+ All Categories
Home > Documents > Advanced Energy Estimations - Project Hunflen

Advanced Energy Estimations - Project Hunflen

Date post: 05-Apr-2018
Category:
Upload: haseebahmad4
View: 222 times
Download: 0 times
Share this document with a friend

of 25

Transcript
  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    1/25

    2011-12-16

    Advanced Energy EstimationsProject Assignment HunflenPaul Hines & Haseeb Ahmad

    12/15/2011

    Examiner: Stefan Ivanell

    This report will use the software WindSim to estimate the annual energy production of 3turbines at Hunflen in Sweden. Turbulence model RNG and Wake Model 1 will be employedin the simulation in WindSim.

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    2/25

    2

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    3/25

    3

    1. Introduction ..................................................................................................................................... 5

    1.1. Background ............................................................................................................................. 5

    1.2. Aim and question formulation ................................................................................................. 5

    1.3. Delimitations ........................................................................................................................... 6

    2. Theoretical framework .................................................................................................................... 6

    2.1. Navier-Stokes equations .......................................................................................................... 6

    2.2. Turbulent Flow Solutions ........................................................................................................ 6

    2.2.1. Reynolds Averaged Navier-Stokes Equations ................................................................. 7

    2.2.2. Direct Numerical Simulation (DNS) ............................................................................... 9

    2.2.3. Large eddy simulation (LES) .......................................................................................... 92.2.4. Detached eddy simulation (DES) .................................................................................... 9

    2.3. Weibull distribution ............................................................................................................... 10

    3. Analysis ......................................................................................................................................... 10

    3.1. Methodology ......................................................................................................................... 10

    3.1.1. Wake effects .................................................................................................................. 13

    3.1.2. Turbulence model .......................................................................................................... 14

    4. Results ........................................................................................................................................... 15

    4.1. Production according to vindstat.nu ...................................................................................... 15

    4.2. WindSim Results ................................................................................................................... 16

    4.2.1. Wake model 1 ................................................................................................................ 16

    4.2.2. Wind Resources ............................................................................................................. 19

    4.3. Comparison with other Turbulence models .......................................................................... 20

    5. Discussion ..................................................................................................................................... 22

    5.1. VilhelmVestas V52 ........................................................................................................... 22

    5.2. FerdinandVestas V52 ........................................................................................................ 22

    5.3. Freja NEG Micon 52 ............................................................................................................. 23

    6. Conclusion ..................................................................................................................................... 23

    Bibliography .......................................................................................................................................... 24

    Appendices ............................................................................................................................................ 25

    Table of Contents

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    4/25

    4

    Figure 1 Turbulence flow solution techniques ........................................................................................ 7

    Figure 2 Sample output of energy module ............................................................................................ 10

    Figure 3 Cell resolution at 5000 ............................................................................................................ 11

    Figure 4 Cell resolution at 300000 ........................................................................................................ 11

    Figure 5 Power curve Vestas V52 850 .................................................................................................. 12

    Figure 6 Power curve NEG Micon 52 900 ............................................................................................ 12

    Figure 7 Objects as placed in terrain module ........................................................................................ 13

    Figure 8 Yearly output 2009-2010 Ferdinand from vindstat.nu ............................................................ 15

    Figure 9 Yearly output Vilhelm 2009-2010 from vindstat.nu ............................................................... 15

    Figure 10 Power output from WindSim frequency table ...................................................................... 16

    Figure 11 Power output per turbineFrequency .................................................................................. 16

    Figure 12 Estimated productionactual production Frequency table .................................................. 17

    Figure 13 Actual vs estimated production frequency table ................................................................... 17

    Figure 14 Power output from WindSim frequency table ...................................................................... 17

    Figure 15 Power output per turbineWeibull distribution................................................................... 18

    Figure 16 Estimated productionactual production Weibull distribution ........................................... 18

    Figure 17 Actual vs estimated production frequency table ................................................................... 19

    Figure 18 Wind Resource map and Wind Rose at 300000 cell resolution ............................................ 19

