IEEE Power & Energy Magazine - November/December 2017 - 48

✔ contract and activate flexibility from distributed en-

ergy resources to solve technical problems (e.g., congestion and voltage problems).
For DSOs, the forecasting challenges are mainly related to
scalability requirements associated with a high number of
time series to forecast and the need to have a spatiotemporal
representation of forecast uncertainty, because the technical
problems are primarily local.

The Value of Forecasts
In the context of uncertainty forecasting, traditional metrics such
as mean absolute or square errors provide only partial information about forecasting skill (or quality). Uncertainty forecasts
represented by probability distributions can be evaluated by
matching empirical and nominal probabilities (called calibration or reliability), size of forecast intervals (called sharpness),
and the ability to provide different conditional probability distributions, depending on the level of the prediction (called resolution). These properties are also used to evaluate the forecasting
skill of NWP ensembles and statistical scenarios.
But new metrics with the ability to discriminate these
properties are still needed to assess the modeling of the
dependence structures. In power industry practice, however,
the more complicated metrics are often too time-consuming
for end users to establish. Another problem is that such metrics often do not measure the value of the forecast for the end
user. More useful, for instance, would be a cost function to
determine how much reserve may be avoided by using probabilistic methods to identify reserve requirements with a longer time horizon. For other applications, such as high-speed
cut-out events where the correct prediction of up-ramping
or down-ramping and the probability of shutting down large
parts of the wind-generating units is essential to avoid grid
stability issues, the value of such predictions lies in avoiding
costs and maintaining grid security.

Is Reserve Allocation Based
on a Forecast?

No

Allocation > One Week?

Yes
No

Allocation > One Week?

Another way to assign a cost value to forecasts is to apply
full production cost modeling, both with and without forecasts.
The difference provides users with an estimate of the
actual monetary cost difference of having the forecast. It is,
however, difficult to distinguish between the effectiveness
of employing only deterministic forecasts versus also taking advantage of the probabilistic information from uncertainty forecasts.
Even traditionally deterministic tasks, like bidding into
the power markets, are no longer captured by standard statistical metrics that measure the success or failure of a forecast. Probabilistic forecasts enable traders to make decisions
based on the expected forecast accuracy and the expected
imbalance costs and thereby avoid costs that reduce their
overall income. So, even though single values are what a
trader has to bid in with, the decision regarding which value
to choose is, today, often the result of a probabilistic forecast.
The correctness of the bid is not only the difference between
the bid and real production at that time but also the cost at
that time. For example, in some markets, traders can earn on
imbalances in their production if the imbalance reduces the
system imbalance.
The challenge, therefore, is to develop problem-specific
metrics that link forecast quality with value. In other words,
we need to understand how calibration and sharpness of
uncertainty forecasts effect markets and whether forecast
quality improvements are cost-effective. In addition, integrating the decision maker's risk profile (e.g., risk averse,
prone, or neutral) in the evaluation of forecast value is indispensable to avoid incorrect assessments.
So, if the value of a forecast no longer has one specific
measure or metric that provides a sufficient means of evaluation, we may have to make a paradigm shift and consider
value in terms of understanding which approach to apply
for a specific problem and how to optimally make use of the

No

Can the Correct Choice
of Forecast Help?

Yes
No

Yes
Yes

How is the Cost
Development: Increasing?

No

Optimized System

No
Yes

Is a Probabilistic Forecast
Applied?
No

Yes
Is the Correct Method
Applied for the Problem
at Hand?

Yes

Other Nonweather-Related
Aspects to be Considered

figure 7. An example of a decision-support scheme for dynamic reserve allocation for a system operator with the target
requirements that 1) allocation needs to be carried out at least one week ahead of time and 2) the cost for allocated reserve is the optimization parameter.
48

