VALIDATION OF COMPUTATIONAL FLUID DYNAMICS METHODOLOGY FOR WIND TURBINE

1

• VALIDATION OF COMPUTATIONAL FLUID DYNAMICS
  METHODOLOGY FOR WIND TURBINE
• José Palma, Fernando Castro, Carlos Santos,
• Álvaro Rodrigues and José Matos

2
Questions in this session

• How can the determination of extreme wind speeds
  be validated ?
• What is the quality of the additional information
  simulated by CFD models ?
• How good is the resource assessment for the site
  in extreme regions ?
• Examples of additional useful information.
• Detailed mapping of
• All 3 components of velocity
• Turbulence intensity
• Shear Factor

3
Contents

• The current situation of CFD (Computational Fluid
  Dynamics)
• Validation and Verification (VV) in CFD
• Methodology
• Example case study
• Conclusions

4
Computational Fluid Dynamics (CFD)

• Increasingly importance of CFD in all areas of
  science and engineering.
• Aeronautics and automotive engineering are on the
  forefront and have triggered the establishment of
  guidelines and standards on the use of CFD in all
  stages of design and product development.
• Wind energy engineering can benefit of
  developments in CFD and computational power to an
  extent that has not happened in the past.

• Reasons for the current situation
• A long and well-established practice based on
  linear flow models (WAsP, MSH3D), which is
  difficult to beat, particularly in wind energy
  resource evaluation.
• Uncertainties and difficulties associated with
  the use of CFD techniques, some of them typical
  of wind energy engineering.
• It is not clear how and when CFD can be used
  within the wind energy engineering.

5
CFD in general

• Commercial interests have raised the expectations
  of the users to unrealistic levels, harming the
  credibility and further use of CFD techniques.
• Expertise needed for proper use of CFD codes.
• The importance of this expertise has been
  hidden by userfriendly interfaces, mousedriven
  menus and high-quality graphic packages. The user
  tends to think that things are easy and the
  results are right, even when they are not.

• Scientific publications and engineering societies
  have been fighting this trend, by requiring
  uncertainty analysis of all computer results and
  writing guidelines and standards on the use of
  CFD techniques.
• AIAA American Institute of Aeronautics and
  Astronautics
• Guide for the Verification and Validation of
  Computational Fluid Dynamics (AIAA G-077-1998)
• ASME American Society of Mechanical Engineers
• Request for an ASME Standard on Verification
  and Validation in Computational Solid Mechanics
  (July 2000)
• ASCI Accelerated Strategic Computing Initiative
  of the US Department of Energys (DOEs)
• DMSS Defence Modelling and Simulation Office of
  the US Department of Defence (DoD)

6
Verification and Validation (VV)
Verification and Validation (VV) emerged has two
major concepts in building our confidence on any
computer code.
Verification is the assessment of the accuracy of
the solution of a computational model by
comparison with known solutions. Verification ?
mathematical issue Validation is the assessment
of the accuracy of a computational simulation by
comparison with real field data. Validation ?
real world (physical issue)

• Validation is crucial, if one has to rely and
  make decisions based on computational results.
• Computer results do not replace experimental
  field data.
• Field data is needed at all stages of computer
  simulations, from setting the boundary conditions
  to validation of the computational results.

7
Steps in a CFD application in wind energy

• Choice and extent of region covered by the
  computer simulation.
• Choice of wind conditions at the edges of
  the integration domain.
• Choice of computer code
• Mathematical model to the fluid flow equations
• Physical modelling (turbulence, stratification)
• System of partial differential equations mass
  and momentum conservation plus turbulence model
  equations
• Steady or time-dependent formulation
• Numerical techniques (discretisation techniques,
  set the maximum accuracy achievable)
• Running
• Validating
• Reporting
• Steps are not sequential trial and error
  procedure.
• Initial choices (1 and 2) may proof wrong,
  assessment of the boundary condition implies some
  degree of validation.

8
Sources of uncertainty in CFD applications in
wind energy

• Not strictly related to the wind energy
  engineering
• Discretization error is known at a point, not
  over the whole domain
• Error propagation within the calculation
• Numerical stability and convergence of the
  equation set
• Turbulence modelling, etc.
• Strictly related to the wind energy engineering
• Terrain digital representation
• Anything better than the digitised maps (10 to 50
  m resolution) is questionable.
• Terrain meshing techniques must be clearly
  stated, anything above linear interpolation is
  bound to introduce details that are not real.
• Wind (boundary) conditions at the domain
  boundaries
• Assumptions are needed on
• ground characterisation (roughness)
• ground effects on turbulence phenomena
• inlet conditions, velocity and turbulence
  profiles as a function of distance a.g.l
• One single wind direction and velocity per
  computer simulation
• Question How many wind directions and speeds are
  needed to characterise the site ?

9
Wind conditions - setting and validation

• VALIDATION AND UNCERTAINTY
• There are no methods for quantifying uncertainty
  in CFD calculations.
• However, we can always
• quantify the agreement between computational
  results and field data.
• assess the influence of parameter choice and
  computational conditions (sensitivity tests)
• WIND CONDITION (velocity direction and magnitude,
  and turbulence)
• Settings
• Wind conditions at the boundaries are such that
  the wind conditions by the computer code at a
  selected location in the field (mast A) are the
  same as measured - boundary condition tuning.
• Validation and uncertainty appraisal
• Question what are the wind conditions at mast
  B for given wind conditions at mast A ?
• END RESULT / CONCLUSION
• Agreement at one point (mast B) under well-known
  conditions allows us, at least, to expect the
  same level of agreement at other points under
  identical conditions.
• Any wind condition at that site, using the same
  code under identical conditions.

10
Validation mean horizontal velocity

• Mast 1 as a reference mast
• Validation at all remaining mass for all 12
  directions
• Differences (uncertainty ?) in average below 10
• Critical or 1 critical direction, 180 degree
  winds
• Why is it so? Is this important, i.e. how
  frequent ? wind rose

11
Flow pattern (180 deg winds)
12
Validation - Flow angle, turbulence and shear
factor

• Access to many otherwise unknown quantities
• Increased knowledge of the wind flow
• Increased confidence in wind turbine layout
• Hopefully, increased park efficiency, i.e. lower
  failures

13
Results
14
Results
Even when we are mainly concerned with
point-to-point correlations, 2D plots covering
the whole area of interest must be shown at
different heights above the ground level The
reader may want to perform further analysis,
which can increase his/her own confidence on the
computational results.
15
Methodology

• Select the domain size and spatial
  discretization.
• Select the boundary conditions (wind direction
  and speed) based on real measurements.
• Perform preliminary calculations and adjust the
  boundary conditions in such a way that the
  measurements at one mast can be replicated.
• Assess the results sensitivity (uncertainty
  appraisal) to numerical, computational and model
  parameters, boundary conditions, etc.
• Provide evidence of these tests.
• Analyse the results, including detailed
  reporting on all parameters and conditions under
  which the calculations were performed.

16
Conclusions

• Complex flow models, namely CFD, can contribute
  to the wind energy engineering practice, if
  carefully used and preceded by proper validation.
• CFD models are particularly useful by uncovering
  flow details and intricacies not available via
  more conventional techniques, including
  experimental techniques.
• Use and confidence in the use of CFD, uncertainty
  determination, can be made only on a case-by-case
  basis.
• Uncertainty to be found by comparison with field
  data.
• Simpler linear models must not be discarded, even
  in complex terrain applications, since they
  provide a basis for results comparison.