Background
The Proudman Oceanographic Laboratory, part of the Natural Environment Research Council, conducts world-class research in:
- estuary, coastal and shelf sea circulation & ecosystem dynamics
- wind-wave dynamics and sediment transport
- global sea level and geodetic oceanography
- marine technology and operational oceanography
The Problem
Simulation of the marine environment is becoming increasingly important to a wide range of human activity including coastal engineering, offshore industry, fishery management, pollution monitoring, climate forecasting and leisure. Sustainable management of coastal environments requires an ability to predict a system which is mobile in three dimensions on a wide range of space and time scales.
The Challenge
The development and execution of a coupled model required efficient linking of the hydrodynamic model to the ecosystem model and the re-casting of all parts of the code into a parallel format such that the resulting coupled model would run at 1km resolution on a large enough area in a reasonable time.
While models have existed for hydrodynamic and for biogeochemical simulation, no coupled model had previously been available. Marine plant variation is becoming potentially hazardous and harmful algal blooms (‘red tides’) are well known and widespread. In this study a coupled hydrodynamic-ecosystem model is developed and applied to the simulation of seasonal variation in algal blooms around the UK coast.
The Solution
Working with the Proudman Oceanographic Laboratory, the Daresbury experts were able to construct parallel algorithms which enabled their existing hydrodynamic code (POLCOMS) to link efficiently to the European Seas Ecosytem Model (ERSEM) and to perform successful simulations at 1km resolution using Daresbury’s high performance computer HPCx. The coupled code overcame the difficulties in linking disparate phenomena by using a 3D hydrodynamic model to provide realistic physical forcing to interact with, and transport, environmental parameters. By integrating from ocean to coast the simulations were able to determine biological production and the fate of contaminants. Area plots of the time evolution of surface chlorophyll levels were carried out successfully for 12 day periods in the months of April and July.
The Benefits
- The customer was able to simulate important marine ecosystem phenomena on a major scale and at high resolution for the first time
- Computational capability was delivered to the customer in terms of code conversion to parallel operation, model coupling and hardware architecture, saving compute time and costs
- A validated model for coastal modelling is now available which can be routinely applied to predict potentially harmful phytoplankton levels. This is of great importance in maintaining marine water quality for a wide variety of commercial activities from food chain security to leisure safety assurance
The Problem
There is a growing need for a method to immobilise radioactive nuclear waste safely. In principle this can be done by encapsulation in a material which forms an effective barrier to potential release into the environment. The waste in question arises primarily from nuclear power stations, with some also from decommissioned nuclear weapons. Safe immobilisation of nuclear waste is critical to the future of the nuclear power industry, but regardless of this the amount of currently stored non-immobilised waste is already large enough to be of concern.
The Challenge
It is important to assess the consequences of the irradiation on the performance of the encapsulating medium over long timescales – up to tens of thousands of years for some isotopes. In order to study how the properties of waste forms can change over time it is necessary to perform massive parallel molecular dynamics (MD) simulations on high energy recoils in materials of interest. These simulations need to be sufficiently large in terms of number of atoms (tens of thousands) requiring significant compute capacity with parallel, distributed memory functionality.
The Solution
Using the Daresbury MD code DL-Poly3, simulations were performed on the UK supercomputer located at Daresbury - HPCx. A highly energetic particle is driven into the system giving rise to cascades of displaced atoms in the encapsulating material. The study covered materials currently proposed for waste containment (glasses) and also new ceramic forms such as zirconium silicate (shown), which promise considerably higher durability. The recoils studied are designed to simulate an alpha decay event during which a heavy recoil causes extensive damage in the structure, resulting in several thousand permanently displaced atoms.
Molecular simulation of zirconium silicate after a highly energetic particle has been driven into the system
The Benefits
- The simulations demonstrate that the permanent residual damage to the containment material can be limited to a few thousand atoms. This is important to the customer when comparing encapsulation materials
- The customer can use such simulations to point the way to the development of novel, high performance containment materials
- The simulations are only possible due to the benefits available from high performance
parallel computing