News item | 2014-10-03 | 15:16 PM

More accurate weather forecasts with models

The weather is becoming increasingly important to many of Vattenfall's operations. Managing the weather is of course impossible, but an R&D programme will aid the company to manage the impact of weather conditions.

In 2011, a programme was initiated with the purpose of developing the capability to run in-house forecasts of weather and wind power production. "We use numerical weather models that represent the complex physics of the atmosphere," says Jens Madsenat Vattenfall.

Today, the internal forecasts are run twice daily for all of Northern Europe, with high resolution predictions that zoom in on areas where Vattenfall's wind farms are concentrated. But within short more frequent and specific forecasts will be run according to Madsen: "When we run our weather models today, we normally use a three kilometre resolution. That is sufficiently accurate for offshore predictions, but in due course we will be using resolutions of one kilometre or less to capture more of the weather effects that occur onshore due to local terrain variations."

In-house data

Madsen underscores the advantage of having a thousand physical measurement stations in the shape of Vattenfall's fleet of wind turbines. "Vattenfall's turbines record important weather parameters that we use to improve our forecasts. In particular the short-term forecasts, what happens over the next 24 hours, improve a lot when we use in-house data.

We are able to customise the service for different purposes, e.g. to help Vattenfall gain a competitive edge when it comes to balancing the intermittent production of wind power".

Competitive edge

Weather conditions are not only important for Vattenfall's wind business but also for trading, asset optimisation, hydro, and even nuclear.

"The forecast outputs are currently being evaluated by different parts of Vattenfall, and we are confident it will be replacing or complementing some of the externally bought weather services that Vattenfall is using in the near future" Madsen says.

He adds that predictions from the same weather models also are being used to replace physical met masts with virtual ones. "That is an advantage because physical masts are expensive and it saves lots of time that you would otherwise spend waiting years for measurements campaigns to be completed."