Variables, Margins and Their ImpactThe International Energy Agency’s Wind Technology Collaboration Programme Task 19 ‘International Recommendations for Ice Fall and Ice Throw Risk Assessments’, published in October 2018, gives a comprehensive overview of the necessary parts of a risk assessment and will hopefully form the basis for a future standard. Although it was obviously created with great care, the variables involved still leave considerable leeway for the results.
By Markus Drapalik, Institute of Safety and Risk Sciences, University of Natural Resources and Life Sciences, Vienna
Why Ice Risk Assessments?
For wind turbines operating in cold climates, aggregation of ice and its subsequent shedding is part of normal operation. Depending on local laws and regulations an assessment of the risk, which results from the aggregated ice being shed from a turbine at standstill or thrown from the moving rotor blades, may be necessary or at least a reasonable precaution. In some regions, ice detection and subsequent stopping of the turbine is mandatory. Even in these cases, large ice amounts of several hundred kilograms can be shed in a single event, where single ice fragments may have a weight of 10kg or more. A comprehensive risk assessment, which generates a risk map for the surroundings of the turbine, creates legal security and helps in effective risk reduction measures.
For wind turbines operating in cold climates, aggregation of ice and its subsequent shedding is part of normal operation. Depending on local laws and regulations an assessment of the risk, which results from the aggregated ice being shed from a turbine at standstill or thrown from the moving rotor blades, may be necessary or at least a reasonable precaution. In some regions, ice detection and subsequent stopping of the turbine is mandatory. Even in these cases, large ice amounts of several hundred kilograms can be shed in a single event, where single ice fragments may have a weight of 10kg or more. A comprehensive risk assessment, which generates a risk map for the surroundings of the turbine, creates legal security and helps in effective risk reduction measures.
What to Include in an Ice Risk Assessment?
Following the International Energy Agency (IEA) recommendations, an assessment consists of at least two parts: an assessment of meteorological risks and an assessment of risks due to the physical throw of ice fragments. Additionally, further risk influencing factors may be included. The relevant outcome for the meteorological assessment is how much ice is shed in a year. This is determined by the icing frequency (how often does ice form on the turbine) and the icing intensity (how much ice is aggregated in a single event), if possible considering ice shedding and reaggregation during a single event. The physical model is usually a ballistic model which calculates the throw distances of single fragments. This is combined with a Monte Carlo simulation which generates several hundred thousand throws. The resulting simulated distribution on the ground is used to generate a risk contour map.
Following the International Energy Agency (IEA) recommendations, an assessment consists of at least two parts: an assessment of meteorological risks and an assessment of risks due to the physical throw of ice fragments. Additionally, further risk influencing factors may be included. The relevant outcome for the meteorological assessment is how much ice is shed in a year. This is determined by the icing frequency (how often does ice form on the turbine) and the icing intensity (how much ice is aggregated in a single event), if possible considering ice shedding and reaggregation during a single event. The physical model is usually a ballistic model which calculates the throw distances of single fragments. This is combined with a Monte Carlo simulation which generates several hundred thousand throws. The resulting simulated distribution on the ground is used to generate a risk contour map.Where Is the Ice?
Several different icing models are currently in use. They have been consequently improved so that reliable predictions of production losses due to icing are possible. These models are consequently suitable for the calculation of icing frequencies. Icing intensity, however, a driving parameter in ice throw risk assessments, is much more difficult to model. Thus, it is recommended to use a predefined ice distribution on the rotor blade, such as the IEC load case. This distribution puts most ice on the outer part of the rotor, which is useful for load testing but is not realistic. Observations show that icing varies with meteorological conditions, blade speed and angle of attack. Depending on the assumed distribution of ice on the blade, the resulting risk is more concentrated towards the tower or more evenly spread out. Therefore, it is not possible to choose the ‘right’ distribution with the current state of knowledge.
Several different icing models are currently in use. They have been consequently improved so that reliable predictions of production losses due to icing are possible. These models are consequently suitable for the calculation of icing frequencies. Icing intensity, however, a driving parameter in ice throw risk assessments, is much more difficult to model. Thus, it is recommended to use a predefined ice distribution on the rotor blade, such as the IEC load case. This distribution puts most ice on the outer part of the rotor, which is useful for load testing but is not realistic. Observations show that icing varies with meteorological conditions, blade speed and angle of attack. Depending on the assumed distribution of ice on the blade, the resulting risk is more concentrated towards the tower or more evenly spread out. Therefore, it is not possible to choose the ‘right’ distribution with the current state of knowledge.
How Far Can It Be Thrown?
The most established ballistic model uses two parameters to characterise a single ice fragment: mass and size. An additional parameter for drag resistance could be used but a fixed value is usually chosen. Thus, the actual geometry of the fragment is not considered, which subsequently ignores autorotation of the fragments as well as lift and turbulence along the fragment.
