Testing the Effect of Different Optimization Parameters on Layout ...
Testing the Effect of Different Optimization Parameters on Layout Design Using openWind®: A GIS Based Wind-Modeling Platform Meagan Krawczyk, Wind Resource Analyst, Shell Wind Nick Robinson, Director of openWind®, AWS Truepower Sara Tyler, Wind Resource Manager, Shell Wind ESRI Petroleum User Group Conference, Houston, TX April 18th 2011
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
1
Overview
What is openWind®? Optimizing wind farm layouts Wind farm layout design basics Test results / conclusions
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
2
openWind® and ArcGIS: Compatible, not competitive openWind®: Software developed by AWS Truepower, LLC A tool for the design, optimization, and assessment of wind power projects Open-source platform Patterned after Geographical Information Systems (GIS) Identical computations to those of other leading wind farm design programs
openWind® GIS functionality: Create and display vectors and rasters Clip vector layers Edit attributes Edit vectors Label features Export vectors/rasters to Google Earth Add/Subtract rasters Perform Vegetated and Non-Vegetated viewshed analysis (ZVI)
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
3
Approaches to wind farm layout design Standard industry practice is to optimize turbines for maximum production
Site Conditions
Wind Conditions
Optimization
Turbine Layout
openWind® Enterprise has the capability to optimize layouts using the Cost of Energy Optimizer (COE) Takes into account some project cost metrics
Site Conditions
Wind Conditions
Plant Cost Estimates
Financial Assumptions
Optimization
Turbine Layout
COE Testing: Hope to find a way to more automate the process of optimizing wind farm layouts for project value Tested whether or not the COE process had measurable impact on the resulting turbine layout
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
4
Wind Farm Layout Design Basics: Create a wind map Inputs to openWind® Digital Elevation Model (DEM) Roughness (RGH) – Values assigned to land cover for the site Meteorology data (TAB file) – Historical, representative met data for the site in the form of wind speed and direction (wind frequency)
Output (.WRG) A grid containing probability (P) and wind speed (U) at each pixel in the grid Defines wind conditions for the site.
Site Land cover (RGH)
Elevation Data (DEM)
Wind Rose (TAB file) Wind Map (.WRG)
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
5
Wind Farm Layout Design Basics: Capture site constraints Constraints: Each site will have specific considerations to avoid, i.e. roads, rivers, steep slopes, etc. Constraints are defined and mapped in order to create a buildable area in raster form In Shell Wind, constraints are handled within ArcGIS
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
6
Wind Farm Layout Design Basics: Define Turbine Specifications Turbine Specifications: Hub height – The height of the turbine at the center of the hub Rotor Diameter – The diameter of the circular path which the turbine blades rotate within Cut in/out – The wind speed in which the turbine will start spinning (cut-in), and discontinue spinning (cut-out) Power Curve – The relationship between power production of the turbine and incoming wind speed Thrust – Used to estimate impact of turbine on downstream wind flow RPM – Rotational speed of the turbine rotor
Turbine Hub Height
Turbine Rotor Diameter
Side View of Turbine Nacelle * Photos courtesy of Vestas V100 Turbine Brochure
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind COE Optimizer
7
Wind Farm layout Design Basics: Choose a spacing Spacing : Turbines need to be properly spaced from each other to lessen wake effects Min. Rotor Spacing
Max. Rotor Spacing
Dominant Wind Direction
Horns Rev Wind Farm : Photographed by Christian Steiness
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
8
Wind Farm Layout Design Basics: Choose an optimizer Optimize Layouts: There are 3 main ways to optimize layouts: Gridded Energy Cost of Energy 1. Gridded Layout Optimizations – Turbines are packed into a linear array defined by angles, spacing and an area
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
9
Wind farm Layout Design Basics: Optimize for energy 2. Optimize for Energy (maximum capacity factor):
Inputs: Wind Resource Grid Turbine Specifications Turbine Spacing Number of Turbines Constraints
Output: Turbine Layout
Capacity Factor: The ratio of estimated actual output of energy over a period of time and it’s output if it had operated at full capacity
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
10
Wind Farm Layout Design Basics: Optimize for cost of energy Inputs: Same as energy optimizer AND: Costs: Roads Cables Turbine Financial Files: Existing roads Water bodies Existing Cables Cost Multipliers Starting points (nodes) for cables and roads must be defined Should be as centrally located as possible Road start node should connect to an existing main road
Outputs: Turbine Layout Road Design Collection System Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
11
