Bolivia Wind Atlas - FINAL · Headquartered in Cochabamba, TDE is Bolivia’s principal power...

!

!

June 5, 2009

Final Report Bolivia Wind Atlas

A project for the International Finance Corporation (IFC)

Copyright 2009 © 3TIER Environmental Forecast Group, Inc. All rights reserved. 3TIER claims a copyright in all proprietary and copyrightable text and graphics in this Report, the overall design of this Report, and the selection, arrangement and presentation of all materials in this Report, including information in the public domain. Reproduction and redistribution inconsistent with the scope of IFC’s contract with 3TIER for this project is prohibited without written permission. Requests for permission may be directed to [email protected].

ph: 206.325.1573 fax: 206.325.1518

[email protected] www.3tier.com 3TIER North America 2001 Sixth Avenue Suite 2100 Seattle, WA 98121

Bolivia Wind Atlas i

TABLE OF CONTENTS

I. Introduction ...............................................................................................................1 A. Background............................................................................................................................ 1 B. Project Objective.................................................................................................................... 2

II. Methodology.............................................................................................................4 TASK I – Inception Report: ........................................................................................................... 4 TASK II – Information Processing:................................................................................................ 4

Atmospheric model ................................................................................................................... 4 Input data .................................................................................................................................. 6 Model Simulations..................................................................................................................... 6

Task III: Data Validation................................................................................................................ 7 Task IV: Draft Final Report ........................................................................................................... 8 Task V – Dissemination ................................................................................................................ 9 Task VI – Final Report and Wind Atlas ....................................................................................... 10 Task VII: Training Workshop and Final Presentation ................................................................. 10

III. Climatology of Bolivia ..........................................................................................11

IV. Results...................................................................................................................14 A. Validation ............................................................................................................................. 14

Summary................................................................................................................................. 14 Validation Statistics................................................................................................................. 15 Tables of Validation Statistics (NCEP ADP Stations) ............................................................. 16 Simulation Bias ....................................................................................................................... 19 Monthly / Seasonal Cycle ....................................................................................................... 20 Diurnal Cycle........................................................................................................................... 20 Validation Conclusion ............................................................................................................. 21

B. Maps .................................................................................................................................... 22

V. Bolivia’s Wind Power Potential ............................................................................25 A. Wind Resource Basics......................................................................................................... 25 B. Spatial Distribution of Wind Resource ................................................................................. 26 C. Wind Power Density............................................................................................................. 27

VI. The Bolivia Wind Atlas in context .......................................................................28 A. Overview .............................................................................................................................. 28 B. Further Analysis ................................................................................................................... 30 C. Free Reference Materials..................................................................................................... 32

Appendix I: Calculation of Weibull parameters.........................................................I-1

Appendix II: Maps........................................................................................................II-1 A. Contents.............................................................................................................................. II-1 B. About the maps................................................................................................................. II-26 C. How to use digital maps in GeoPDF format ...................................................................... II-27

Appendix III: Additional Validation (Wind Roses) ...................................................III-1

Bolivia Wind Atlas 1!

!" !#$%&'()$*&#+

!

This report summarizes the wind resource over the country of Bolivia at three heights above ground level (20, 50, and 80 meters) based on the results of sophisticated meteorological simulations completed by 3TIER Environmental Forecast Group, Inc. (3TIER) for the International Finance Corporation (IFC) as part of IFC’s ongoing financing of Transportadora de Electricidad (TDE), Bolivia’s largest electricity transmission company. The results presented in this report shall serve as the basis for preliminary site assessment during the prospecting phase of wind project development. Before making investment decisions about specific project sites, any party wishing to harness wind energy in Bolivia should, as part of its due diligence process:

• Visit the location and review topographical, environmental, or logistical factors that could affect the successful installation and operation of wind turbines or transmission lines (as applicable);

• Collect on-site measurements at wind turbine hub height (varies with equipment) using a properly calibrated meteorological device (anemometer or other) or verify the observed wind resource at a neighboring location from data collected there. To gauge the availability of wind resources with the greatest possible precision, collecting wind measurements over extended periods of time is necessary;

• For off-grid systems, confirm that wind “supply” (times when the wind blows and with what strength) corresponds with anticipated energy demand and/or plan for back-up supply using batteries or other energy sources.

• Consider the long-term variability of the wind resource; and

• Obtain the necessary environmental and engineering permissions from Bolivian government authorities.

!" #$%&'()*+,-#

!

