Hatch Ltd. Validation and comparison WAsP and meteodyn 2011
1. Validation and Comparison between
WAsP and Meteodyn Predictions for
a Project in Complex Terrain
Meteodyn WT users meeting– Paris (France), March 21 and
22, 2011
Gilles Boesch, Wind Project Analyst
j y
Salim Chemanedji, Senior Project Manager
Martin Hamel, Project Manager
Hatch (Montreal), Canada
2. Hatch
• Employee-owned, Projects in more than 150 countries
– 8000 employees worldwide
l ld id
– EPCM, integrated teams, project and construction
management
– Consulting – process technologies and business
process,
– Serving mining & metals, infrastructure and energy
• For Wind Power projects:
– Wind resource assessment
– Geotech engineering, foundation design
– Turbine evaluation and selection
– Total project and construction management
– Interconnection assessment, Electrical engineering
– Environmental assessment
2
3. Overview
• Review of models
• Presentation of a test case
• Results and comparisons
• Conclusions and investigations
3
5. Why is CFD a good
alternative to linear models ?
• CFD is now well recognized by the wind
community
• Overpass linear model limitations for
p
complex terrain
• Reduce the modelling uncertainty
• R d
Reduce financial risks
fi i l i k
But CFD must be used with care since it
is more complex
5
6. Why is CFD a good
alternative to linear models ?
• Some questions remains:
– Can we quantify the uncertainty and errors
associated to these models ?
– What are the criteria for chosing linear or CFD
models ?
– Do CFD models always perform better than
linear models ?
Usually difficult to assess because only few
meteorological masts are available within a project
to perform cross-predictions
6
7. Why is CFD a good
alternative t linear models ?
lt ti to li d l
CFD Models (Meteodyn) Linear Models (WAsP)
• Pros • Pros
– Suitable for complex terrain – Easy and fast computation
– Calibration f the it
C lib ti of th site – Good performance in
possible (forest, stability, relatively flat terrain
mesh etc.) – Is already a standard
– Built-in features (energy, • Cons
extreme winds turbulence
winds,
etc.) – High errors for complex
terrain
• Cons – Calibration is difficult to
– Solid expertise needed perform (when possible)
– Calculation time
C l l ti ti
7
9. A test case
• Comparison between WAsP and
Meteodyn on a potential project
• Project covers an area of 11km x 8km
j
• Equipped with 12 meteorological masts
(recording from 6 months to 6 years of
data)
• Relatively complex (deep valleys, ridges,
rolling mountains)
g )
• Mix of coastal and inland areas
9
10. A test case
• Forest diversity, varies among :
diversity
– Completely logged area (no trees)
– 15m high trees
– Regrowth
• RIX variations (Ruggedness Index)
– % of slopes >30% in a 3500m radius
– 2 to 25 over the entire project
– 2.7 to 22.4 at the meteorological masts
Variety of conditions to evaluate the
behavior of the models
10
14. Meteodyn settings
• Topographical information :
– Roughness : 0.6 for trees
– Elevation Contour : 5m within project area
• Mesh :
– Mapping area covering all met masts
– Mesh independency tests (variation of the
Radius)
– Minimum horizontal resolution : 30m
– Minim m vertical resol tion : 5m
Minimum ertical resolution
– 3 460 000 cells in the prevailing direction
14
15. Meteodyn settings
• Model:
– Robust forest model (convergence issues with
the dissipative model)
– Near neutral stability class
– 30 degrees directional steps
• Data:
– Measured data
– Quality controlled
– At 50m or 60m high
– Extrapolated to long term with standard MCP
method
15
18. Results - Errors
• Cross-Prediction Matrix
– 12 x 12 matrix = 132 cross predictions
– For both WAsP and Meteodyn
– No correction is applied to both models output
– Correction often applied with WAsP because
of wind speed inconsistencies in complex
terrain
• Converted into a Relative Error Matrix :
V predicted − Vmeasured
%E =
Vmeasured
• Resulting in 132 relative error values for
each cross-prediction
h di ti
18
19. Results - Errors
• Absolute errors
WAsP Meteodyn
Min Error 0.0% 0.0%
Max Error 34.0% 14.1%
Average 7.1% 4.7%
• On average, Meteodyn reduces the error
by 35%.