    Figure 19 Turbulence model comparisonFerdinand .......................................................................... 21Figure 20 Turbulence model comparison - Freja .................................................................................. 21

    Figure 21 Turbulence model comparison - Vilhelm ............................................................................. 22

    List of Figures

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    5/25

    5

    1. Introduction1.1. BackgroundIn spite of early discussion questioning the profitability of wind power in forest

    environments (Wizelius) interest in harvesting wind resource from complex/ and or hilly

    terrain is growing in for example in Sweden with a number of projects in planning. Vindkraft

    Norr is a joint venture between Statkraft, a large energy company and SCA a company

    owning large amounts of forest land (vindkraftnorr.se) . The absence of a nearby residential

    population can making planning easier but often the terrain can be more challenging.

    The first software providing wind resource estimations was developed in the 1980s.

    Windpro, a modular based Windows compatible software that can be used for design and

    planning of individual wind turbines or wind farms was developed over 20 years ago in

    lborg in Denmark. WAsP, the Wind Atlas Analysis and Application Program enables wind

    simulation and estimation of power output from wind turbines through the use of linear

    equations and has been in present in the industry for over 25 years (Facts about Ris DTU).

    However, the limitations of this software in complex terrain have been recognized (Wallbank,

    2008)

    Software models such as Windpro using computational fluid dynamics (CFD) have beenseen to have considerable advantages when mapping complex terrain. The founder of

    WindSim, Arne Grawdahl was working on the project to establish the Norwegian Wind Atlas.

    The use of CFD was required to simulate the complex Norwegian coastline. The first

    commercially available version of WindSim was launched in 2003.

    CFD will be examined later in a discussion of the theoretical framework underlying this

    study.

    1.2. Aim and question formulationThis report will use the software WindSim to estimate the annual energy production of 3

    turbines at Hunflen in Sweden. Hunflen lies in Dalarna in mid Sweden. The turbines are two

    Vestas V52 and one NEG Micon 52.

    Turbulence model RNG and Wake Model 1 will be employed in the simulation in WindSim.

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    6/25

    6

    1.3. DelimitationsThis report is produced under limited time frame and by users who are not well skilled in

    the use of CFD software. The users have the benefit of direction and assistance from members

    of the department of wind power at Gotland University but even given WindSims user

    friendly interface both of the above limitations much be acknowledged.

    2. Theoretical frameworkCFD uses a non-linear flow model based on Navier-Stokes equations. Navier-Stokes

    equations describe fluid flow based on the laws of conservation of momentum, mass and

    energy. (Karl Nilsson, Stefan Ivanell, 2010)

    2.1. Navier-Stokes equationsNavier-Stokes equations are used to explain the motion of a fluid i.e. liquid or gas. These

    equations are based on Newtons Second Law which describes the relation between force,

    mass and acceleration on a fluid. Navier Stock equations are quite useful in the modeling of

    weather, understanding the flow behavior of fluids, designing of wind turbines blades,

    aircraft, and in many other useful applications.

    Navier-Stokes Equations are non-linear, partial differential equations which do notexplicitly describe the variables but these present how variables change with time. The

    solution of Navier-Stokes Equations is velocity field which describe the velocity of fluid at a

    point in time. (T. Wallbank, 2008). The assumption, on which Navier-Stokes equations are

    based, is the continuous nature of fluid. The derivation of Navier-Stokes equations starts with

    the conservation of mass, momentum and energy conservation for a finite arbitrary volume.

    (T. Wallbank, 2008)

    2.2. Turbulent Flow SolutionsThe turbulence can be defined as the state of motion of a fluid which is characterized by

    apparently random and chaotic three dimensional vorticity. (Introduction to turbulence/Nature

    of turbulence, 2011) Vorticity can be defined as the measure of the rate of rotational spin in a

    fluid. Turbulence dominates all other flow phenomena, and results in increased energy

    dissipation, mixing, heat transfer, and drag.

    Turbulent flows can be computed either by solving the Reynolds Average Navier-Stokes

    (RANS) equations with suitable models or with direct computation.

