A wind power forecast is the expected production of one or more wind turbines in the near future. Usually power is forecasted for periods between 1 and 48 hours ahead, with the emphasis on the next day, but forecasts at a range as short as 15 minutes are made as well. Longer range forecasts (3 to 9 days), on the other hand, can be made but are not used. Wind power forecasts are comparable to weather forecasts, but then expressed in power (in kilowatts) or energy (kilowatthour) instead of temperature or precipitation. Another difference is that wind power forecasts are issued as a continuous series of 15 minutes values.

Contents 1 Weather forecasts for wind turbines 2 Atmospheric models 3 Physical versus statistical methods 4 Why wind power forecasts?

Weather forecasts for wind turbines

The weather is described by several physical quantities. Four of them are needed to forecast wind power: wind speed, wind direction, temperature and pressure. In addition, and here comes the main difference with weather forecasting, the wind turbine power curve plays a role. This power curve gives the relation between wind speed at hub height and power production of a wind turbine.

A wind power forecast is constructed as follows:

• A weather forecast gives the expected wind speed at hub height,
• The turbine power curve gives the expected power at the forecasted wind speed for the standard value of the air density,
• The weather forecast allows to calculate the expected air density via the expected temperature and the expected pressure,
• The expected power is corrected for the difference between the standard and the expected air density.

_________________________________________________________________

So there are 3 factors that influence wind power: wind speed, power curve and air density.

Weather forecasts are issued by meteorological institutes. For example, in the Netherlands KNMI publishes 4 times per day expected values of wind speed, wind direction, temperature and pressure for the period the between 0 and 48 hours after initialisation of the atmospheric model Hirlam with measured data. Since it takes some time to verify the measurements, to run the model and to check and distribute the output there is a gap of about 4 hours between the moment the measurements were acquired and the model output is available to the user. This gap is a blind spot in the forecasts from an atmospheric model.

Successive forecasts can alter significantly if the weather is changing over time. It is recommended always to use the latest forecasts.

Atmospheric models

Meteorological institutes apply atmospheric models for weather forecasts on short and medium term periods. An atmospheric model is a numerical approximation of the physical description of the state of the atmosphere in the near future, and usually is run on a supercomputer. Each computation starts with initial conditions originating from recent measurements. The output consists of the expected average value of physical quantities at various vertical levels in a horizontal grid and stepping in time up to several hours after initiation.

There are several reasons why atmospheric models only approximate reality. First of all, not all relevant atmospheric processes are included in the model. Also, the initial conditions may contain errors (which in a worse case propagate), and the output is only available for discrete points in space (horizontal as well as vertical) and time. Finally, the initial conditions age with time - they are already old when the computation starts let alone when the output is published. By these limitations atmospheric models are only an aid to weather forecasts. (In addition to atmospheric models meteorologist use other methods.)

Many different atmospheric models are available, ranging from academic research tools to fully operational instruments. Besides for the very nature of the model (physical processes or numerical schemes) there are some clear distinctive differences between them: time domain (from several hours to 6 days ahead), area (several 10.000 km² to an area covering half the planet), horizontal resolution (1 km to 100 km) and temporal resolution (1 hour to several hours).

One of the atmospheric models is the High Resolution Limited Area Model, abbreviated HiRLAM, which is frequently used in Europe. HiRLAM comes in many versions, that’s why it is better to speak about "a" HiRLAM rather than "the" HiRLAM. Each version is maintained by a national institute such as the Dutch KNMI, the Danish DMI or Finnish FMI. And each institute has several versions under her wing, divided into categories such as: operational, pre-operational, semi operational and for research purposes.

Other atmospheric models are UKMO in the UK, Lokalmodell in Germany, Alladin in France (Alladin and Lokalmodell are also used by some other country’s within Europe), and MM5 in the USA.

In order to make weather forecasts up to 48 hours ahead KNMI uses a rather rough operational HiRLAM (horizontal grid points 22 km apart). In addition, for 24 hours forecasts KNMI operates a (more refined) HiRLAM with grid points 11 km apart and more vertical levels, whereas higher resolution versions are available for research. In all cases, the forecasts consist of the hourly or 3-hourly expected value of the 10-minute average of a quantity.