The contract between 3TIER and IFC for this project was initiated as part of an ongoing relationship between IFC and Transportadora de Electricidad S.A. (TDE), Bolivia’s largest electricity transmission company and a recipient of IFC loan financing. Under its corporate social responsibility program, TDE, part of the Spanish consortium Red Electrica de España, has undertaken a program to promote the access to and use of renewable energy in Bolivia, especially in areas near its installations in isolated rural communities. As part of its contribution to wind energy development in Bolivia, TDE


processed information collected during 10 years at 201 meteorological stations and began disseminating their analysis in April 2008. IFC, a member of the World Bank Group, fosters sustainable economic growth in developing countries by financing private sector investment, mobilizing private capital in local and international financial markets, and providing advisory and risk mitigation services to businesses and governments. IFC’s vision is that people should have the opportunity to escape poverty and improve their lives. In FY08, IFC committed $11.4 billion, and mobilized an additional $4.8 billion through syndications and structured finance for 372 investments in 85 developing countries. IFC also provided advisory services in 97 countries. For more information, visit www.ifc.org. Headquartered in Cochabamba, TDE is Bolivia’s principal power transmission company. TDE presently owns and operates 74% of the transmission network for Bolivia’s national interconnected system. The system consists of some 2,187 km of 230 kV, 115 kV, and 69 kV transmission lines and substations in various parts of Bolivia. The system interconnects the departments of La Paz, Cochabamba, Santa Cruz and Oruro, Chuquisaca and Potosi and the system facilitates the transmission of power and energy between the country’s generators and distributors and unregulated customers. TDE is owned by a subsidiary of the Spanish grid operator Red Electrica de España. For more information, visit: www.tde.com.bo. Founded in 1999, Seattle, Washington-based 3TIER is one of the largest independent providers of wind, solar and hydro energy assessment and power forecasting worldwide. 3TIER’s Regional Office for Latin America and the Caribbean in Panama City, Panama played an important role in the development of the Bolivia Wind Atlas and the company also has offices in Germany, India, and Australia. For more information, visit www.3tier.com.

!" #$%&'()*+,&'()-.'*

!

According to the Terms of Reference specified in 3TIER’s contract with IFC:

• The objective of this short-term contract is to prepare a geographic database of the wind energy potential of Bolivia.

• The assignment will create a map of wind energy resources throughout the

country. Once completed, data from this work will be shared not only within TDE’s project development offices, but also with public authorities responsible for rural electrification initiatives and with organizations and companies interested in the energy sector. TDE, through an agreement with the Catholic University of Bolivia (UCB – Universidad Católica de Bolivia) will provide indefinite universal access to the maps and database that supports the information that 3TIER provided to IFC.


• Information on wind potential will be prepared in easy to use and accessible

electronic format that will be made available over the Internet, or CD’s. The consultant firm is expected to prepare a plan for TDE, such that this database is properly maintained and/or improved over time. Once these studies are concluded, TDE in coordination with IFC will promote the dissemination of this information to all interested parties.


II. !"#$%&%'%()

!

!"#$%&%'%&()*+,-.(%/*+.0,1%

3TIER generated this report based on a joint mission to Bolivia in April 2008 as well as 3TIER’s internal activities. It offered a detailed plan for project execution.

!"#$%&&%'%&(2.034,-.(%50.)*66-(71%

*#+%,-$"./01+%&"'!

The spatial assessment of the wind resource across Bolivia presented in this report is

based on a randomized year of simulated data (January 01 through December 31)

using a regional nonhydrostatic primitive equation model of the atmosphere. 3TIER used version 3.0 of the Numerical Weather Prediction (NWP) model called the Weather Research and Forecasting (WRF) Model.". The WRF model has been developed in a collaborative partnership between federal agencies and universities in the United States and represents the next generation in weather forecast models. 3TIER uses this model for both its forecasting and resource assessment projects. With the WRF model, 3TIER

constructed a full year of data by individually simulating each calendar day of the year,

where the year is chosen at random from the last 10 years (1998–2007).

The WRF model utilizes a nested grid layout. The extent of the coarsest model grid was

selected to capture the e!ect of synoptic weather events on the wind resource in the

region of interest, as well as to allow the model to develop regional thermally-driven

circulations. The increasingly fine 54.0 km, 18.0 km, 6.0 km and 2.0 km grids were used

to simulate the e!ect of local terrain and local scale atmospheric circulations. Table 1

below lists some details of the final configuration of the Numerical Weather Prediction

(NWP) model.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1 Skamarock, W.C., J.B. Klemp, J. Dudhia, D.O. Gill, D.M. Barker, W. Wang, J.G. Powers, 2005: A

description of the Advanced Research WRF Version 2. NCAR Technical Note, NCAR/TN-468+STR, Boulder, Colorado, 88p.


!

Figure 1: The bold red box around Bolivia denotes the valid study area of the 2 km resolution grid domain used for this project.

Table 1: NWP Model Configuration


23-4#1&5#5!