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    7/25

    7

    2.2.1. Reynolds Averaged Navier-Stokes EquationsThe Reynolds Averaged Navier-Stokes Equations are the simplification of Navier-Stokes

    Equations by taking the time average of the velocity terms in the equations. RNS equations

    are used to describe turbulent flows. The basic tool which is required to derive the RNS

    equations from Instantaneous Navier-Stokes equations is the Reynolds decomposition. The

    Reynold decomposition means the separation of variable into the mean (time averaged)

    component and fluctuating component. By this transformation, we get a set of unknowns

    called Reynold Stresses which are the functions of velocity fluctuations and which require a

    turbulence model to produce a closed system of solvable equations. The computational

    requirements for RANS equations are far less than Navier-Stokes equations. (symscape, 2009)

    Turbulence ModelsTurbulence modeling is used to calculate the effects of turbulence in fluids. By taking

    average, the solution of turbulence equations can be simplified but models are required to

    represent scales of the flow that are not resolved. (Ching Jen Chen, 1998)

    Figure 1 Turbulence flow solution techniques

    RANS based

    turbulence

    models

    Large eddy

    simulation (LES)

    Detached eddy

    simulation (DES)

    Direct Numerical

    Simulation DNS

    Linear eddy

    viscosity Models

    Non-Linear eddy

    viscosity Models

    Reynolds stress

    Models

    Algebric Models One equation

    Models

    Two equation

    Models

    k-epsilon Model k-omega ModelRealisability

    Issues

    Near Wall

    Treatment

    RNG k-epsilon

    Model

    Realisable k-

    epsilon Model

    Standard k-

    epsilon Model

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    8/25

    8

    Following turbulence model is used in WindSim

    K epsilon turbulence model

    k-epsilon turbulence model is one of the most common turbulence models however it does not

    work well in the cases where large pressure gradient occurs. (Wilcox, 1998, p. 174). Thismodel came from the main branch of turbulent solution techniques i.e. RANS based

    turbulence models. The sub-branch of RANS based turbulence models is the linear eddy

    viscosity models, it can be seen in figure below. It is a two equation turbulence model which

    means it employs two extra transport equations to describe turbulent flow behavior.

    The first transported variable is turbulent kinetic energy and second transported variable is

    turbulent dissipation.

    Turbulent kinetic energy

    The turbulent kinetic energy is simply the energy in the turbulence. If the flow can be

    partitioned into mean and turbulent parts, then the total kinetic energy of the flow will simply

    be the sum of the kinetic energy of the mean and turbulent flows. (Turbulence Intensity and

    Turbulent Kinetic Energy, 2011)

    Turbulent dissipation

    Turbulent dissipation describes the scale of the turbulence.

    Some usual models of k-epsilon models are

    Standard k-epsilon model

    Realisable k-epsilon model

    RNG k-espsilon model

    WindSim uses k-epsilon as the turbulent model with standard form as well as modified

    forms. The standard k-epsilon model is widely used turbulent model and has been verified

    and validated for a wide variety of flows. It has less computational costs and is numerically

    more stable than the more advanced and complex stress models, it is more successful in flow

    where the normal Reynolds stresses are less important. In wind engineering, k-epsilon model

    doesnt perform well because its inability to cope w ith normal stresses which are more

    dominant in wind flows. (Veersteeg, H.K., and Malalasekera, W., 1995)

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    9/25

    9

    WindSim uses some modified k-epsilon models like RNG k-epsilon model, k-epsilon

    model with YAP correction. The RNG k-epsilon is based on the renormalization group

    analysis of the Navier-Stoke equations. The transport equations for turbulence generation and

    dissipation are the same as those for the standard model but the model differs because of one

    additional constant which improves the performance for separating flow and recirculation

    regions.

    One of the main inadequacies of k-epsilon model is the over estimation of turbulent

    kinetic energy however the slight improvement has been achieved after the development of

    modified k-epsilon models.

