What if a 50-year, site-specific wave hindcast for offshore wind design took just over a day instead of a week? That is what we are working towards at MetOceanWorks, with a significant investment in new modelling infrastructure built around the way our models actually run.

Reliable metocean data underpins almost every offshore wind design decision, from extreme wave and fatigue loads to foundation sizing, cable routing and installation windows. A good design-basis hindcast means high-resolution modelling of the site and its local bathymetry over several decades, validated against measured data, refined and re-run. Every iteration builds confidence, but every iteration also requires significant computing power.

We are moving from a conventional HPC setup to purpose-built servers, designed around where the time in a hindcast really goes. Key highlights include:

• GPU acceleration for hydrodynamics, using NVIDIA H200 GPU well suited to our flexible-mesh models
• Hundreds of CPU cores for spectral wave modelling, with SWAN being recompiled for the new processors and each model allocated dedicated cores
• A dedicated job queue that keeps processor capacity efficiently utilised and gives priority to time-critical project work
• Striped enterprise NVMe storage, easing the write bottleneck that can account for up to half the runtime of output-heavy runs
• Standardised NetCDF output, streamlining data handling and analysis
• Professional data centre hosting with redundant power, a gigabit connection and secure remote access, plus cloud capacity when a project needs more

We expect output-heavy 50-year hindcasts to drop from around seven days to just over one, with less storage-intensive runs about three times faster. Every model run will log its own timings, allowing us to measure performance, identify bottlenecks and continually improve how we use the system.

"Quality in metocean work comes from iteration: calibrating, testing sensitivities and validating against measurements until the data can be relied on," said Andrew Watson, Managing Director at MetOceanWorks. "This investment means we can do more of that within the same programme, or deliver projects faster when deadlines are tight. The less time we spend waiting on models, the more time we can spend on analysis."

If you are planning metocean work for an offshore wind project, we'd be happy to discuss what this additional modelling capacity could mean for your programme.