Global Weather Archive: The main input data for wind resource assessment simulations are historic global weather archives, which are maintained by operational weather forecasting centers around the world including the United States National Center for Environmental Prediction (NCEP). These global archives represent the overall state of the atmosphere over the entire planet and are themselves the result of a sophisticated computer analysis of available surface and upper air observations. Each time period of analysis combines tens of thousands of individual measurements around the globe into a consistent physical state. The NCEP/NCAR reanalysis2 includes the NCEP global spectral model operational in 1995, with 28 sigma vertical levels and a horizontal triangular truncation of 62 waves, equivalent to about 210 km. The analysis scheme is a 3-dimensional variational (3D-Var) scheme cast in spectral space. Due to the necessity to represent the entire globe, the NCEP/NCAR reanalysis data set is maintained at a relatively coarse horizontal resolution and, by itself, does not contain the level of detail necessary to resolve the wind flow patterns over smaller geographic regions or over a single project. However, these data do provide a good representation of the history of large-scale spatial patterns in the atmosphere (i.e. the position of high and low pressure systems; the location of the jet stream) as well as the general state of the ocean (e.g. sea surface temperatures) and land surface condition (e.g. soil moistures). 3TIER maintains an archive of 40+ years of global weather data from NCEP at its Seattle headquarters. By combining these coarse data with high-resolution land-use data and a high-resolution numerical weather simulation model, 3TIER will create an accurate reconstruction of regional and site-specific wind fields.

High Resolution Terrain, Soil and Vegetation Data: For this project, 3TIER used high resolution 3 arc-second (roughly 90 m) terrain data from the Shuttle Radar Topography Mission (SRTM)3. In addition, WRF employs a 30 arc second global 24-category land use map (USGS), a 5 arc minute soil texture (FAO), and a 0.15-degree monthly climatology green vegetation fraction (NESDIS).

!%&"'16/+4'5#/%3,

3TIER implemented the WRF model in a nested-grid configuration. The simulation contained four nested domains. The outer domain has a spatial resolution of 54 km, and the other three had progressively finer resolutions of 18 km, 6 km, and 2 km. The innermost grid has a spatial resolution of 2 km covering Bolivia in its entirety (see Figure 1), plus a buffer zone in each direction to avoid grid edge effects. For each individual day, a 47-hour simulation was generated. 3TIER discarded the first 23 hours of each simulation to allow for proper model initialization, and used the last 24 hours of each simulation for the subsequent wind resource analysis. Sensitivity studies were done to determine the necessary initialization lead-time. One year was simulated by sampling

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!2 Available online at www.cpc.ncep.noaa.gov/products/wesley/reanalysis.html

3 Information available at http://srtm.usgs.gov


individual days from the 10-year period 1998-2007 of the NCEP/NCAR reanalysis archive to serve as boundary conditions for our NWP simulations. 3TIER sampled only from the satellite period of the reanalysis archive (1979 onwards) to avoid statistical discrepancies due to changes in the global upper air observing system. For this project, 3TIER applied version 3.0 of the WRF model. Having completed test runs with different configurations and reviewed them in light of rough-scale open-source data, 3TIER completed the 2km resolution model simulations (which yielded the so called “raw” simulation data that later was subjected to validation).

!468%&&&1%94,4%:4;-<4,-.(!

3TIER’s analysis of model output for this project focused on two aspects of the modeled wind fields: internal consistency and comparison with observations. The first part of the evaluation determined whether the modeled fields were subject to numerical instabilities and modeling artifacts, which would have been directly related to model setup and implementation. Model output was also subjected to quantitative controls of numerical stability based on Courant limits; visual inspection of wind speed fields focusing on detection of two-delta-t instabilities and/or spurious standing waves; and a qualitative analysis of the wind distribution at each grid point.

The second part of 3TIER’s evaluation relied on the availability of observations with which to compare the model simulations. 3TIER analyzed model output for this project by comparing it to data collected in Bolivia during time periods that overlap with the simulations (1998-2007). This process included a comparison of modeled and observed means, variances, diurnal distributions, and Weibull parameters, as well as correlation statistics computed on different time scales. From IFC, 3TIER received raw data from the JICA data set, raw data from the CRE wind resource assessments near Santa Cruz, and data from Bolivia’s Government Meteorological Agency (SENAMHI). Ideally, observed data to be used for comparison with simulation results should have been collected using an instrument located at least 20 meters above the ground surface for which the height above the ground surface is specified for each collection location. Data should have been recorded at hourly intervals during a full consecutive 12-month period. The data set provided should also include the coordinates for data collection. Unfortunately, none of the above-noted observed data sets had all of these three characteristics. Furthermore, the CRE, SENAMHI, and JICA data provided by IFC did not coincide with the random year of meteorology that 3TIER simulated during the 10-year record period enough to permit the detailed analysis called for in the Terms of Reference for this project. 3TIER’s summary analysis of its simulation results considered the JICA observed data and data from 3 additional tall meteorological towers in northern Chile within the rectangular Bolivia simulation domain4. To permit a

