IEEE Power & Energy Magazine - November/December 2017 - 45

Nearly all forecast systems have one thing in common:
they rely on numerical weather predictions to calculate
the expected RES power production.

200
160
120
80
40

Time (UTC)

0
. 2 0:0
01 0
3
ov

9

N

1
. 2 6:0
01 0
3
ov

8

N

0
. 2 8:0
01 0
3
ov

8

N

0
. 2 0:0
01 0
3
ov

8

N

1
. 2 6:0
01 0
3
ov

7

N

0
. 2 8:0
01 0
3
ov

7

N

ov
N
7

ov

0
. 2 0:0
01 0
3

0

N
7

If deterministic forecasts are used [Figure 5(a)], the result
is a scenario that provides information on areas where grid
congestion is to be expected, e.g., overloads of lines or transformer stations. Depending on the computing resources, this
can be repeated to cover all time scales, from two days ahead
down to minutes ahead.
The next step is then estimating the forecast uncertainty
in advance to become informed about the range of possible
grid congestion areas. For this purpose, ensemble forecasts are
required as input [Figure 5(b)] because they provide a range
of different realistic scenarios consistent with the approaching
weather situations. Although this is computationally expensive,
it is deemed necessary to include the correct spatiotemporal

Power Rated (MW)

70
60
50
40
30
20
10
0

0
. 2 0:0
01 0
3
7
N
ov 08
. 2 :0
01 0
3
7
N
ov 16
. 2 :0
01 0
3
8
N
ov 00
. 2 :0
01 0
3
8
N
ov 08
. 2 :0
01 0
3
8
N
ov 16
. 2 :0
01 0
3
9
N
ov 00
. 2 :0
01 0
3

Power Rated (MW)

others, a more advanced approach is applied to consider the
grid topology, i.e., taking into account the structure of the
grid. Using this topology information can inform which
RES units have an impact on critical parts of the grid, e.g.,
by dynamic load-flow calculations. Although this topology information about the DSO as well as the TSO grid is
often not available to forecasters, many grid operators have
started to enhance their power-flow calculation tools to be
able to accommodate forecast information. RES forecasts
for individual sites or groups of plants with a common grid
connection point can already be used as input to power-flow
calculations on the DSO level and, in a second step, on the
TSO level.

Time (UTC)

N
7

N
7

210
180
150
120
90
60
30
0

ov 0
. 2 0:0
01 0
7
N
ov 0 3
. 2 6:0
01 0
7
N
ov 1 3
. 2 2:0
01 0
7
N
ov 1 3
. 2 8:0
01 0
8
N
ov 0 3
. 2 0:0
01 0
8
N
ov 0 3
. 2 6:0
01 0
8
N
ov 1 3
. 2 2:0
01 0
8
N
ov 1 3
. 2 8:0
01 0
9
N
ov 0 3
. 2 0:0
01 0
3

Power Rated (MW)

70
60
50
40
30
20
10
0

ov 0
. 2 0:0
01 0
7
N
ov 0 3
. 2 6:0
01 0
7
N
ov 1 3
. 2 2:0
01 0
7
N
ov 1 3
. 2 8:0
01 0
8
N
ov 0 3
. 2 0:0
01 0
8
N
ov 0 3
. 2 6:0
01 0
8
N
ov 1 3
. 2 2:0
01 0
8
N
ov 1 3
. 2 8:0
01 0
9
N
ov 0 3
. 2 0:0
01 0
3

Power Rated (MW)

(a)

Time (UTC)

Time (UTC)
(b)

figure 5. Wind power forecasts for two different transformer stations in the same German DSO grid about 250 km apart.
(a) The red line shows a deterministic forecast indicating the most probable scenario at both sides. (b) The colored lines
show ensemble forecasts representing different scenarios: an equal color belongs to an identical scenario at both sites.
The black lines represent the measurement. UTC: coordinated universal time.
november/december 2017

ieee power & energy magazine

45
Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - November/December 2017
