16 lorenz local_and_regional_pv_power

October 22nd 2015, 4th PV Modelling Workshop, Köln, Germany 11
Local and regional PV power forecasting
based on
PV measurements, satellite data
and numerical weather predictions
Elke Lorenz, Jan Kühnert, Björn Wolff,
Annette Hammer, Detlev Heinemann
1)Energy Meteorology Group, Institute of Physics,
University of Oldenburg
1

Outline
grid integration of PV power in Germany
overview on PV power prediction system
evaluation:
different data and models for different forecast horizons
combination of different models
summary and outlook
2

October 22nd 2015, 4th PV Modelling Workshop, Köln, Germany
Grid integration of PV in Germany
installed Power: 38.5GWpeak (end of 2014)
3

marketing of PV power at the European Energy Exchange
3
control areas

by transmission system operators
regional forecasts
3
control areas

regional forecasts
direct marketing
local forecasts
3
control areas

regional forecasts
direct marketing
local forecasts
day ahead: for the next day
3
control areas

regional forecasts
direct marketing
local forecasts
intraday: until 45 minutes
before delivery
3
control areas

regional forecasts
direct marketing
local forecasts
intraday: until 45 minutes
before delivery
here: 15 min to 5 hours ahead
3
control areas

PV power
measurement
hoursforecast horizon
Overview of forecasting scheme
4

PV power
measurement
satellite cloud motion
forecast CMV
hoursforecast horizon
4

PV power
measurement
forecast CMV
dayshours
NWP: numerical
weather prediction
forecast horizon
4

PV power predictions
PV power
measurement
forecast CMV
dayshours
PV power forecasting:
PV simulation and
statistical models
NWP: numerical
weather prediction
forecast horizon
4

Numerical weather predictions
global model forecast (IFS)
of the European Centre
for Medium-Range Weather Forecasts
(ECWMF)
regional model forecasts (COSMO EU)
of the German Meteorological service
(DWD)
COSMO EU, dir. irradiance
2104-05-02, 12:00
5

Numerical weather predictions
global model forecast (IFS)
of the European Centre
for Medium-Range Weather Forecasts
(ECWMF)
regional model forecasts (COSMO EU)
of the German Meteorological service
(DWD)
Post processing:
 bias correction and combination
with linear regression
COSMO EU, dir. irradiance
2104-05-02, 12:00
5

Meteosat Second Generation (high resolution visible range)
cloud index from
Meteosat images with
Heliosat method*
resolution in Germany
1.2km x 2.2 km
15 minutes
Irradiance prediction
based on satellite data
6
*Hammer et al 2003

cloud index from
Heliosat method
cloud motion vectors by
identification of matching
cloud structures in
consecutive images
extrapolation of
cloud motion to predict
future cloud index
Meteosat Second Generation (high resolution visible range)
6

satellite derived irradiance maps
cloud index from
Heliosat method
cloud motion vectors by
identification of matching
cloud structures in
consecutive images
extrapolation of
cloud motion to predict
future cloud index
irradiance from predicted
cloud index images with
Heliosat method
200W/m2 900W/m2
6

Measurement data
March- November 2013
15 minute values
921 PV systems1) in Germany
information on PV system
tilt and orientation
19
1)Monitoring data base of Meteocontrol GmbH
7

PV power
measurement
forecast CMV
dayshours
PV power forecasting:
PV simulation and
statistical models
NWP: numerical
weather prediction
forecast horizon
8

PV power
measurement
forecast CMV
dayshours
NWP: numerical
weather prediction
forecast horizon
8
Pmeas(t-Dt)
Pclear(t-Dt)
persistence:
constant ratio of
measured PV power Pmeas
to clear sky PV power Pclear
Ppers (t)= Pclear(t)

Different input data and models
22
PV power
measurement
forecast CMV
dayshours
NWP: numerical
weather prediction
forecast horizon
persistence
8
PV simulation:
 Diffuse fraction model:
Skartveith et al, 1998
 Tilt model: Klucher 1970
 Parametric model for MPP
efficiency: Beyer et al 2004
 Linear regression

PV power
measurement
forecast CMV
dayshours
PV simulation
NWP: numerical
weather prediction
forecast horizon
PV simulationpersistence
8