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

4 3TIER obtained this data from the website of a GEF-funded UNDP and Chilean Government study of renewable energy resources in Chile. (See http://www.renovables-rural.cl/actividades/fr_actividades.html).


more robust statistical validation analysis, 3TIER also downloaded observed data corresponding to 22 meteorological stations from the National Center for Environmental Prediction (NCEP) Automated Data Processing (ADP) Global Surface Observations.5 3TIER evaluated the quality of the wind resource simulations against observed data both qualitatively and quantitatively. Qualitative evaluation included visual inspection of the annual, seasonal, monthly and diurnal characteristics of 3TIER’s model simulations and a review of what is known about the meteorology of Bolivia. The amount and quality of meteorological observations did not permit 3TIER to perform a model-output-statistics (MOS) correction to the simulated wind fields for Bolivia. For validation results, please see section IV, a., which begins on page 12.

!468%&:1%9042,%=-(4;%/*+.0,

The deliverable associated with this task was a series of data sets that 3TIER calculated using the model simulation output and delivered to IFC. They included gridded data sets of each of the following:

• On a horizontal 2 km grid: annual and monthly average wind speed, wind power density, Weibull k, air temperature and air density calculations at 20 m, 50 m and 80 m above the ground;

• The model elevation, roughness, and land use data;

• Annual and monthly wind and power rose (16 sectors) and speed frequency distribution (1 m/s bins) for each model grid point;

• Surface inclination in the direction of maximum slope. For this deliverable 3TIER

calculated the value of the slope at the 1-arcminute (approximately 2 km)

resolution of the Bolivia dataset using the NASA SRTM topography with a

resolution of 90 m. First, 3TIER defined elevation values for each point on the

2km grid through bilinear interpolation from the four SRTM data points that

surrounded it; 3TIER then calculated the slope between each 2 km grid point and

the 8 points surrounding it. The result was a fraction, such that a value of 0.2

indicates a 20% slope (400 m change in elevation averaged over the 2 km

horizontal distance). The data set delivered to IFC contains for each 2 km grid

point the result of the 8 slope calculations that had the greatest absolute value6; and

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!5 3TIER obtained this information from the National Center for Atmospheric Research (NCAR) Data

Support Section at http://dss.ucar.edu/datasets/ds464.0/

6 A spatial representation of this data set on a map of Bolivia would alert wind atlas users to complex

terrain and probable local turbulence in those areas with extreme slopes (>20%). Both TDE and UCB can provide the data files necessary to create such a representation.


• Annual average energy output calculations based on a GE 1.5 SLE turbine at 80-meter hub height7.

!468%:%'%9-66*3-(4,-.(

3TIER facilitated workshops in Cochabamba and La Paz from March 9-10, 2009 in order to present the results of this project and made the Bolivia Wind Atlas available at http://firstlook.3tier.com in advance of those activities to facilitate interaction with the information prior to the workshops. Final dissemination of the Bolivia Wind Atlas will permit FirstLook users to interactively

explore the wind resource of Bolivia by taking the following steps:

1) Register as a user of FirstLook® by visiting http://firstlook.3tier.com by following

the instructions available at the FAQ link under the heading “Account.” The

registration process begins upon a user!s first visit to FirstLook when he or she

clicks either of two hyperlinks at the wind tab: (i) “Please login to display free

wind speed data” in the center of the map pane or (ii) “Login or register” at the

bottom of the map window.8

2) Browse the clickable map available online at FirstLook

3) Review additional details for each location of interest such as wind power

density, monthly and average wind speed, and power capacity factor from the

expanded popup window accessible by clicking the words “Click Here” at the

bottom right-hand corner of the standard popup window

4) Obtain free Standard Reports using coupons available from TDE 5) Obtain Professional Reports (which contain a time series for a specific location)

for 50% off using coupons available from TDE.

Pursuant to the contract between IFC and 3TIER, coupons will be made available

through TDE to qualified Bolivia Wind Atlas users during the first 12 months following

the publication of this Final Report. Interested parties should visit the “Atlas Eólico”

hyperlink at www.tde.com.bo to request FirstLook coupons for Bolivia.

Registration for FirstLook is a one-time free process. Table 2 compares FirstLook

Standard and Professional Reports for the Bolivia Wind Atlas:

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!7 Rather than using power curves available from GE for each air density, 3TIER used a power curve for a

standard atmosphere at 15 degrees Celsius and 10% factor for turbulence intensity, which in turn was multiplied by the result of the change in air density. Therefore, the energy production in this data set may tend toward an upward bias for high air density and a downward bias for low air density.

8 3TIER will use information provided during the registration process in accordance with the Privacy

Policy published at the FirstLook website.


