Validation of wind speed prediction methods at offshore sites

Document Sample
Validation of wind speed prediction methods at offshore sites Powered By Docstoc
					Resource analysis for waters offshore
             of the UK

             Dougal McQueen
        Garrad Hassan and Partners Limited
             Dr Simon Watson
        CREST, Loughborough University
                    Motivation
• The cost of offshore
  monitoring is large
  (difficult access, harsh
  environment…)
• Concerns over “industry
  standard” methods
  (WAsP)
• Lack of quality data
  coverage (emerging
  industry)
• Many methods and
  technologies; but do they
  work?
                                                                    z             10m 
                                        usea z   usea 10m  ln 
                                                                   z
                                                                            
                                                                               ln 
                                                                                   z
                                                                                            
                                                                                            
                                                                    0, sea        0, sea 

                                                                                       u*2
                 Methodology                                             z 0, sea   
                                                                                       g


1) Identify possible methods for
   resource prediction

2) Make predictions

3) Validate against measurements

No attempt made to increase knowledge
  of “Boundary Layer Meteorology”
Charnock, Log-law
                        Methods
Met station wind speeds:
   – Empirical as described by
     Hsu.

   – Wind Atlas as described in
     European Wind Atlas (used
     in WAsP), no obstacles,
     topography, only boundary
     layer development model.
   – Proximity, boundary layer
     similarity?
                                                                 2
                                        u  u                
                                     G  *  ln  *   Az L   B 2 z L 
                                           fz0 
                                                  
                                                               
                                                               

                  Methods cont.
Gridded pressure product
   – Nogaps
   – Geostrophic drag law


Triangulated pressure
   gradients
   – Pressure measurements at
     Met stations.
   – Geostrophic drag law
   – Spatial separation??                  1 p
                                fG x  
                                            y
                                                              1 p
                                                   fG y  
                                                               x
                       Methods cont.
Reanalysis data
   – Assimilated: surface pressure,
     radiosonde, satellite, ship,
     aircraft, pibal etc…
   – Global Frozen state Numerical
     Prediction model
   – Reanalysis 2 (NCEP / NCAR)
       • 925 mb winds
       • 10m surface winds (boundary
         layer model)
   – ECMWF, World Wind Atlas,
     Reanalysis 1, NARR not
     used…!
                    Methods cont.

Operational Forecast model
   – Evolving state, data
     assimilation?
   – Nogaps (US Navy run
     global model, 10m winds)
   – ECMWF Operation model
   – UKMO MESOSCALE
     model (high resolution)
     10m winds & 925mb winds
            Validation cases

~30 m measurement height
• Scroby Sands - 5 year
• Kish Bank lighthouse - 18
  month (obstacle corrected)
• Arklow Bank - 2 year
• Shell Flats (2 masts) - 1 year
                            Offshore Mast
                            Met station
                            Met station
                            (MSLP)

Kish Bank             Shell Flats



                          Scroby Sands




        Arklow Bank
                                                         x         xo 
                                                    1
                                                                p
                                                    N
                                               m       N
                                                                             100
                           Mean bias                        
                                                                             1
                                                                      
                                                                             N
                                                                                 x
                                                                                 N
                                                                                     o



                                                    Scroby Sands
         Hsu (empirical)                            Kish Bank
                                                    Shell Flats 1
             Wind Atlas                             Shell Flats 2
Triangulated pres. grads                            Arklow Bank

NOGAPS gridded MSLP
   NOGAPS 10m winds
    ECMWF 10m winds
      NCEP 10m winds
    NCEP 925mb winds
            UKMO 10m
          UKMO 950mb

                       -40 -30 -20 -10 0   10 20 30 40 50 60                 %
                                                        x    p  xo 
                                                   1                  2

                                                   N
                                              r       N
                                                                          100
                                                           
                            RMS error
                                               Scroby Sands
         Hsu (empirical)                       Kish Bank
                                               Shell Flats 1
             Wind Atlas                        Shell Flats 2
                                               Arklow Bank
 Triangulated pres. grads
NOGAPS gridded MSLP
  NOGAPS 10m winds
   ECMWF 10m winds
      NCEP 10m winds
   NCEP 925mb winds
            UKMO 10m
         UKMO 950mb

                       0    10 20 30 40 50 60 70 80 90 100                 %
                   
                    1
                                                                             x           xo ,i 
                                                                            N
                                                                        1
                                             xb  a1  xo  a0
             T
  a  Xo Xo              X o  xp
                           T                                                                      2
                                                                                   b ,i
                                                                        N   i i
                                                                  b                                  100
                                                                                   
     1 xo ,1 
     1 x 
Xo  
         o,2 
                               Unbiased RMS error                                                 x p ,1 
                                                                                                 x 
                                                                                               Xp
                                                                                                     p,2 
                                                                                               
                                                                                                       
     1 xo , N                                                                                   x p,N 
                                                                                                         
                                                                 Scroby Sands
                   Hsu (empirical)                               Kish Bank
                                                                 Shell Flats 1
                          Wind Atlas                             Shell Flats 2
                                                                 Arklow Bank
 Triangulated pres. grads
   NOGAPS gridded MSLP
       NOGAPS 10m winds
            ECMWF 10m winds
              NCEP 10m winds
         NCEP 925mb winds
                         UKMO 10m
                    UKMO 950mb

                                       0   10 20 30 40 50 60 70 80 90 100 %
                           Conclusions
Range of methods for Wind Speed prediction presented.
Masts offshore of UK used as a case study.
Boundary layer modelling specifics avoided.

• Some methods show systematic bias (could be removed).
• Promise shown for reanalysis and forecast models where Wind Atlas method
  is limited.
• Methods OK for site screening / feasibility.
• More applicable when sites move further offshore.
• No substitute for in site measurements.

Future direction
    – Consistency
    – Use of Reanalysis for MCP?
    – Boundary layer modelling… sensitivities…
                Thank you

• Particular thanks to CREST, Loughborough
  University (Dr Simon Watson).
• Dr Dave Pearce, E.ON Renewables UK
• Dr Chris Ziesler, Shell Renewables
• Dr Brian Hurley, Airtricity

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:10
posted:8/7/2012
language:
pages:14