boundary migration and grain rotation introduces a physical length scale, Rc, into the system, enabling the growth process to be characterized by two regimes. If the average grain size is smaller than Rc, as is the case in nanocrystalline materials, grain growth is dominated by the grain-rotation-coalescence mechanism. By contrast, if the average grain size is greater than Rc, then growth is dominated by curvature-driven grain-boundary migration. This is an outstanding example of successfully bridging length and time scales in simulation; moreover the “MD-to-mesoscale” scheme may be applied in any area of science in which multiple time and length scales are important.
Further details of this work may be found in Haslam et al, Mat. Sci. and Eng. A-Struct., 2001; Haslam et al, Comp. Mater. Sci., 2002 and Moldovan et al, Acta Mater., 2002 (see Publications).