Time Dependent CFD Analyses of Wind Quality in Complex Terrain

1
Time Dependent CFD Analyses of Wind Quality in
Complex Terrain
Claude Abiven1, José M. L. M. Palma2 and Oisin
Brady1 (1) Natural Power, Scotland (2)
FEUP/CEsA, Faculty of Engineering University of
Porto, Portugal
2

• Outline

• Site overview
• Wind characteristics ? critical sectors
• Steady state versus time dependent analyses
• 300º winds
• 0º winds
• 180º winds
• Conclusions

3

• Site overview - topography

Highly complex topography No trees Three
planned turbines
4

• Site overview wind rose

One year dataset of 10-min average
measurements 140-day dataset of 1Hz wind
measurements 50 m mast
5

• CFD code

• VENTOS CFD code
• written by researchers from the university of
  Porto, Portugal
• 3D Reynolds-averaged Navier-Stokes CFD solver
• ?-e turbulence model
• transport equation discretised by finite volume
  techniques
• References
• 1 Simulation of the Askervein flow. Part 1
  Reynolds averaged Navier-Stokes equations (k-e
  turbulence model).
• Boundary Layer Meteorology. V.107, 501-530,
  2003.
• 2 Linear and nonlinear models in wind resource
  assessment and wind turbine micro-siting in
  complex terrain.
• Journal of Wind Engineering and Industrial
  Aerodynamics. V. 96, 2308-2326, 2008.

6

• Wind characteristics critical sectors

Measured Computed
Values of turbulence intensity are large for
sectors 0º, 180º, 300º
Values of veer are large for sectors 0º, 150º,
300º ? In-depth analysis of sectors 0º, 180º,
300º is carried out

7

• 300º winds steady state wind direction

Y
This can be seen on the wind rose
Large values of veer are caused by the topography
8

• 300º winds steady state wind turbulence

High turbulence coincides with flow divergence at
the channel outlet
9

• 300º winds power spectrum

measured
Measured and simulated peak positions are in good
agreement T 1000s ? Simulations can help us
understand the reason for these peaks
simulated
10

• EOF analysis

EOF Empirical Orthogonal Functions Widely used
in climate sciences From a spatial variable
evolving with time (i.e. map of wind speed as
a function of time) EOFs split the signal into
spatial patterns of this variable associated with
a time series. Each pattern explains part of the
variance of the original signal. (i.e. maps of
wind speed high and lows and their evolution in
time)
11

• 300º winds EOF of wind speed

The first EOF is associated with an oscillation
of period 1000s followed by a steady
state Most of the variability occurs at and
downwind of the turbines High-lows are the sign
of an oscillation
12

• 300º winds frame by frame analysis

13

• 0º winds steady state wind direction

A system of vortices forms downwind of the hill
14

• 0º winds power spectrum

measured
Measured and simulated peak positions are in good
agreement T 1000s ? Simulations can help us
understand the reason for these peaks
simulated
15

• 0º winds frame by frame analysis

16

• 190º winds steady state wind direction

This can be seen on the wind rose
The wind is forced around the hill and appears as
a wind from direction 210 on site
17

• 150º winds steady state wind direction

The wind is forced around the hill and appears as
a wind from direction 90 on site
This can be seen on the wind rose
18

• 170º winds power spectrum

measured
Measured and simulated peak positions agree
reasonably well 100s lt T lt 300s ? Simulations
can help us understand the reason for these peaks
simulated
19

• 170º winds frame by frame analysis

20

• Conclusions

• Low wind occurrences are caused by nearby
  mountains, which divert the flow from its
  original direction.
• Spectral analyses show the preferred time scales.
• EOF and frame by frame analysis are used to
  relate the preferred time scale to a physical
  event.
• The model is able to reproduce time-dependent
  phenomena in complex terrain, as measured by the
  met mast.
• 10-min averaged data and conventional analysis
  hide important flow features that can impair the
  wind farm operation.