Table 2: Comparison of FirstLook Report Features

FEATURE Standard Report

Professional Report

Map X X Wind Speed and Power Capacity by Month X X Hourly Wind Speed and Power Distribution X X Annual Average Wind and Power Directions (Wind and Power Roses – Annual)

X X

Monthly Average Wind and Power Directions (Wind and Power Roses – Monthly)

X

Annual Average Diurnal Wind and Power Variation X X Monthly Average Diurnal Wind and Power Variation X Hourly Mean Wind and Power Table by year X X Hourly Mean Wind and Power Table by Month X Data File in CSV format X Price Without Coupon (USD) $1000 $2500 Price With Coupon (USD) $0 $1250

!468%:&%'%=-(4;%/*+.0,%4(<%>-(<%",;46!

This report was prepared for this task by compiling the results of analysis completed during the project.

!468%:&&1%!04-(-(7%>.086?.+%4(<%=-(4;%50*6*(,4,-.(!

3TIER facilitated training workshops in Cochabamba and La Paz to present the Final Report during the week of March 9-10. As stipulated in the Terms of Reference, the training workshops were conducted in Spanish and addressed:

• How the wind atlas was prepared • Strengths and weaknesses of the approach • Explanation of the validation results • Explanation of the internet-based viewer (FirstLook) • How to interpret the results of FirstLook reports • Recommendations regarding suitable next steps for in-depth assessments of

wind potential at specific locations.


III. 7'/+5#%'%()1%819%'/:/5

This section addresses Bolivia’s climatology and how it relates to the Bolivia Wind Atlas. The American Meteorological Society’s second edition Glossary of Meteorology defines climatology as, “The description and scientific study of climate.” A fundamental distinction is the time scale associated with the study of weather (short-term) and climate (medium-long-term). The same source notes under its definition of weather that, “As distinguished from climate, weather consists of the short-term (minutes to days) variations in the atmosphere.”9

According to C. David Whiteman in the textbook, Mountain Meteorology, the four determining factors behind climate are latitude, altitude or elevation, ‘continentality’ (distance from the sea), and exposure to regional circulations.10 The defining characteristic of Bolivia’s climate that affects the wind resource is the complexity and variability of the country’s terrain. Bolivia’s 1,098,580 km! of total surface area varies greatly in elevation from a low point of approximately 90 meters above sea level near Bolivia’s border with Paraguay to a high point of 6,542 meters above sea level (Nevado Sajama – 18.11° S, 68.88° W). Depending on the altitude, the climate and wind resources also vary greatly. The WRF model that 3TIER employed addresses this complex topography through use of a sigma terrain following vertical coordinate system. In complex terrain the sigma coordinate system allows a high vertical resolution just above ground level, whatever altitude the ground may be. Additionally the “nested grid” methodology described earlier in this report, allows the WRF to identify the impacts of meteorological factors outside of Bolivia (regional circulations) on the wind resource within the country.

Ismael Montes de Oca’s Spanish-language article entitled “Geografía y Clima de Bolivia” notes, “The country presents a great variety of climates that depend mainly on the latitude, the altitude, and the proximity of high mountains or flat zones and mostly the circulation of the trade winds. [ . . .] The temperatures are related to three latitudinal climatic regions: tropical, subtropical, and temperate.”11 Montes de Oca also details the different climate conditions in each of Bolivia’s seven physiographical units. Those units include the so-called Altiplano highlands between 3500-4000 meters above sea level, which are located in Bolivia’s Southwestern corner to the lowland tropical flatlands, all below 500 meters above sea level, which bisect Bolivia from Cobija in its Northwest Corner, pass through Trinidad at the center of the country, and also include Santa Cruz.

Mesoscale models like WRF are highly skilled at identifying the spatial distribution of wind resources, but are still just models based on fundamental assumptions. Therefore,

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!9 This glossary is available at the following URL: http://amsglossary.allenpress.com/glossary

10 C. David Whiteman, Mountain Meteorology Fundamentals and Applications (New York, NY: Oxford

University Press, 2000) 3.

11 Article in PDF format downloaded on 10/21/2008 from

http://www.ifeanet.org/publicaciones/anfitrionoai.php?art=749.


they should not be the sole basis for identifying annual average wind speeds. As noted elsewhere in this report, collecting on-site measurements with a properly calibrated device is a necessary next step at any location that appears to have a promising wind resource to seek greater precision about specific wind speeds. 3TIER paid particular attention to conditions in mountainous areas in developing the Bolivia Wind Atlas. This Atlas has spatial resolution of 2 kilometers, which means that 3TIER’s simulation generated values for every point on a 2 km x 2 km horizontal grid. 3TIER’s FirstLook tool displays a range of annual average wind speed values at its clickable online map rather than a single value to address the issue of uncertainty.