ieee power & energy magazine

november/december 2017



Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - November/December 2017

IEEE Power & Energy Magazine - November/December 2017 - Cover1
IEEE Power & Energy Magazine - November/December 2017 - Cover2
IEEE Power & Energy Magazine - November/December 2017 - 1
IEEE Power & Energy Magazine - November/December 2017 - 2
IEEE Power & Energy Magazine - November/December 2017 - 3
IEEE Power & Energy Magazine - November/December 2017 - 4
IEEE Power & Energy Magazine - November/December 2017 - 5
IEEE Power & Energy Magazine - November/December 2017 - 6
IEEE Power & Energy Magazine - November/December 2017 - 7
IEEE Power & Energy Magazine - November/December 2017 - 8
IEEE Power & Energy Magazine - November/December 2017 - 9
IEEE Power & Energy Magazine - November/December 2017 - 10
IEEE Power & Energy Magazine - November/December 2017 - 11
IEEE Power & Energy Magazine - November/December 2017 - 12
IEEE Power & Energy Magazine - November/December 2017 - 13
IEEE Power & Energy Magazine - November/December 2017 - 14
IEEE Power & Energy Magazine - November/December 2017 - 15
IEEE Power & Energy Magazine - November/December 2017 - 16
IEEE Power & Energy Magazine - November/December 2017 - 17
IEEE Power & Energy Magazine - November/December 2017 - 18
IEEE Power & Energy Magazine - November/December 2017 - 19
IEEE Power & Energy Magazine - November/December 2017 - 20
IEEE Power & Energy Magazine - November/December 2017 - 21
IEEE Power & Energy Magazine - November/December 2017 - 22
IEEE Power & Energy Magazine - November/December 2017 - 23
IEEE Power & Energy Magazine - November/December 2017 - 24
IEEE Power & Energy Magazine - November/December 2017 - 25
IEEE Power & Energy Magazine - November/December 2017 - 26
IEEE Power & Energy Magazine - November/December 2017 - 27
IEEE Power & Energy Magazine - November/December 2017 - 28
IEEE Power & Energy Magazine - November/December 2017 - 29
IEEE Power & Energy Magazine - November/December 2017 - 30
IEEE Power & Energy Magazine - November/December 2017 - 31
IEEE Power & Energy Magazine - November/December 2017 - 32
IEEE Power & Energy Magazine - November/December 2017 - 33
IEEE Power & Energy Magazine - November/December 2017 - 34
IEEE Power & Energy Magazine - November/December 2017 - 35
IEEE Power & Energy Magazine - November/December 2017 - 36
IEEE Power & Energy Magazine - November/December 2017 - 37
IEEE Power & Energy Magazine - November/December 2017 - 38
IEEE Power & Energy Magazine - November/December 2017 - 39
IEEE Power & Energy Magazine - November/December 2017 - 40
IEEE Power & Energy Magazine - November/December 2017 - 41
IEEE Power & Energy Magazine - November/December 2017 - 42
IEEE Power & Energy Magazine - November/December 2017 - 43
IEEE Power & Energy Magazine - November/December 2017 - 44
IEEE Power & Energy Magazine - November/December 2017 - 45
IEEE Power & Energy Magazine - November/December 2017 - 46
IEEE Power & Energy Magazine - November/December 2017 - 47
IEEE Power & Energy Magazine - November/December 2017 - 48
IEEE Power & Energy Magazine - November/December 2017 - 49
IEEE Power & Energy Magazine - November/December 2017 - 50
IEEE Power & Energy Magazine - November/December 2017 - 51
IEEE Power & Energy Magazine - November/December 2017 - 52
IEEE Power & Energy Magazine - November/December 2017 - 53
IEEE Power & Energy Magazine - November/December 2017 - 54
IEEE Power & Energy Magazine - November/December 2017 - 55
IEEE Power & Energy Magazine - November/December 2017 - 56
IEEE Power & Energy Magazine - November/December 2017 - 57
IEEE Power & Energy Magazine - November/December 2017 - 58
IEEE Power & Energy Magazine - November/December 2017 - 59
IEEE Power & Energy Magazine - November/December 2017 - 60
IEEE Power & Energy Magazine - November/December 2017 - 61
IEEE Power & Energy Magazine - November/December 2017 - 62
IEEE Power & Energy Magazine - November/December 2017 - 63
IEEE Power & Energy Magazine - November/December 2017 - 64
IEEE Power & Energy Magazine - November/December 2017 - 65
IEEE Power & Energy Magazine - November/December 2017 - 66
IEEE Power & Energy Magazine - November/December 2017 - 67
IEEE Power & Energy Magazine - November/December 2017 - 68
IEEE Power & Energy Magazine - November/December 2017 - 69
IEEE Power & Energy Magazine - November/December 2017 - 70
IEEE Power & Energy Magazine - November/December 2017 - 71
IEEE Power & Energy Magazine - November/December 2017 - 72
IEEE Power & Energy Magazine - November/December 2017 - 73