Wind Farm Layout Design Basics: Assess your design Generate layout-specific producible energy estimates based on: Input wind map (WRG) Wind speed distribution (.tab file) Turbine model specifications Site specific air density Site reference height
Estimates Include: Gross Energy : Energy before wake effects Net Energy : Gross Energy – site specific losses (i.e. wake effects) Capacity Factor : Net Energy / (#MW *8760) Example: 10 Turbine – 2 MW Machine:
Cost of energy
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
12
Testing Methodology: COE vs. Energy Two test sites were selected to run both the energy optimizer and the COE optimizer Inputs were kept the same for both optimizers, other than inputs specific to COE COE inputs were defined by engineers within Shell Wind Each optimization was run for the same number of iterations Reports were run for each scenario and results compared
Reported conclusions are based on averaged results from 2 sites (4 scenarios) Each of the 4 scenarios had a different total number of turbines/MW : 100, 250, 400 and 380 MW scenarios This methodology was chosen to see if an increase in available land (by decreasing the number of turbines) would increase the difference between costs
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
13
Results: COE vs. Energy Designs were compared on basis of the following parameters: Gross Energy
Net Energy
Capacity Factor
Array Efficiency
COE ~0.08% lower
COE ~1.41% lower
COE ~1.40% lower
COE ~0.61% lower
Cost/MWh
Total Cable Cost
Total Road Cost
COE ~1.90% lower
COE ~17.7% lower
COE ~30.9% lower
Total Cable Length
Total Road Length
COE ~12.0% lower
COE ~17.7% lower
Across the board, COE optimized layouts were lower in each category than energy optimized layouts Scenarios with a higher ratio of land to turbines had a higher difference between energy and cost between the COE and energy optimized layouts
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
14
Results: 100 MW Vestas V112 Scenario: Turbine Road Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Cable Node
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
15
Results: 400 MW Vestas V112 Scenario: Road Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Cable Node
Turbine
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
16
Results: 250 MW Vestas V100 Scenario: Road Node
Cable Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Turbine
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
17
Results: 380 MW SWT 101
Scenario: Road Node Cable Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Turbine
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
18
Testing Methodology: Sensitivity to DEM resolution and start nodes 50 meter and 10 meter resolution DEMs were tested Start nodes for cables and roads were shifted The start nodes were shifted from the 1st location to different locations that were still feasible for the project
50 m DEM, 1st set start nodes
10 m DEM, 2ND set start nodes
*Red circle represents Cable Start Node (Substation), Yellow circle represents Road Start Node
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
19
Results: Sensitivity to DEM resolution and start nodes Comparison 1: Variable is DEM Site 1: 50 m DEM, 1st set start nodes (base case) Site 2: 10 m DEM, 1st set start nodes (comparator)
Comparison 2: Variables are DEM and start nodes Site 1: 50 m DEM, 1st set start nodes (base case) Site 3: 10m DEM, 2nd set start nodes (comparator)
Comparison 3: Variable is start nodes Site 2: 10 m DEM, 1st set start nodes (base case) Site 3: 10m DEM, 2nd set start nodes (comparator)
Comparison
Gross Energy
Net Energy
Capacity Factor
Array Efficiency
Cost/MWh
Total Cable Cost
Total Road Cost
Total Cable Length
Total Road Length
1
+ 0.16%
- 0.14%
Negligible
-0.34%
+ 0.32%
-1.52%
+ 4.38%
-2.02%
+ 1.02%
2
+ 0.11%
- 0.17%
- 0.28%
- 0.34%
- 0.04%
- 4.83%
- 0.26%
- 3.32%
- 2.18%
3
- 0.05%
- 0.03%
Negligible
Negligible
- 0.35%
- 3.26%
- 4.85%
-1.28%
- 3.24%
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
20
Results: Sensitivity to DEM resolution and start nodes
250 MW Vestas V100 sensitivities:
10 m DEM, 1st set start nodes
50 m DEM, 1st set start nodes
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
10 m DEM, 2ND set start nodes
21
Conclusions: The COE optimizer produces layouts that are more linear and less like the energy optimized layouts for sites that are less constrained (or have more available land) Based on the cost calculator within openWind®, COE optimized layouts cost less to build than energy optimized layouts per MW installed capacity COE optimized layouts produce less energy per MW installed than energy optimized layouts Energy optimized layouts are optimized for energy, and therefore tend to be more spread out to minimize wake effects COE optimized layouts minimize the cost of energy rather than maximizing the energy capture and tend to be more tightly packed
COE optimized layouts are generally more linear in their design, which could make construction and maintenance easier Using a higher resolution DEM impacts COE optimized designs, indicating that the highest resolution of DEM available to the user should be used Shifting the start nodes for cables and roads has shown to affect the outcome of the COE optimized layout. The user should be aware of this relationship and test various scenarios to see what fits best for the site