One force that drives Bolivia’s climate and shapes its wind resource is the strong solar radiation that reaches the surface in several parts of the country, particularly its southwest corner. The attached image from 3TIER’s online map of global horizontal irradiance marks the annual average values over the noted region, which begins north of Lake Titicaca at Bolivia’s border with Peru, extends to the southeast until approximately 100 km southwest of Santa Cruz, and then southward through Tarija in Bolivia’s wine-producing region.

!

Figure 2: Global Horizontal Irradiance over Bolivia as visualized at FirstLook Through its validation of simulated wind resource values for Bolivia against observed data, 3TIER identified several simulation parameters used for test simulations that had to be configured differently to properly capture the impact of solar radiation on the intensity and variability of Bolivia’s wind resource. The end result of simulation with the


revised initialization parameters was a greater consistency in the seasonal cycle of the wind resource when compared to observed data.


IV. ;",4'#,

"@ :4;-<4,-.(%

64++5.)1

3TIER compared simulated wind speeds to three sets of wind speed observations collected at thirty-five locations within the model domain in Bolivia, Brazil, and Chile:

1) Wind speeds at 20 meters, measured at 10 tall tower locations between late January and early December of the year 2000 as part of a study commissioned by the Japan International Cooperation Agency (JICA).12 Tower locations are marked at the left of Figure 3.

2) Wind speeds at 20 meters measured between October 2003 and December 2004 at 3 additional towers in northern Chile that are within the model domain for Bolivia.13 Colored dots show these tower locations at the left of Figure 3.

3) Wind speeds at 10 meters, measured at 22 meteorological stations from the National Centers for Environmental Prediction Automated Data Processing (NCEP ADP) data set. These stations are located primarily in Bolivia, but also include some in Brazil. The right of Figure 3 shows the locations of these meteorological stations and Table 2 contains their names and geo-referencing.

3TIER also reviewed the consistency between simulated wind speed data and a fourth set of wind speed observations collected near Santa Cruz, Bolivia between 1993 and 1995 for the Rural Electrical Cooperative (Cooperativa Rural de Electricidad – CRE in Spanish); however, those observations are not featured here because:

(i) A full time series was not available,

(ii) Coordinates for the anemometers used to collect wind speed data were unavailable, and

(iii) Data for the CRE study was collected outside of the 10-year simulation period (1998-2007).

These factors precluded 3TIER from making an “apples to apples” comparison between the simulated data and the CRE observations. Nonetheless, differences between the

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!12

The full study, entitled “The study on rural electrification implementation plan by renewable energy in the Republic of Bolivia” is available though JICA’s online library at the following URL http://lvzopac.jica.go.jp/library/indexeng.html through a “Catalog Search” for the key words “Bolivia Renewable Energy”. 3TIER obtained the tower data from IFC.

13 3TIER obtained this data from the website of a Global Environment Facility (GEF) -funded United

Nations Development Program (UNDP) and Chilean Government study of renewable energy resources in Chile. (See http://www.renovables-rural.cl/actividades/fr_actividades.html).


simulated data and the observed wind speeds from the CRE study appear consistent with those between the simulated data and the above-noted 35 locations. !

<5'/&5#/%316#5#/,#/0,1

Figure 3 presents color-coded maps of the simulated wind speeds minus observed annual mean wind speed for the observed data sets. Positive values indicate that the simulated wind speeds exceeded observed wind speeds and negative values indicate that the simulated wind speeds were less than observed wind speeds. Because the tall towers (left) are located exclusively in western Bolivia and neighboring regions of Chile, 3TIER chose the meteorological stations used for this analysis for their complementary spatial distribution. The meteorological stations (right) cover the rest of Bolivia and neighboring regions of Brazil that are within the rectangular simulation domain. 3TIER’s validation process only compared observed data with simulated data at moments when values were available from both data sets. The scarcity of observed data from tall towers during much of the simulation period may have contributed to an increase in error statistics at tall tower locations (below left).

Figure 3: Maps of simulated wind speeds minus observed annual mean wind speed differences at observation locations for (left) tall tower observations at 20 meters and (right) meteorological station observations at 10 meters. Units are meters per second (m/s) and each color represents the range of bias shown in the legend at the bottom of the figure. For example, yellow dots represent validation sites where 3TIER’s simulated wind speeds were between 0.5 m/s below and 0.5 m/s above observed wind speeds.


=5>'",1%81<5'/&5#/%316#5#/,#/0,1?@7AB1*CB16#5#/%3,D1

3TIER calculated validation statistics for the 22 NCEP ADP meteorological stations, displayed below in Table 3.

Bolivia Wind Atlas 17

Table 3: Summary Validation Statistics by Observation Station

Station ID

Station Name Lat.

