Time variant energy yield analysis

renewable energy projects
by renewable energy people
Our Services
Full lifecycle services across renewable energy sectors
2
03/06/2015
Time variant energy yield analysis
A case study
Presenter: Daniel Marmander
Date: 3rd June 2015
Introduction
What we set out to do
To assess the effect on P50 yield of the standard method using time
invariant compared to a time variant energy yield calculation
To determine which parameters would have the greatest effect if
changed from time invariant to time variant
To see if binning by time of day and month of year would produce
the same result as a time step by time step approach
4
03/06/2015
Introduction
What we have done
We have assessed:
Impact of TI-specific power curves
Impact of rotor equivalent wind speed
Impact of time variant wind flow
Impact of different binning strategies
This is a case study, for a single site.
5
03/06/2015
Introduction
What we have not done
This does not reflect:
The absolute accuracy of the various
scenarios
Quantative statistics – it is a case study
We have not used turbulence renormalisation
to compensate for high/low turbulence, only
TI-specific power curves
6
03/06/2015
Introduction
What, when, where
The site is thermally driven, with low inter
annual variation but large diurnal variations
We have in practise only one wind direction,
from the east
We have diurnal variations in wind speed
ratios between measurements on site
We have based our assessment on actual
measured wind data and calculated power
curves, however the results have been
anonymised
7
03/06/2015
Introduction
Measured wind speed by hour of day and month of year
8
03/06/2015
Introduction
Measured wind speed by hour of day and month of year
2 m/s
16 m/s
9
03/06/2015
Introduction
Base case
The base case represents an (almost)
standard setup:
EV wake model
Sector wise TI for wakes
One flow resource grid, 100 m resolution
Hub height (80 m) wind speed used
Normal TI power curve
Correction for average air density
Run on time series, in contrast to
frequency distribution
10
03/06/2015
Introduction
Scenarios
Base case
Rotor equivalent wind speed, using average shear, 12x24 and instant
TI specific power curves, using average TI, 12x24 and instant
Air density correction, 12x24 and instant
Parameters are changed once at a time (e.g. base case, but with 12x24
air density correction), as well as all at once (e.g. everything on 12x24).
11
03/06/2015
Shear
Time of day, measured shear
12
03/06/2015
Shear
Time of year, measured shear
13
03/06/2015
Shear
Impact of time variant compensation for shear conditions
Rotor equivalent wind speed compensation
8.60
Very small difference,
all below 0.5%
8.55
Relates to available
wind energy
8.50
8.45
Does not predict effect
on turbine capacity to
convert the energy
8.40
8.35
8.30
Base case
14
Average: 0.4%
24x12: -0.4% Instant: -0.2%
03/06/2015
Turbulence intensity
By speed, measured TI
15
03/06/2015
Turbulence intensity
By time of day, measured TI
16
03/06/2015
Turbulence intensity
By time of year, measured TI
17
03/06/2015
Turbulence intensity
Impact of using TI specific power curves
Turbulence intensity specific power
curves
8.60
Very small differences
Turbulence
renormalisation
method has not been
used
8.55
8.50
8.45
Does not reflect the
influence on wake
losses
8.40
8.35
8.30
Average: 0.0%
18
24x12: 0.2%
Instant: 0.2%
03/06/2015
Wind speed
By time of day, measured wind speed
19
03/06/2015
Wind speed
Impact of using time variant flow conditions derived from meso-scale model
Wind flow
8.60
8.55
Average wind speed at
hub height is 0.2 m/s
higher using time
invariant wind flow
8.50
Almost no difference in
yield
8.45
8.40
Wind speeds are either
very low or very high
8.35
8.30
Base case
20
Time of day: -0.3%
Unusually low
sensitivity for wind
speed changes
03/06/2015
Temperature
By time of day, measured temperature
21
03/06/2015
Air density
Impact of using time variant air density
Air density compensation
8.60
8.55
Significant yield
difference
No difference between
12x24 and instant
8.50
Caused by strong
diurnal trend for both
speed and temperature
8.45
8.40
8.35
8.30
Base case
22
24x12: 2.5%
Instant: 2.5%
03/06/2015
Combined results
Impact of adjusting for all parameters
Very similar to when
adjusting for varying air
density only
All combined
8.60
8.55
Significant difference
against base case
8.50
8.45
Very small difference
between 12x24 and
instant
8.40
8.35
8.30
Base case
23
24x12: 2.0%
Instant: 2.2%
03/06/2015
Conclusions
For this case study
Very small difference between 12x24 binning and instant time
series adjustments. Indicates that 12x24 binning is adequate
Air density adjustment is by far the most important factor for
energy yield
Time variant wind flow makes a large difference in mean wind
speed over the site, but has small impact on final energy yield
24
03/06/2015
Conclusions
In general
Time variant modelling might be able to more precisely predict the
energy yield
It is likely that a 12x24 approach is adequate for most sites
There is a need to assess the absolute accuracy of the different
approaches, to assure that no over all bias is accidentally introduced
By assessing the sensitivity for using 12x24 binning instead of
averages, we can get a reasonable idea about the need of a timevariant approach
25
03/06/2015
Thank you for listening
Questions?
26
03/06/2015