Physical versus statistical methods

The forecasted physical quantities are valid on a standard level and in a grid point, and have to be translated to turbine hub height and turbine location. To this end local roughness and atmospheric stability are taken into account. Furthermore, the hourly or 3-hourly 10-minute averages (meteorological standard) must be converted into a continuous series of 15-minute averages (standard in the electricity sector). There are 2 essentially different conversion approaches available: physical and statistical.

The physical method consists of several sub-model that altogether deliver the translation from the wind forecast at a certain grid point and model level, to the forecasted energy of one or more wind turbine(s). Every sub-model contains the mathematical description of the physical processes relevant to the translation. For example a sub-models treats the effect of local roughness and atmospheric stability on the wind speed. Also wake interaction in a wind farm, where several wind turbines may interfere with each other, is put into a sub-model. Knowledge of all relevant processes is therefore important when constructing a pure physical model (such as the early versions of the Danish Prediktor).

The statistical method also consists of sub-models for the translation from grid point and model level to local energy, but now based on the mathematical description of estimators of the relevant quantities. Since the parameters in the estimative relations are not universal these have to be obtained from measurements, for example with recursive least square estimators or neural networks. For a purely statistical model (such as the early versions of the Danish WPPT) it is necessary to have a continuous stream of observations in order to keep the parameter values up to date.

In practise there is no such thing as a purely physical or statistical model. For example, a physical model often has a statistical sub-model in order to make corrections for systematic errors. Later versions of the Danish Prediktor, the German Previento, the Dutch AVDE or the American eWind are examples of the mixed method. On the other hand, parameters in a statistical model may be obtained from mathematical formulations describing a physical process. This is the case in for instance the Danish Zephyr, later versions of the Danish WPPT and the German AWPT.

In the Netherlands the translation from the HIRLAM is offered by the meteorological service providers Meteo Consult (together with Ecofys) and Aeolis Forecasting Services. In addition ECN performs such translations for research or evaluation purposes.

Why wind power forecasts?

In the electricity grid at any moment balance must be maintained between electricity consumption and generation - otherwise disturbances in power quality or supply may occur. Balance on time scales shorter than 15 minutes is the responsibility of a System Operator (SO). Balance on time scales of 15 minutes and longer, on the other hand, is the responsibility of market parties. The 15-minute period usually is referred to as the Programme Time Unit (PTU), whereas a market party is also known as a Programme Responsible Party (PRP).

Balancing of the 15-minute averaged power is required from all electrical producers and consumers connected to the grid, who for this purpose may be organised in sub-sets. Since these sub-sets are referred to as Programmes, balancing on the 15-minute scale is referred to as Programme Balance. Programme Balance now is maintained by using production schedules (issued the day before delivery) and measurement reports (distributed the day after delivery). When the measured power is not equal to the scheduled power, there is so-called Programme Imbalance:

Programme Imbalance
is
Realised sum of production and consumption
minus
Forecasted sum of production and consumption.

Programme Imbalance is settled by the System Operator, with different tariffs for negative Programme Imbalance and positive Programme Imbalance. What therefore counts is the absolute value of the Programme Imbalance.

If only production from wind energy is taken into account, Programme Imbalance reduces to:

Programme imbalance by wind energy
is
Realised wind production  
minus  
Forecasted wind production.

In the case of a positive Programme Imbalance by wind energy the realised wind production is bigger than the forecasted wind production. And vice versa, in the case of a negative Programme Imbalance by wind energy.

If all other strategies to control Programme Imbalance are not considered, Programme Imbalance due to wind energy boils down to:

Programme imbalance by wind energy
is
Wind production forecast error.

Because of the asymmetrical tariffs of Programme Imbalance, the absolute value of the wind power forecast error would tell whether or not such a forecast is good. This means that it is not sufficient to forecast well on average - the probability of a hit must be large. In order to have a small wind Programme Imbalance, it is therefore necessary to have both a small systematic forecast error and a small random forecast error.

So far the terminology of the Dutch set-up of balance control has been used. This set-up identifies clusters of producers and consumers, and is equal to the set-up in other countries. In the Netherlands the set-up however is effectuated in a different way: PRP's are responsible for wind energy whereas everywhere else this responsibility lays by a SO.