    2.2.2. Direct Numerical Simulation (DNS)Direct Numerical Simulation is used in computational fluid dynamics to solve the Navier-

    Stoke equations numerically without any turbulence model. This means that the whole range

    of spatial and temporal scales of the turbulence must be resolved. The power required to

    resolve such models with current computational capabilities, makes them inappropriate for

    large CDF applications. (Direct numerical simulation (DNS), 2007)

    2.2.3. Large eddy simulation (LES)Large eddy simulation is a popular technique used for the simulation of turbulent flows.

    This feature allows one to explicitly solve for the large eddies in a calculation and implicitly

    account for the small eddies by using a subgrid-scale model (SGS model). (Large eddy

    simulation (LES), 2007). The power requirement for LES is less than DNS but more than

    RNS. The RANS methods give a time averaged result while LES methods are able to resolve

    turbulent flow structures and predict instantaneous flow characteristics.

    2.2.4. Detached eddy simulation (DES)It is the hybrid technique which combines the best aspects of RANS and LES

    methodologies in a single solution strategy. There are some difficulties associated with the

    use of the standard LES models, particularly in near-wall regions. These issues lead to the

    development of hybrid models like DES. This model attempts to treat near-wall regions in a

    RANS-like manner, and treat the rest of the flow in an LES-like manner. (Detached eddy

    simulation (DES), 2007)

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    10/25

    10

    2.3. Weibull distributionWindSim uses Weibull distribution to create a wind frequency table from met mast

    information (Wallbank, 2008)

    Figure 2 Sample output of energy module

    3.Analysis3.1. Methodology

    WindSim contains the following modules:

    Terrain module

    Establish the numerical model based on height and roughness data

    Wind Fields module

    Calculation of the numerical wind fields

    Objects module

    Place and process wind turbines and climatology data.

    Results module

    Analyse the numerical wind fields

    Wind Resources module

    Couple the numerical wind fields with climatology data by statistical means to provide the

    wind resource map

    Energy module

    Couple the numerical wind fields with climatology data by statistical means to provide the

    Annual Energy Production (AEP); including wake losses. Determine the windcharacteristics used for turbine loading. (WindSim, 2011)

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    11/25

    11

    The terrain model generates a 3D model of the area under examination. Input includes

    coordinates, height and roughness. A map is first converted using the terrain module. In

    Refinement Type the Refinement area is detailed along with the number of cells to be used this

    allows greater accuracy in computations for the chosen area. (WindSim, 2011) Resolutions in

    the range 5000, 10000, 50000, 80000, 100000, 150000, 200000, 250000, 300000 will be

    selected and the results recorded.

    Cell resolution at 300000 and 5000 are shown below.

    Figure 3 Cell resolution at 5000

    Figure 4 Cell resolution at 300000

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    12/25

    12

    The Objects module is used to specify each of the proposed turbines. The turbine models are

    as follows:

    Vestas V52 Ferdinand, 850 KW, hub height 65m

    Vestas V52 Vilhelm, 850 KW, hub height 65m

    NEG Micon 52, Freja, 900 KW, hub height 49m

    Figure 5 Power curve Vestas V52 850

    Figure 6 Power curve NEG Micon 52 900

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    13/25

    13

    Figure 7 Objects as placed in terrain module

    The Energy results for each increasing resolution will then be presented and discussed. A

    wind resource map for the highest resolution will also be recorded.

    3.1.1. Wake effectsThe WindSim wind resource model provides for the calculation of wake effects based on

    analytical models (WindSim, 2011). In this report Model 1 has been chosen. This model is

    based on momentum deficit theory and gives a simple linear expansion of the wake on the

    basis of the wake factor, k.