• S
Some exceptions : 33 cases out of 132
ti t f
show better results with WAsP
19
20. Results - Errors
• Generally, errors have the same sign
(positive/negative)
40.0%
30.0%
20.0%
Relativ Error (%)
WAsP
10.0%
Meteodyn
y
ve
0.0%
-10.0%
-20.0%
• The difference is in the magnitude
g
20
22. Results - Errors
• RIX dependency:
– WAsP : Error increase sharply when RIX >
15%
– Meteodyn : Error is more constant
RIX influence on cross-prediction errors
25.00%
20.00%
15.00%
Averag Error (%)
Wasp
Meteodyn
10.00%
ge
5.00%
0.00%
0.0 5.0 10.0 15.0 20.0 25.0
RIX (%)
22
23. Results - Errors
• RIX dependency:
– Ri Ø suggests correcting WAsP with ∆RIX
Ris t ti WA P ith
(between 2 masts)
– Correction based on a correlation between
Error and ∆RIX for each cross-prediction
E d f h di ti
– Open question : Can we correct Meteodyn
based on the RIX ?
Error vs dRIX - Meteodyn Error vs dRIX - Wasp
40.0% 40.0%
30.0% y = 0.5552x 30.0% y = 1.0632x
R² = 0.6345 R² = 0.7025
20.0% 20.0%
%)
Error (%
%)
10.0%
10 0% 10.0%
10 0%
Error (%
0.0% 0.0%
‐30.0% ‐20.0% ‐10.0%
-10.0% 0.0% 10.0% 20.0% 30.0% ‐30.0% ‐20.0% ‐10.0%
-10.0% 0.0% 10.0% 20.0% 30.0%
-20.0% -20.0%
-30.0% -30.0%
∆RIX (%) ∆RIX (%)
23
24. Results - Errors
• RIX dependency:
– E
Error increases when ∆RIX increases
i h i
– Error and ∆RIX seem to be correlating
– The slope is lower for Meteodyn
Meteodyn is less sensitive to site topography
differences
Error vs dRIX - Meteodyn Error vs dRIX - Wasp
40.0% 40.0%
30.0% y = 0.5552x 30.0% y = 1.0632x
R² = 0.6345 R² = 0.7025
20.0% 20.0%
%)
Error (%
%)
10.0%
10 0% 10.0%
10 0%
Error (%
0.0% 0.0%
‐30.0% ‐20.0% ‐10.0%
-10.0% 0.0% 10.0% 20.0% 30.0% ‐30.0% ‐20.0% ‐10.0%
-10.0% 0.0% 10.0% 20.0% 30.0%
-20.0% -20.0%
-30.0% -30.0%
∆RIX (%) ∆RIX (%)
24
26. Results - Uncertainty
• Caracterize the repeatability of an
estimate
• Uncertainty can be reduced on average
y g
by 1.5 when using Meteodyn.
• No trend with regards to the RIX
• The separation distance is more
important regarding the uncertainty
Numbers are site-specific and must
be considered with care !
26
28. Conclusions
• For this project Meteodyn shows better
project,
results for error and uncertainty
compared to WAsP
• Significant advantages :
– Cost reduction : Need for less meteorological
masts per project
– Financial risks reduction : Lower uncertainty
increases P75 or P99 value
28
29. Conclusions
• However :
– WAsP results are without any correction which
is often performed (like RIX correction for
example)
– Results are specific to this site
– Some cases are better predicted with WAsP
– Other projects with lower RIX show equivalent
results between both models
29
30. Conclusions
• Further investigations and questions :
– How do they compare when correcting WAsP
y g
with the RIX ?
– Can we correct Meteodyn’s results in a certain
way ? (RIX or other)
– Why does WAsP better predict the wind speed
at some points ?
• Mesh refinement ?
• Forest model ?
• Roughness ?
– What are the criteria for defining a terrain in
terms of complexity (use of WAsP vs
Meteodyn) ?
– How many met tower should we use in
complex terrain ?
30