Regional forecasts:
persistence, CMV and NWP based forecasts
9

Regional forecasts:
persistence, CMV and NWP based forecasts

Rmse in dependence of forecast horizon
15 minute values
normalization to installed power Pinst
only daylight values, calculation time of CMV: sunel > 10°
only hours with all models available included in dependence of forecast horizon
 






N
i
inst
pred
inst
meas
P
P
P
P
N
rmse 1
2
1
10

forecasts for German average
CMV forecasts better than
NWP based forecast
up to 4 hours ahead
10
 






N
i
inst
pred
inst
meas
P
P
P
P
N
rmse 1
2
1
15 minute values

forecasts for German average
CMV forecasts better than
NWP based forecast
up to 4 hours ahead
persistence better than
CMV forecasts
up to 1.5 hour ahead
10
 






N
i
inst
pred
inst
meas
P
P
P
P
N
rmse 1
2
1
15 minute values

11
comparison of German average and single site forecasts:
rmse for German about 1/3 of single sites rmse for NWP forecasts
German average single sites

11
comparison of German average and single site forecasts:
rmse for German about 1/3 of single sites rmse for NWP forecasts
improvements with persistence and CMV larger for regional forecasts

PV power
measurement
forecast CMV
dayshours
PV simulation*
NWP: numerical
weather prediction
forecast horizon
PV simulation*persistence
12
*)PV simulation
with bias correction

PV power
measurement
forecast CMV
dayshours
PV simulation
NWP: numerical
weather prediction
forecast horizon
Combination of different models
PV simulationpersistence
combination
12

Combination of forecasting methods
combination of forecast models with linear regression:
Pcombi=aNWPPNWP + aCMVPCMV + apersistPpersist + a0
coefficients aNWP, aCMV, apersist, a0 are fitted to measured data

in dependence of
 forecast horizon
 hour of the day

in dependence of
 hour of the day
training data:
 for single site forecasts: each PV system separately

in dependence of
 hour of the day
training data:
 for single site forecasts: each PV system separately
 for regional forecasts: average of sites

How many days to train forecast combination?
Improvement score:
with respect to best single model
𝑟𝑚𝑠𝑒 𝑟𝑒𝑓 − 𝑟𝑚𝑠𝑒 𝑐𝑜𝑚𝑏𝑖
𝑟𝑚𝑠𝑒 𝑟𝑒𝑓
all sites average, May to November, 2012

Improvement score:
independent test year for model configuration

Improvement score:
last 30 days

Regional forecasts
regression coefficients in dependence of forecast horizon
15
bars:
standard deviation for all hours
regression coefficients (weights) reflect horizon dependent
forecast performance of different models

Regional forecasts
Rmse in dependence of forecast horizons
15
Considerable improvement with combined model
over single model forecasts

Single site forecasts
regression coefficients in dependence of forecast horizon
bars:
standard deviation
for all hours & sites
regression coefficients (weights) reflect horizon dependent
forecast performance of different models for single sites

Regression coefficients
16
bars:
standard deviation
for all hours & sites
horizon dependent regression coefficients different for
regional and single site forecasts

forecast combination outperforms single model forecasts
for all horizons
improvements with combination larger for regional forecasts
17

Summary
PV power prediction contributes to successful
grid integration of more than 38 GWpeak PV power in Germany
18

Summary
PV power forecasts based on satellite data (CMV) significantly
better than NWP based forecasts up to 4 hours ahead
18

Summary
PV power forecasts based on satellite data (CMV) significantly
better than NWP based forecasts up to 4 hours ahead
significant improvement by combining different
forecast models with PV power measurements,
in particular for regional forecasts
18

Outlook
combination of machine learning with PV simulation
for integration of additional data sources:
additional meteorological parameters
additional NWP systems
uncertainty information
19

Thank you for your attention!
EU-FP7 grant agreement no: 308991
20

16 lorenz local_and_regional_pv_power

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to 16 lorenz local_and_regional_pv_power

Similar to 16 lorenz local_and_regional_pv_power (20)

More from Sandia National Laboratories: Energy & Climate: Renewables

More from Sandia National Laboratories: Energy & Climate: Renewables (20)

Recently uploaded

Recently uploaded (13)

16 lorenz local_and_regional_pv_power