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    14/25

    14

    V = (1 - SQRT(1 - CT))/(1 + (2kx/D))2

    Where:

    CT = thrust coefficient (-)

    k = A/LOG(h/z0)

    A = 0.5

    h = hub height (m)

    z0 = roughness height (m)

    (Source: WindSim.com)

    3.1.2. Turbulence modelThe wind fields module allows for the selection of a Turbulence model. The default

    model is the standard k- model belonging to the family of eddy viscosity models. An eddy

    viscosity is calculated by an analytical equation (WindSim, 2011). The standard form of the

    k- model is summarized as follows, with, t denoting differentiation with respect to time and,

    i denoting differentiation with respect to distance:

    k),t + ( Ui k - {t/PRT(k)} k,i ),i = (Pk - )

    (),t + ( Ui - {t/PRT()} ,i ),i = {/k} (C1 Pk - C2 )

    t = C k2/

    Here k is the turbulent kinetic energy; is the dissipation rate; is the fluid density; t is the

    turbulent kinematic viscosity. C,C1, C2, PRT(k), PRT() are the model constants.

    (WindSim, 2011)

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    15/25

    15

    4. Results4.1. Production according to vindstat.nu

    The production data for NEG Micon 52 Freja is not available on vindstat.nu

    Vestas V52 850 KW Hunflen Ferdinand 800

    2009 2010

    January 104097

    February 64299 112608

    March 151877 158710

    April 125396

    May 165342

    June 114888

    July 132393

    August 141227

    September 235166

    October 156036

    November 191045

    December 92123

    Figure 8 Yearly output 2009-2010 Ferdinand from vindstat.nu

    Average yearly output Feb 2009Jan 2010 is 1673, 8 MW

    Vestas V52 850KW Hunflen Vilhelm 801

    2009 2010

    January 98847

    February 50779 87414

    March 132320 142738

    April 109980

    May 107988June 92666

    July 118927

    August 125998

    September 215377

    October 141055

    November 180530

    December 87431

    Figure 9 Yearly output Vilhelm 2009-2010 from vindstat.nu

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    16/25

    16

    Average yearly output Feb 2009Jan 2010 is 1461,8 MW

    4.2. WindSim Results4.2.1. Wake model 1

    Frequency Table Power production in MWh/y for the differing cell resolutions based on

    frequency table is presented below.

    WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000

    Ferdinand 1867,1 2003,7 2110,4 2152,8 2214,8 2236,3 2088,7 2241,3 2259,6

    Vilhelm 1946,5 1995,7 2086,9 2164,5 1972,3 2088,9 1847,1 2089,8 1960,4

    Freja 1337,7 1508,0 2036,0 2221,4 2098,2 2200,6 2012,2 2224,3 2101,4

    Total 5151,3 5507,4 6233,3 6538,7 6285,3 6525,8 5948,0 6555,4 6321,4

    Figure 10 Power output from WindSim frequency table

    The table below shows the variation in estimated energy output across differing cell

    resolutions.

    Figure 11 Power output per turbine Frequency

    The table below shows difference in actual production and forecast production for each of the

    two turbines for which information is present on vindstat.nu.

    Difference Ferdinand = Estimated production - Actual production 1673,8

    Difference Vilhelm = Estimated productionActual production 1461,8

    1000.0

    1200.0

    1400.0

    1600.0

    1800.0

    2000.0

    2200.0

    2400.0

    P

    o

    w

    e

    r

    o

    u

    t

    p

    u

    t

    Cell Resolution

    Frequency

    ferdinand

    wilhelm

    freja

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    17/25

    17

    Figure 12 Estimated production actual production Frequency table

    WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000

    Ferdinand 1867,1 2003,7 2110,4 2152,8 2214,8 2236,3 2088,7 2241,3 2259,6Actual 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9Difference 193,2 329,8 436,5 478,9 540,9 562,4 414,8 567,4 585,7

    % 11,5426411 19,70328 26,07766 28,61068 32,31463 33,59906 24,78127 33,89777 34,99103

    Figure 13 Actual vs estimated production frequency table

    Weibull Distribution Power production in MWh/y for the differing cell resolutions based on

    Wiebull distribution is presented below.

    WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000

    Ferdinand 1908,6 2045,3 2146,3 2198,0 2252,0 2280,8 2125,0 2284,3 2305,5Vilhelm 1986,7 2035,9 2125,2 2211,6 2010,3 2135,4 1885,7 2134,7 2006,6Freja 1375,1 1548,4 2077,8 2266,2 2139,2 2246,0 2053,5 2268,2 2145,7

    Total 5270,4 5629,6 6349,3 6675,8 6401,5 6662,2 6064,2 6687,2 6457,8Figure 14 Power output from WindSim frequency table

    0.0

    100.0

    200.0

    300.0

    400.0

    500.0

    600.0

    700.0

    800.0

    po

    w

    e

    r

    o

    u

    t

    p

    u

    t

    Cell resolution

    Difference between estimated and Actual

    production Frequency table

    Ferdinand

    Vilhelm

    WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000

    Wilhelm 1946,5 1995,7 2086,9 2164,5 1972,3 2088,9 1847,1 2089,8 1960,4Actual 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9Difference 484,6 533,8 625,0 702,6 510,4 627,0 385,2 627,9 498,5% 33,1488243 36,51431 42,75278 48,06095 34,91365 42,88959 26,34944 42,95115 34,09964

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    18/25

    18

    Figure 15 Power output per turbine Weibull distribution

    The table below shows the variation in estimated energy output across differing cell

    resolutions. The table below shows difference in actual production and forecast production for

    each of the two turbines for which information is present on vindstat.nu.

    Difference Ferdinand = Estimated production - Actual production 1673,8

    Difference Vilhelm = Estimated productionActual production 1461,8

    Figure 16 Estimated production actual production Weibull distribution

    1000.0

    1200.0

    1400.0

    1600.0

    1800.02000.0

    2200.0

    2400.0

    P

    o

    w

    e

    r

    o

    ut

    p

    u

    t

    Cell Resolution

    Weibull distribution

    Ferdinand

    Wilhelm

    Freja

    0.0

    200.0

    400.0

    600.0

    800.0

    1000.0

    5000 10000 50000 80000 100000 150000 200000 250000 300000

    P

    o

    w

    e

    r

    o

    u

    t

    p

    u

    t

    Cell resolution

    Difference estimated vs Actual production

    Weibull distribution

    Ferdinand

    Vilhelm

    Freja

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    19/25

    19

    Figure 17 Actual vs estimated production frequency table

    4.2.2. Wind Resources

    WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000

    Ferdinand 1908,6 2045,3 2146,3 2198,0 2252,0 2280,8 2125,0 2284,3 2305,5Actual 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9 1673,9Difference 234,7 371,4 472,4 524,1 578,1 606,9 451,1 610,4 631,6

    % 14,0218975 22,18851 28,22236 31,31098 34,537 36,25754 26,94988 36,46664 37,73315WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000Wilhelm 1986,7 2035,9 2125,2 2211,6 2010,3 2135,4 1885,7 2134,7 2006,6Actual 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9 1461,9Difference 524,8 574,0 663,3 749,7 548,4 673,5 423,8 672,8 544,7% 35,8986742 39,26416 45,37266 51,28278 37,51301 46,07038 28,98985 46,0225 37,25992WTG 5000 10000 50000 80000 100000 150000 200000 250000 300000Freja 1375,1 1548,4 2077,8 2266,2 2139,2 2246,0 2053,5 2268,2 2145,7Actual 1317,1 1317,1 1317,1 1317,1 1317,1 1317,1 1317,1 1317,1 1317,1Difference 58,0 231,3 760,7 949,1 822,1 928,9 736,4 951,1 828,6