IEEE Power & Energy Magazine - November/December 2017 - 74
IEEE Power & Energy Magazine - November/December 2017 - 75
IEEE Power & Energy Magazine - November/December 2017 - 76
IEEE Power & Energy Magazine - November/December 2017 - 77
IEEE Power & Energy Magazine - November/December 2017 - 78
IEEE Power & Energy Magazine - November/December 2017 - 79
IEEE Power & Energy Magazine - November/December 2017 - 80
IEEE Power & Energy Magazine - November/December 2017 - 81
IEEE Power & Energy Magazine - November/December 2017 - 82
IEEE Power & Energy Magazine - November/December 2017 - 83
IEEE Power & Energy Magazine - November/December 2017 - 84
IEEE Power & Energy Magazine - November/December 2017 - 85
IEEE Power & Energy Magazine - November/December 2017 - 86
IEEE Power & Energy Magazine - November/December 2017 - 87
IEEE Power & Energy Magazine - November/December 2017 - 88
IEEE Power & Energy Magazine - November/December 2017 - 89
IEEE Power & Energy Magazine - November/December 2017 - 90
IEEE Power & Energy Magazine - November/December 2017 - 91
IEEE Power & Energy Magazine - November/December 2017 - 92
IEEE Power & Energy Magazine - November/December 2017 - 93
IEEE Power & Energy Magazine - November/December 2017 - 94
IEEE Power & Energy Magazine - November/December 2017 - 95
IEEE Power & Energy Magazine - November/December 2017 - 96
IEEE Power & Energy Magazine - November/December 2017 - 97
IEEE Power & Energy Magazine - November/December 2017 - 98
IEEE Power & Energy Magazine - November/December 2017 - 99
IEEE Power & Energy Magazine - November/December 2017 - 100
IEEE Power & Energy Magazine - November/December 2017 - 101
IEEE Power & Energy Magazine - November/December 2017 - 102
IEEE Power & Energy Magazine - November/December 2017 - 103
IEEE Power & Energy Magazine - November/December 2017 - 104
IEEE Power & Energy Magazine - November/December 2017 - 105
IEEE Power & Energy Magazine - November/December 2017 - 106
IEEE Power & Energy Magazine - November/December 2017 - 107
IEEE Power & Energy Magazine - November/December 2017 - 108
IEEE Power & Energy Magazine - November/December 2017 - 109
IEEE Power & Energy Magazine - November/December 2017 - 110
IEEE Power & Energy Magazine - November/December 2017 - 111
IEEE Power & Energy Magazine - November/December 2017 - 112
IEEE Power & Energy Magazine - November/December 2017 - 113
IEEE Power & Energy Magazine - November/December 2017 - 114
IEEE Power & Energy Magazine - November/December 2017 - 115
IEEE Power & Energy Magazine - November/December 2017 - 116
IEEE Power & Energy Magazine - November/December 2017 - Cover3
IEEE Power & Energy Magazine - November/December 2017 - Cover4
http://www.nxtbook.com/nxtbooks/pes/powerenergy_050619
http://www.nxtbook.com/nxtbooks/pes/powerenergy_030419
http://www.nxtbook.com/nxtbooks/pes/powerenergy_010219
http://www.nxtbook.com/nxtbooks/pes/powerenergy_111218
http://www.nxtbook.com/nxtbooks/pes/powerenergy_091018
http://www.nxtbook.com/nxtbooks/pes/powerenergy_070818
http://www.nxtbook.com/nxtbooks/pes/powerenergy_050618
http://www.nxtbook.com/nxtbooks/pes/powerenergy_030418
http://www.nxtbook.com/nxtbooks/pes/powerenergy_010218
http://www.nxtbook.com/nxtbooks/pes/powerenergy_111217
http://www.nxtbook.com/nxtbooks/pes/powerenergy_091017
http://www.nxtbook.com/nxtbooks/pes/powerenergy_070817
http://www.nxtbook.com/nxtbooks/pes/powerenergy_050617
http://www.nxtbook.com/nxtbooks/pes/powerenergy_030417
http://www.nxtbook.com/nxtbooks/pes/powerenergy_010217
http://www.nxtbook.com/nxtbooks/pes/powerenergy_111216
http://www.nxtbook.com/nxtbooks/pes/powerenergy_091016
http://www.nxtbook.com/nxtbooks/pes/powerenergy_070816
http://www.nxtbook.com/nxtbooks/pes/powerenergy_050616
http://www.nxtbook.com/nxtbooks/pes/powerenergy_030416
http://www.nxtbook.com/nxtbooks/pes/powerenergy_010216
http://www.nxtbook.com/nxtbooks/ieee/powerenergy_010216
http://www.nxtbook.com/nxtbooks/pes/powerenergy_111215
http://www.nxtbook.com/nxtbooks/pes/powerenergy_091015
http://www.nxtbook.com/nxtbooks/pes/powerenergy_070815
http://www.nxtbook.com/nxtbooks/pes/powerenergy_050615
http://www.nxtbook.com/nxtbooks/pes/powerenergy_030415
http://www.nxtbook.com/nxtbooks/pes/powerenergy_010215
http://www.nxtbook.com/nxtbooks/pes/powerenergy_111214
http://www.nxtbook.com/nxtbooks/pes/powerenergy_091014
http://www.nxtbook.com/nxtbooks/pes/powerenergy_070814
http://www.nxtbook.com/nxtbooks/pes/powerenergy_050614
http://www.nxtbook.com/nxtbooks/pes/powerenergy_030414
http://www.nxtbook.com/nxtbooks/pes/powerenergy_010214
http://www.nxtbookMEDIA.com