Post on 31-Dec-2016
Background Computational Methodology Results Conclusions + future directions
Investigation of the influence ofturbine-to-turbine interaction
on their performance using OpenFOAM
Dr Gavin Tabor, Mulualem Gebreslassie, Prof Mike Belmont
CEMPS, University of Exeter
Background Computational Methodology Results Conclusions + future directions
Background: Lift/Drag Turbine
Novel design for tidal turbine based on cycloidal turbine involvingcomplex rotating airfoil blades
• Blades act in drag mode on one side; rotate (0.5Ω) to developlift on other side
• Unit operates as cross-flow turbine
• Energy extracted through volume – high efficiency (measuredefficiency of ∼ 50%
• High blockage factor; suitable for near-surface (eg. esturine)sites.
• Development backed by AquaScientific Ltd – someexperimental and esturine testing done
Background Computational Methodology Results Conclusions + future directions
Project aims
Ultimate aim : to model large (100+) farms of units
Proximate aim : low-cost CFD model of multiple units to studyinteractions. However; detailed turbine blade motion too costly;simple actuator disk models insufficiently detailed.
• Developed new Immersed Body Force technique to treat blademotion
• LES turbulence formulation – need to examine large scaletransient motions
• VOF – free surface important for turbine behaviour
Computational code used : OpenFOAM
Background Computational Methodology Results Conclusions + future directions
LES and VOF
Filtered NSE including body force terms :
∇.u = 0,
∂tu +∇.(u ⊗u ) = ∇.(S − B) + F
S = −pI + 2νD
B is SGS Stress term : effect of SGS turbulence on GS flowrepresented by 1-equation eddy viscosity model.
F represents artificial body force term.
Background Computational Methodology Results Conclusions + future directions
VOF
FSF represented by indicator function α
∂α
∂t+∇.(αu ) +∇.(α(1− α)ur ) = 0
Final term is artificial compression term active only on interface.
Physical properties calculated as weighted average of individualcomponents;
µ = αµw + (1− α)µa
Background Computational Methodology Results Conclusions + future directions
Turbine modelling
Immersed body force method :
• Blades represented by bodyforces
F = FD + F L
• Compromise between accuracyand efficiency
• Capable of representing largescale vortexes
Background Computational Methodology Results Conclusions + future directions
What is OpenFOAM?
OpenFOAM is an Open Source CCM code/code library :
• Written in C++
• Based on FVM on arbitrary unstructured (polyhedral cell)meshes
• Originally developed by Henry Weller and others at IC (1990 –2000); Nabla Ltd (2000 – 2004) as FOAM
• Now released (2004 –) under Gnu GPL by OpenCFD Ltd.(http://www.opencfd.co.uk/)
• Extensive user community
• Extensions and variants released by 3rd parties (-dev, pyFoam)
• Academic and commercial usage.
Background Computational Methodology Results Conclusions + future directions
Strictly, OpenFOAM is not a CFD code – it is a C++ library ofclasses for writing CFD codes.
OpenFOAM uses the full range of the C++ language –inheritance, polymorphism, templating, operator overloading etc –where appropriate :
• Class mechanism – define new “types” for CFD
• Interface vs implementation : segregation of effort.
• Operator Overloading – provides standard mathematicalsyntax
• Inheritance, polymorphism etc – encodes relationshipsbetween conceptual entities in code
Effective result is a high level “language” for encoding CFD.
Background Computational Methodology Results Conclusions + future directions
Validation
Laboratory testing carried outin flow channel; flow rate andturbine rotation under a rangeof mechanical torqueconditions :
• Flow rate measured withrotormeter
• Torque output usingmechanical system
• Rotation rate recordedoptically.
Compared with functionallyequivalent CFD simulations.
Background Computational Methodology Results Conclusions + future directions
Single Turbine simulation
Background Computational Methodology Results Conclusions + future directions
Two turbine simulations
• Reduced efficiencyby 18% at 15Dturbine spacing
• Reduced efficiencyby at least 7% at20D spacing
Background Computational Methodology Results Conclusions + future directions
Three turbine simulations
• Performance of middleturbine improved due toblockage effect at 2Dlateral spacing
• As the lateral spacingincreased to 4D theblockage effect wasreduced
Background Computational Methodology Results Conclusions + future directions
Seven turbine simulations
• 3D spacing inflicted highenergy shadowing ondownstream row
• Increased lateral spacing(6D) reduces the wakeinteraction and theperformance of thedownstream row improved
• There was a blockageeffect on the performanceof the base turbine in themiddle row
Background Computational Methodology Results Conclusions + future directions
Seven turbine simulations
Background Computational Methodology Results Conclusions + future directions
Summary
• IBF model constructed and successfully validated – goodcomparison with experimental results
• Successfully generates large scale vortex behaviour – LES andfree surface simulation to capture other important effects
• Low cost method allows turbine/turbine interaction to besimulated. Wake decay also shown (single and multipleturbines)
• Able to examine other turbine issues (eg. effects of venturiplates)
Background Computational Methodology Results Conclusions + future directions
Future directions
Intention to take this work in two directions – higher detail andlarger scale.
Higher detail – PhD project (Matt Berry) to simulate detailedblade motion – MRF, GGI, overset meshing
Larger scale – EPSRC-funded project to develop farm modelling;ROM for individual turbines in linked array, poweroptimisation and flood risk modelling