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
22
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
1
Overview
What is openWind®? Optimizing wind farm layouts Wind farm layout design basics Test results / conclusions
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
2
openWind® and ArcGIS: Compatible, not competitive openWind®: Software developed by AWS Truepower, LLC A tool for the design, optimization, and assessment of wind power projects Open-source platform Patterned after Geographical Information Systems (GIS) Identical computations to those of other leading wind farm design programs
openWind® GIS functionality: Create and display vectors and rasters Clip vector layers Edit attributes Edit vectors Label features Export vectors/rasters to Google Earth Add/Subtract rasters Perform Vegetated and Non-Vegetated viewshed analysis (ZVI)
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
3
Approaches to wind farm layout design Standard industry practice is to optimize turbines for maximum production
Site Conditions
Wind Conditions
Optimization
Turbine Layout
openWind® Enterprise has the capability to optimize layouts using the Cost of Energy Optimizer (COE) Takes into account some project cost metrics
Site Conditions
Wind Conditions
Plant Cost Estimates
Financial Assumptions
Optimization
Turbine Layout
COE Testing: Hope to find a way to more automate the process of optimizing wind farm layouts for project value Tested whether or not the COE process had measurable impact on the resulting turbine layout
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
4
Wind Farm Layout Design Basics: Create a wind map Inputs to openWind® Digital Elevation Model (DEM) Roughness (RGH) – Values assigned to land cover for the site Meteorology data (TAB file) – Historical, representative met data for the site in the form of wind speed and direction (wind frequency)
Output (.WRG) A grid containing probability (P) and wind speed (U) at each pixel in the grid Defines wind conditions for the site.
Site Land cover (RGH)
Elevation Data (DEM)
Wind Rose (TAB file) Wind Map (.WRG)
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
5
Wind Farm Layout Design Basics: Capture site constraints Constraints: Each site will have specific considerations to avoid, i.e. roads, rivers, steep slopes, etc. Constraints are defined and mapped in order to create a buildable area in raster form In Shell Wind, constraints are handled within ArcGIS
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
6
Wind Farm Layout Design Basics: Define Turbine Specifications Turbine Specifications: Hub height – The height of the turbine at the center of the hub Rotor Diameter – The diameter of the circular path which the turbine blades rotate within Cut in/out – The wind speed in which the turbine will start spinning (cut-in), and discontinue spinning (cut-out) Power Curve – The relationship between power production of the turbine and incoming wind speed Thrust – Used to estimate impact of turbine on downstream wind flow RPM – Rotational speed of the turbine rotor
Turbine Hub Height
Turbine Rotor Diameter
Side View of Turbine Nacelle * Photos courtesy of Vestas V100 Turbine Brochure
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind COE Optimizer
7
Wind Farm layout Design Basics: Choose a spacing Spacing : Turbines need to be properly spaced from each other to lessen wake effects Min. Rotor Spacing
Max. Rotor Spacing
Dominant Wind Direction
Horns Rev Wind Farm : Photographed by Christian Steiness
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
8
Wind Farm Layout Design Basics: Choose an optimizer Optimize Layouts: There are 3 main ways to optimize layouts: Gridded Energy Cost of Energy 1. Gridded Layout Optimizations – Turbines are packed into a linear array defined by angles, spacing and an area
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
9
Wind farm Layout Design Basics: Optimize for energy 2. Optimize for Energy (maximum capacity factor):
Inputs: Wind Resource Grid Turbine Specifications Turbine Spacing Number of Turbines Constraints
Output: Turbine Layout
Capacity Factor: The ratio of estimated actual output of energy over a period of time and it’s output if it had operated at full capacity
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
10
Wind Farm Layout Design Basics: Optimize for cost of energy Inputs: Same as energy optimizer AND: Costs: Roads Cables Turbine Financial Files: Existing roads Water bodies Existing Cables Cost Multipliers Starting points (nodes) for cables and roads must be defined Should be as centrally located as possible Road start node should connect to an existing main road
Outputs: Turbine Layout Road Design Collection System Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
11
Wind Farm Layout Design Basics: Assess your design Generate layout-specific producible energy estimates based on: Input wind map (WRG) Wind speed distribution (.tab file) Turbine model specifications Site specific air density Site reference height