The most established ballistic model uses two parameters to characterise a single ice fragment: mass and size. An additional parameter for drag resistance could be used but a fixed value is usually chosen. Thus, the actual geometry of the fragment is not considered, which subsequently ignores autorotation of the fragments as well as lift and turbulence along the fragment.
A direct comparison between experiments that the institute conducted with specimens reproduced from found ice fragments and simulations using this simple ballistic model shows an underestimation of up to 30% for shed distances. Similar experiments for ice throw from small wind turbines result in errors of 40% over- and underestimation.
When Is an Ice Fragment Dangerous?It is obvious that ice fragments below a certain threshold can be ignored, since they will not inflict serious harm. One possibility is a hard 40J limit for the impact energy. Another choice is a continuous probit function, which attributes an increasing ‘deadliness’ of up to 1 to a fragment. This tries to include injuries in a reasonable manner. Depending on the choice of limit, certain ice fragments contribute to the total risk or not.
This is strongly linked to the choice of fragment mass and size distribution. Different studies have found different distributions for these parameters, possibly depending on the local conditions. Again, the choice of distribution changes the resulting risk map.
Risk Reduction Factors Make the Difference
Adhering to the ALARA (As Low As Reasonably Achievable) principle, it is sensible to install additional safety measures such as warning signs, warning lights or physical barriers. It can be assumed that these measures result in a further risk reduction, since fewer people will go near an iced turbine. This is incorporated in the results using risk reduction factors (RRFs), which are applied globally on all calculated risk values. Depending on the measure, RRFs of up to 100 are proposed, meaning a reduction of the risk by two orders of magnitude. Due to the lack of corresponding studies, these factors are solely expert judgements and cannot consistently be applied.
Adhering to the ALARA (As Low As Reasonably Achievable) principle, it is sensible to install additional safety measures such as warning signs, warning lights or physical barriers. It can be assumed that these measures result in a further risk reduction, since fewer people will go near an iced turbine. This is incorporated in the results using risk reduction factors (RRFs), which are applied globally on all calculated risk values. Depending on the measure, RRFs of up to 100 are proposed, meaning a reduction of the risk by two orders of magnitude. Due to the lack of corresponding studies, these factors are solely expert judgements and cannot consistently be applied.
There Is No Worst CaseConsidering the uncertainties arising from a lack of knowledge, it seems that the prudent choice would be to generate a worst case with large ice loads and big fragments on the outer parts of the rotor. Unfortunately, this is not a sensible option for several reasons. Fragments thrown further are distributed over a larger area and thus the local risk is lower, while larger fragments are more dangerous but are thrown less far. Additionally, relevant structures, such as roads or cross-country ski tracks, may be near the turbine and need to be considered. Furthermore, apart from the risk to the public, the risk to service personnel also needs to be considered. In this case, ice fragments thrown near the turbine are relevant. In summary, constructing a worst case is not possible. Thus, it is necessary to generate a risk map which is as realistic as possible.
Conclusion
The IEA recommendations present a useful framework for ice throw and shed risk assessments but are greatly restricted by the current state of knowledge. Using the same data and the best methods available, different researchers may still produce results differing by several orders of magnitude. This cannot be seen as a reason to reject the use of assessments but rather to treat the results with caution. The ALARA principle demands (very simplified) that additional safety measures should be applied as long as the cost–benefit analysis is reasonable. Thus, easy measures for risk reduction, like warning signs, should always be installed regardless of risk assessment results. Finally, there is a moral obligation to ensure the highest reasonably possible safety at the current state of knowledge, which cannot be quantified.
The IEA recommendations present a useful framework for ice throw and shed risk assessments but are greatly restricted by the current state of knowledge. Using the same data and the best methods available, different researchers may still produce results differing by several orders of magnitude. This cannot be seen as a reason to reject the use of assessments but rather to treat the results with caution. The ALARA principle demands (very simplified) that additional safety measures should be applied as long as the cost–benefit analysis is reasonable. Thus, easy measures for risk reduction, like warning signs, should always be installed regardless of risk assessment results. Finally, there is a moral obligation to ensure the highest reasonably possible safety at the current state of knowledge, which cannot be quantified.
Biography
Markus Drapalik is a physicist at the Institute of Safety and Risk Sciences at the University of Natural Resources and Life Sciences, Vienna. He has been working on wind turbine icing for the last seven years and received his PhD for his work on the risk of ice shed from turbines.
Markus Drapalik is a physicist at the Institute of Safety and Risk Sciences at the University of Natural Resources and Life Sciences, Vienna. He has been working on wind turbine icing for the last seven years and received his PhD for his work on the risk of ice shed from turbines.