    % 4,40 17,56 57,76 72,06 62,42 70,53 55,91 72,21 62,91

    Figure 18 Wind Resource map and Wind Rose at 300000 cell resolution

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    20/25

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    21/25

    21

    Figure 19 Turbulence model comparison Ferdinand

    Figure 20 Turbulence model comparison - Freja

    1900.0

    1950.0

    2000.0

    2050.0

    2100.0

    2150.0

    2200.0

    2250.0

    2300.0

    10000 100000 150000 200000 250000 300000

    Po

    w

    e

    r

    o

    u

    t

    p

    u

    t

    Resolution

    Comparison RNG with Standard and Modified

    Turbulence models

    Ferdinand

    RNG

    Ferdinand

    Modifiied

    Ferdinand

    Standard

    1400.0

    1500.0

    1600.0

    1700.0

    1800.0

    1900.0

    2000.0

    2100.0

    2200.0

    2300.0

    10000 100000 150000 200000 250000 300000

    P

    o

    w

    e

    r

    o

    u

    t

    p

    u

    t

    Resolution

    Comparison RNG with Standard and

    Modified Turbulence models

    Freja RNG

    Freja Mod

    Freja

    Standard

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    22/25

    22

    Figure 21 Turbulence model comparison - Vilhelm

    5. DiscussionThe number of simulations within the limited time frame may affect the validity of the

    results obtained. The processing power available did not allow for production of energy

    estimations for cell resolution in excess of 300000.

    As production data for the NEG Micon Freja is unavailable discussion against actual

    production will be limited to the above results and is perhaps best examined per turbine.

    5.1. Vilhelm Vestas V52As cited above yearly power output based on vindstat.nu was 1461,8 MW. The nearest

    result to actual production comes at 200000 cell resolution at an estimated production of

    1847, 1 MWh/y for frequency table and 1885, 7 MWh/y for Weibull distribution. The power

    estimation is next closest at 5000 cell resolution at 1946, 5 MWh/y and 1986,7 MWh/yrespectively. The results in the range 10000- 150000 cell resolution show a fluctuation in

    power output as can be seen in Fig 11 and 14 above. At cell resolution 300000 the power

    output decreases again and it is possible with further resolution it would have decreased

    further.

    5.2. Ferdinand Vestas V52As cited above yearly power output based on vindstat.nu was 173,8 MW. The nearest

    result to actual production comes at 5000 cell resolution at an estimated production of 1867,1

    1700.0

    1750.0

    1800.0

    1850.0

    1900.0

    1950.0

    2000.0

    2050.0

    2100.0

    2150.0

    2200.0

    10000 100000 150000 200000 250000 300000

    P

    o

    w

    e

    r

    o

    u

    t

    p

    u

    t

    Resolution

    Comparison RNG with Standard and

    Modified Turbulence models

    Vilhelm RNG

    Vilhelm Mod

    Vilhelm

    standard

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    23/25

    23

    MWh/y for frequency table and 1908,6 MWh/y for Weibull distribution. The power

    estimation is next closest at 10000 cell resolution at 2003,7 MWh/y and 2045,3 MWh/y

    respectively. The results in the range 50000- 250000 cell resolution show a fluctuation in

    power output as can be seen in Fig 11 and 14 above. As opposed to the result for Vilhelm at

    cell resolution 300000 the power output does not decrease again however it is possible with

    further resolution it would have decreased.

    5.3. Freja NEG Micon 52The energy estimation results for Freja show a very similar pattern to Vilhelm. However,

    there is quite a marked jump in estimation from 5000 and 10000 cell resolutions producing a

    very low estimate when compared with 50000 and beyond. There is little stability that can be

    observed in the estimations with both frequency table and Weibull distribution generatingenergy estimates varying in size across the range of sampled resolutions.

    6. ConclusionThe energy estimations from WindSim clearly overestimate the production from the

    turbines as compared to actual production. As has been discussed in theoretical framework the

    k-epsilon model has a tendency to over-estimate. The impact of availability may have a

    significant impact on the difference in results along with the unavailability of simulations athigh resolution.

