Weather simulation at the HornsRev offshore wind farm in Denmark. Blue shades indicate 10m wind speeds. The turbine wake effects are visible.


then looked into the possibilities of ‘translating’ the Eddy model to work on GPUs. “It was amazing – around 100 times faster than normal super computers.”


The company’s atmospheric simulations are designed to run on massively parallel supercomputing platforms and can exploit the spectacular acceleration offered by GPUs. “This breakthrough then made Whiffle’s practical model applications possible in 2015.”


Remco admits that embedding the technology wasn’t entirely without problems at first. “Of course predicting the weather of tomorrow is extremely complex, with all kinds of chaotic phenomena coming into play. We still have a great respect for the accuracy of global weather models. It is not easy to beat state-of-the-art models - the bar is set so high. But at some point we realised our ultra-high resolution approach could improve upon these traditional models and we could beat them!”


Ultra-high resolution at small-scale The key difference is the ultra-high resolution Whiffle uses for its finecasting, Remco stresses. This facilitates an accurate and natural representation of small-scale processes such as turbulence, surface


interactions, cloud formation and precipitation at a certain location and time with unprecedented spatial and temporal detail. The massive computer power offered by the GPUs is crucial for weather predictions over the desired area at a resolution of 100 m or finer. It can actually go to 50 m or even 10 m. For example, it can perfectly capture phenomena that one wind turbine has on another one. Small cumulus clouds of 1 km would be impossible to capture by a traditional model with a resolution of 10 km. “We can represent the laws of physics in a very direct way. We totally mimic nature. Our atmospheric simulations do not reduce reality anywhere near as much as the traditional weather models do.”


Customers include many Dutch wind farms Whiffle was ready to send data to its first customer in the summer of 2016. Then swiftly afterwards the company started forecasting for the Gemini Wind Park, which is the largest offshore wind farm in the Netherlands. The finecasting system first uses a pre-processing engine to convert large-scale weather data into customised input fields. Forecasting wind energy production depends on the ability to predict wind speeds at the height of the turbine blades. Whiffle’s ultra-high


“ We have a 3D model of weather parameters like turbulence, cloud formation, wind speed and surface interactions”


resolution atmospheric model is fully geared towards an accurate modelling of local wind conditions, making it ideal for wind energy forecasts. For example a wind farm operator then knows what it can expect to produce the next day. Most wind farms use the information for trading energy on the day- ahead markets. They can predict how many megawatt hours they will produce, so they will not be over or under selling their generation levels.


Olympic team In addition to the energy sector, Whiffle believes that it can develop a widely applicable model for other markets. Together with the Sailing Innovation Centre, it is currently applying its model for the Dutch Olympic sailing team to simulate Tokyo Bay, which will be used for the 2021 games. Additionally, the technology can be used to model the dispersion of pollutants through the atmosphere.


report 7


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28