A very brief and necessarily limited overview of the state-of-the-art in hydrate molecular simulation will be presented below. Unfortunately, it is not possible to describe these areas in any satisfactory level of detail in this short review with limited references; in many cases below, the names of those who have contributed to particular fields of study are given rather than listing out many references.
The use of various water models, including polarisable potentials, has been investigated for methane hydrate by English and MacElroy and also Jiang and Jordan. In their studies, they found that the TIP4P, TIP4P-FQ and recent Drude-based charge-on-spring models were effective. However, specific guest-host parameterisation of potentials for hydrates or the general guest-host fluid system has been performed less often, but Anderson and Trout, Mastny and de Pablo, and Jordan’s group have made valuable contributions here.
There have been many developments in revising and improving the statistical thermodynamic model of van der Waals and Platteeuw since the 1970’s, with Holder, Prausnitz, Tse, Rodger and Tanaka active in this field. More recently, Anderson and Trout , and Monson have refined the statistical mechanics models with valuable input from molecular simulation.
Many structural and dynamic equilibrium properties of a wide variety of type I, II and H hydrates have been studied by MD and Monte Carlo techniques by a wide variety of researchers since the 1980’s, following the initial work of Tse et al . These have included radial distribution functions and density of states, with comparison to experimental data whenever possible. As mentioned previously, most equilibrium properties of hydrates are relatively similar to those of ice, with the exception of thermal conductivity. There have been a few MD studies into the anomalous thermal conductivity of clathrates using both equilibrium and non-equilibrium approaches, chiefly by Inoue and Tanaka, Jiang and Jordan, English, and Tse . These studies have had mixed results with respect to agreement with experimental data, but have helped to develop insights into resonance scattering effects of guests on the propagation of thermal phonons in the hydrate lattice.
Molecular simulation has also contributed to the investigation of the viability of carbon dioxide sequestration in hydrates, and Peters and Trout, and Chialvo and co-workers  have been particularly active in this area. Developments in free energy estimation methods and transition state theory have been used by these workers to assess the methane diffusivity in hydrates by ‘jumping’ between cages with water vacancies present, and ability to replace methane in hydrates by carbon dioxide from a high-pressure phase in contact with the hydrate.
Hydrate nucleation has also been studied by MC and MD techniques by Rodger and co-workers, Tse, Radhakrishnan, Anderson and Trout . Molecular simulation has led to the development of an order parameter approach to probe the free energy surface of available states, and has advanced the view that instantaneous hydrate-like molecular configurations from random thermal motion lead to sudden, unpredictable, nucleation events (the ‘local structure’ model), which appears consistent with available experimental data. These simulation-based insights have helped to assess the validity or otherwise of earlier, speculative models such as the cluster and gas phase nucleation hypotheses , underlining the value of molecular simulation in providing insights into fundamental molecular mechanisms of hydrate behaviour.
The growth of hydrate crystals is another area where MD simulation has had a particular impact, with contributions from Báez and Clancy, Rodger and co-workers, English and MacElroy, and Kusalik and co-researchers . Although growth is likely to take place at the water-methane interface, most studies before that of Zhang et al  concentrated on aqueous solutions highly supersaturated in methane (or another guest) in contact with the clathrate hydrate, so this work, which does consider methane-water interfaces, is of particular relevance and interest. Molecular simulation has proven particularly useful to modelling the diffuse hydrate-liquid boundary and for identifying the importance of transport of methane to the interface in allowing further hydrate growth. Numerous interesting studies of Kusalik and co-workers have also identified the occasional formation of different methane hydrate structure types (II) on existing templates (type I), with water vacancies and empty as well as doubly-occupied cages. The large activity in recent years in MD studies of hydrate crystallisation has been highly valuable in identifying underlying molecular mechanisms and adding to our understanding of polymorphism. Anderson and Trout and also Rodger and co-workers have also been very active in recent years in the study of inhibition mechanisms in hydrate growth by the presence of chemical additives, which is of particular relevance to flow assurance in the petrochemical industry. The recent study of Zhang et al.  has also probed the influence of inhibitors and temperature on growth or dissolution from hydrate crystallites at water-methane interfaces, and captures quite realistically the full complexity of conditions for both inhibitory and non-inhibitory clathrate crystallisation.
Decomposition of hydrates has also been studied using MD simulation by Báez and Clancy, Rodger and co-workers, English and co-researchers, Myshakin et al . The driving forces were either conventional thermal heating or exposure to electromagnetic fields, in an effort to model earlier descriptions by experiment, industrial practice, or reservoir-level macroscopic simulation [1-3]. In these MD studies , the importance of cage occupancy, interfacial relaxation, over-temperature vis-à-vis the hydrate melting point and methane diffusion through the liquid interfacial layers was highlighted, and these studies have complemented our understanding of hydrate break-up. This is of importance to optimising approaches for large-scale methane exploitation in ocean depths, and to determining its operational (and, ultimately, economic) viability. The inhibition effect has been studied by Tse ; it was suggested the formation of an ice layer on the surface of the decomposed hydrate delayed the kinetics of decomposition.
In all of the outlined molecular simulation studies of nucleation, growth, break-up and inhibition, the memory effect often plays an important underlying role. It is fair to state that there is currently little comprehensive understanding in the scientific community about this, and it is the opinion of many in the field of hydrate simulation that this requires substantial further investigation [12-14]. It is likely that simulation can contribute more to our understanding of this effect, and to investigate hypotheses further.
It is perhaps appropriate to mention just a few other interesting recent trends and developments in hydrate simulation. For instance, these concern the study of new hydrate formers, such as those of structure H by Alavi and co-workers, and new hydrate structures and insights into polymorphism by Kusalik and co-investigators. Jordan and co-workers are currently investigating DFT approaches in hydrates and fitting classical potentials, and there is increasing industry-based activity in hydrate molecular simulation.