(Obs.) Long. (Obs.)

Lat. (Model)

Long. (Model)

Start date End date

Model overlap

with Obs.

(hours)

Obs. mean wind

speed (m/s)

Model mean wind

speed (m/s)

% Error

(Model- Obs.) / Obs.

SBCR Corumbá, Brazil -19.00 -57.65 -19.0083 -57.6417 1998/03/31 2007/02/05 2950 2.99 2.76 -7.6

SBRB Rio Branco, Brazil -10.00 -67.80 -10.0083 -67.8083 1998/03/31 2007/02/26 4737 2.21 1.69 -23.5

SBVH Vilhena Aeroporto, Brazil

-12.73 -60.13 -12.7250 -60.1250 1998/03/31 2006/04/15 1516 3.08 2.48 -19.5

SLCB Cochabamba, Bolivia -17.45 -66.10 -17.4583 -66.0917 1998/03/31 2007/02/26 2888 3.22 3.00 -7.0

SLCO Cobija, Bolivia -11.08 -68.87 -11.0750 -68.8750 1998/08/16 2007/02/26 1942 2.85 2.23 -21.7

SLCP Concepcion, Bolivia -16.25 -62.10 -16.2583 -62.0917 1998/08/16 2007/02/26 1608 4.08 4.10 +0.7

SLET Santa Cruz/El Trompillo, Bolivia

-17.80 -63.17 -17.8083 -63.1750 1998/08/16 2007/02/26 2807 6.03 5.87 -2.6

SLJE San Jose De Chiquitos, Bolivia

-17.83 -60.75 -17.8250 -60.7417 1998/08/16 2007/02/26 1490 3.46 4.18 +20.8

SLJO San Joaquin, Bolivia -13.07 -64.67 -13.0750 -64.6750 1998/08/16 2007/02/10 1574 3.32 2.42 -27.2

SLLP La Paz/Alto, Bolivia -16.52 -68.18 -16.5250 -68.1750 1998/03/31 2007/02/26 7178 3.34 3.75 +12.3

SLOR Oruro, Bolivia -18.05 -67.07 -18.0583 -67.0750 1998/08/16 2007/02/26 1833 4.03 5.19 +28.8

SLPO Potosi, Bolivia -19.53 -65.72 -19.5250 -65.7250 1998/08/26 2007/02/26 1365 4.60 4.79 +4.0

SLPS Puerto Suarez, Bolivia -19.00 -57.73 -19.0083 -57.7250 1998/08/16 2007/02/26 2354 3.55 3.07 -13.6

SLRB Robore, Bolivia -18.32 -59.75 -18.3250 -59.7417 1998/08/16 2007/02/26 1609 4.29 4.01 -6.4

SLRI Riberalta, Bolivia -11.02 -66.12 -11.0250 -66.1250 1998/08/16 2007/02/26 1515 2.95 2.03 -31.1

SLRY Reyes, Bolivia -14.30 -67.37 -14.3083 -67.3750 1998/08/16 2007/02/26 1874 3.44 2.97 -13.8

SLSA Santa Ana, Bolivia -13.72 -65.58 -13.7250 -65.5750 1998/08/16 2007/02/26 2867 4.22 2.53 -40.1

SLSI San Ignacio De Velasco, Bolivia

-16.37 -60.95 -16.3750 -60.9417 1998/08/16 2007/02/26 2182 3.71 4.12 +11.2

SLSU Sucre, Bolivia -19.02 -65.27 -19.0250 -65.2750 1998/08/16 2007/02/26 1929 3.11 4.30 +38.5

SLTJ Tarija, Bolivia -21.53 -64.72 -21.5250 -64.7250 1998/03/31 2007/02/26 1965 4.29 4.26 -0.6

SLVR Viru-Viru, Bolivia -17.65 -63.13 -17.6583 -63.1250 1998/03/31 2007/02/26 6441 5.30 5.48 +3.4

SLYA Yacuiba, Bolivia -22.02 -63.70 -22.0250 -63.6917 1998/08/16 2007/02/26 1966 4.00 3.77 -5.8


Data Source: 3TIER obtained this observed data from the National Center for Environmental Prediction (NCEP) Automated Data Processing (ADP) Global Surface Observations, provided by the National Center for Atmospheric Research (NCAR) Data Support Section at http://dss.ucar.edu/datasets/ds464.0/ Measurement Height: The measurement height is assumed to be the standard meteorological station wind speed height

of 10 meters.

Table 3 Abbreviations: In the table above marked Validation Statistics by Observation Station,

• “Lat (Obs.)” are the latitudes and “Lon (Obs.)” are the longitudes for the NCEP ADP data stations. The NCEP ADP data set only provided these to the nearest 0.01 degree.