Estimates Include: Gross Energy : Energy before wake effects Net Energy : Gross Energy – site specific losses (i.e. wake effects) Capacity Factor : Net Energy / (#MW *8760) Example: 10 Turbine – 2 MW Machine:
Cost of energy
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
12
Testing Methodology: COE vs. Energy Two test sites were selected to run both the energy optimizer and the COE optimizer Inputs were kept the same for both optimizers, other than inputs specific to COE COE inputs were defined by engineers within Shell Wind Each optimization was run for the same number of iterations Reports were run for each scenario and results compared
Reported conclusions are based on averaged results from 2 sites (4 scenarios) Each of the 4 scenarios had a different total number of turbines/MW : 100, 250, 400 and 380 MW scenarios This methodology was chosen to see if an increase in available land (by decreasing the number of turbines) would increase the difference between costs
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
13
Results: COE vs. Energy Designs were compared on basis of the following parameters: Gross Energy
Net Energy
Capacity Factor
Array Efficiency
COE ~0.08% lower
COE ~1.41% lower
COE ~1.40% lower
COE ~0.61% lower
Cost/MWh
Total Cable Cost
Total Road Cost
COE ~1.90% lower
COE ~17.7% lower
COE ~30.9% lower
Total Cable Length
Total Road Length
COE ~12.0% lower
COE ~17.7% lower
Across the board, COE optimized layouts were lower in each category than energy optimized layouts Scenarios with a higher ratio of land to turbines had a higher difference between energy and cost between the COE and energy optimized layouts
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
14
Results: 100 MW Vestas V112 Scenario: Turbine Road Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Cable Node
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
15
Results: 400 MW Vestas V112 Scenario: Road Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Cable Node
Turbine
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
16
Results: 250 MW Vestas V100 Scenario: Road Node
Cable Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Turbine
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
17
Results: 380 MW SWT 101
Scenario: Road Node Cable Node
Energy Optimization
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
Turbine
Cost of Energy Optimization
ESRI PUG Presentation - openWind ® COE Optimizer
18
Testing Methodology: Sensitivity to DEM resolution and start nodes 50 meter and 10 meter resolution DEMs were tested Start nodes for cables and roads were shifted The start nodes were shifted from the 1st location to different locations that were still feasible for the project
50 m DEM, 1st set start nodes
10 m DEM, 2ND set start nodes
*Red circle represents Cable Start Node (Substation), Yellow circle represents Road Start Node
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
19
Results: Sensitivity to DEM resolution and start nodes Comparison 1: Variable is DEM Site 1: 50 m DEM, 1st set start nodes (base case) Site 2: 10 m DEM, 1st set start nodes (comparator)
Comparison 2: Variables are DEM and start nodes Site 1: 50 m DEM, 1st set start nodes (base case) Site 3: 10m DEM, 2nd set start nodes (comparator)
Comparison 3: Variable is start nodes Site 2: 10 m DEM, 1st set start nodes (base case) Site 3: 10m DEM, 2nd set start nodes (comparator)
Comparison
Gross Energy
Net Energy
Capacity Factor
Array Efficiency
Cost/MWh
Total Cable Cost
Total Road Cost
Total Cable Length
Total Road Length
1
+ 0.16%
- 0.14%
Negligible
-0.34%
+ 0.32%
-1.52%
+ 4.38%
-2.02%
+ 1.02%
2
+ 0.11%
- 0.17%
- 0.28%
- 0.34%
- 0.04%
- 4.83%
- 0.26%
- 3.32%
- 2.18%
3
- 0.05%
- 0.03%
Negligible
Negligible
- 0.35%
- 3.26%
- 4.85%
-1.28%
- 3.24%
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
20
Results: Sensitivity to DEM resolution and start nodes
250 MW Vestas V100 sensitivities:
10 m DEM, 1st set start nodes
50 m DEM, 1st set start nodes
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
10 m DEM, 2ND set start nodes
21
Conclusions: The COE optimizer produces layouts that are more linear and less like the energy optimized layouts for sites that are less constrained (or have more available land) Based on the cost calculator within openWind®, COE optimized layouts cost less to build than energy optimized layouts per MW installed capacity COE optimized layouts produce less energy per MW installed than energy optimized layouts Energy optimized layouts are optimized for energy, and therefore tend to be more spread out to minimize wake effects COE optimized layouts minimize the cost of energy rather than maximizing the energy capture and tend to be more tightly packed
COE optimized layouts are generally more linear in their design, which could make construction and maintenance easier Using a higher resolution DEM impacts COE optimized designs, indicating that the highest resolution of DEM available to the user should be used Shifting the start nodes for cables and roads has shown to affect the outcome of the COE optimized layout. The user should be aware of this relationship and test various scenarios to see what fits best for the site
Copyright of Royal Dutch Shell plc
Date 9 May, 2011
ESRI PUG Presentation - openWind ® COE Optimizer
22