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    24/25

    24

    Bibliography

    Detached eddy simulation (DES). (Feburary 2007). Hmtat frn CFD-Online: http://www.cfd-

    online.com/Wiki/Detached_eddy_simulation_(DES) den 8 December 2011

    Direct numerical simulation (DNS). (den 2 Feburary 2007). Hmtat frn CFD-Online: http://www.cfd-online.com/Wiki/Direct_numerical_simulation_(DNS) den 8 December 2011

    Large eddy simulation (LES). (February 2007). Hmtat frn CFD-Online: http://www.cfd-

    online.com/Wiki/Large_eddy_simulation_(LES) den 8 December 2001

    (2011). Hmtat frn WindSim: http//www.WindSim.com den 7 December 2011

    Introduction to turbulence/Nature of turbulence. (den 3 September 2011). Hmtat frn Introduction to

    turbulence/Nature of turbulence: http://www.cfd-online.com/Wiki/Introduction_to_turbulence/Nature_of_turbulence#What_is_turbulence.3F

    den 7 December 2011

    Turbulence Intensity and Turbulent Kinetic Energy. (den 9 August 2011). Hmtat frn

    http://apollo.lsc.vsc.edu: http://apollo.lsc.vsc.edu/classes/met455/notes/section3/3.html den 8

    December 2011

    Ching Jen Chen, S.-Y. J. (1998). Fundamentals of Turbulence Modeling. Taylor & Francis.

    Facts about Ris DTU. (u.d.). Hmtat frn Risoe DTU National Laboratory for Sustainable Energy:

    http://www.risoe.dtu.dk/About_risoe/fakta_risoe.aspx den 6 12 2011

    Karl Nilsson, Stefan Ivanell. (2010). Wind Energy. Gotland University.

    Stangroom, P. (2004). CFD Modelling of Wind Flow over Terrain. University of Nottingham.

    symscape. (2009). symscape. Hmtat frn Reynolds-Averaged Navier-Stokes Equations:

    http://www.symscape.com/reynolds-averaged-navier-stokes-equations 2011

    Veersteeg, H.K., and Malalasekera, W. (1995).An Introduction to Computational Fluid Dynamics.

    Prentice Hall.

    vindkraftnorr.se. (u.d.). Hmtat frn http://www.vindkraftnorr.se/morttjarnberget.asp?id=7

    Wallbank, T. (2008). WindSim Validation Study: CFD Validation in Complex Terrain. WindSim.

    Wilcox, D. C. (1998). Turbulence Modeling for CFD; 2nd edition. D C W Industries.

    Wizelius, T. (u.d.). Warning for wind power in forests. Hmtat frn Nyteknik:http://www.nyteknik.se/nyheter/energi_miljo/vindkraft/article253329.ece den 5 12 2011

  • 7/31/2019 Advanced Energy Estimations - Project Hunflen

    25/25

    Appendices

    Table 1. Vindstat data, 850 Vestas Hunflen Ferdinand 800

    2005 2006 2007 2008 2009 2010 2011

    January 135509 198817 291270 277887 267069 104097 232270

    February 151846 145239 107846 241300 64299 112608 164087

    March 64267 126711 150467 215211 151877 158710 264617

    April 147311 154722 205126 115484 125396 140118 170173

    May 254261 100726 206088 93358 165342 108958 173498

    June 192459 152794 82501 135849 114888 102062 109826

    July 246063 132559 141528 87003 132393 187878 86645

    August 128883 90937 147009 118517 141227 127678 103911

    September 167647 249708 87951 235166 160690 187234

    October 149726 217889 244696 156036 212775 257853

    November 294222 225500 201147 191045 202575 200503

    December 349604 233641 151972 92123 132287

    Table 2. Vindstat data, 850 Vestas Hunflen Vilhelm 800

    2005 2006 2007 2008 2009 2010 2011

    January 103766 188613 259690 279079 244542 98847 200899

    February 131170 48898 100040 215324 50779 87414 145993

    March 57522 16772 153524 192645 132320 142738 190854

    April 115597 137374 181073 109201 109980 116687 130356

    May 236364 103011 186266 82239 107988 94455 147528

    June 176052 134896 81121 119988 92666 88379 99229

    July 224016 116525 126854 82524 118927 158707 74223

    August 169196 81255 137331 100446 125998 112743 92149

    September 146726 202549 84835 215377 124641 169207

    October 135108 70606 179583 141055 193865 233937November 239495 186387 190209 180530 183378 187747

    December 306282 211737 118489 87431 119595


Recommended