• “Lat. (Model)” is the latitude and “Lon. (Model)” is the longitude of the nearest data set grid point. • “Model overlap with Obs (hours)” is the total number of hours of overlapping data between the observed and

model-based data, which were the hours used for validation. Most stations have many missing hours of data. • “Obs. mean wind speed” is the observed mean wind speed in m/s, averaged over all hours of overlapping data. • “Model mean wind speed” is the model-based mean wind speed in m/s, averaged over all hours of overlapping

data • “%Error” is the percentage error of the model mean wind speed, calculated as (Model mean wind speed – Obs.

mean wind speed) / Obs. Mean Wind Speed.

Table 4: Summary Bias & Error Statistics (NCEP ADP Data)

Mean wind speed bias -0.14 m/s -4.58%

Mean absolute error 0.53 m/s 15.46 %

Root Mean Squared Error 0.67 m/s 19.46 %

In the summary table above, the mean wind speed bias, mean absolute error, and RMS error are provided in both m/s

and percentage errors, where the percentage errors are calculated as (Model mean wind speed – Obs. mean wind speed)

/ Obs. Mean Wind Speed.


!"#$%&'"()*+"&,*

Below are histograms of the annual mean differences between simulated wind speeds and observed wind speeds for each data set. The magnitude of the differences is consistent with those that 3TIER has encountered for simulated data sets in other parts of the world. The differences relative to meteorological stations are smaller than typically found for 5km resolution simulation data sets. While the difference between simulated data and observations collected at tall towers is larger than 3TIER has found elsewhere, the towers were located in relatively rough terrain, which tends to increase

the errors found in the simulated data set. Also, the number of samples at individual

data points varied. At the tall towers (below left), the number of hourly samples used in

the annual mean ranged from 436 to 793 and averaged 627. At the NCEP ADP stations

(below right), the number of hourly samples used in the annual mean ranged from 1490

to 7178 and averaged 2483. The minimum number of hourly samples per month for

was 42 for the tall towers and 72 for the NCEP ADP stations. 3TIER’s analysis indicates that the simulated wind speeds have a systematic low bias near Bolivia’s border with Chile and in northern Bolivia. In other parts of Bolivia, simulated wind speeds have a small or slightly high positive bias. Overall, based on the slightly low average bias of simulated wind speeds relative to observed wind speeds 3TIER regards simulated Bolivia wind speeds as somewhat conservative.

!

Figure 4: Histograms of simulated wind speeds minus observed annual mean wind speed for: (left) simulated wind speeds minus tall tower observations at 20 meters, and (right) simulated wind speeds minus meteorological station observations at 10 meters. Units for bias are meters per second (m/s). The colors of the vertical bars are the same as those used in the spatial representation of bias according to validation site in Figure 3. For example, yellow bars represent validation sites where 3TIER’s simulated wind speeds were between 0.5 m/s below and 0.5 m/s above observed wind speeds.


!"#$%&'()(*+,-"#,&(.'/&+(

3TIER compared the seasonal cycles of wind speed between simulated data and observed data, averaged across each data set. The seasonal cycle for the simulated data compares favorably with the seasonal cycle for both sets of observations even though the small number of overlapping observations from tall towers results in a noisy seasonal cycle for direct comparison with the towers.

!

Figure 5: Seasonal cycle of monthly mean wind speed for simulated data and observations, for: (left) 20 meter wind speed, averaged over 13 tall tower locations, and (right) 10 meter wind speed, averaged over 22 meteorological station locations. Units are meters per second (m/s).

0123#,&(.'/&+(

3TIER compared the diurnal cycles of wind speed for simulated data with the diurnal cycles of wind speed for observed data, as averaged across each data set. The comparison with meteorological stations indicates that the timing of the diurnal cycle is good at these locations, although the amplitude of the diurnal cycle for the simulated data diverges from that of the observed data. The comparison with tall towers indicates that the maximum and minimum in the simulated diurnal cycle tend to come a few hours too late at these locations. These results are generally consistent with the validation of simulated data sets for other projects. While the seasonal cycle tends to be a close approximation to observed data, the diurnal cycle tends to be a reasonable approximation, but not as close to observed data as the seasonal cycle. Although the diurnal cycle was quite different at the tall towers (below left) than at the meteorological stations (below right), the simulation captured the differences in relative magnitude between the two, albeit with the time lags noted above.


!

Figure 6: Diurnal cycle of hourly mean wind speed for simulated data and observations, for: (left) 20 meter wind speed, averaged over 13 tall tower locations, and (right) 10 meter wind speed, averaged over 22 meteorological station locations. Units are meters per second (m/s).

!"#$%"&$'()*'(+#,-$'()

Validation of simulated data against observed data from 35 locations throughout the rectangular Bolivia model domain yielded findings similar to those obtained in other regions where 3TIER has performed validation analysis on its simulations. Overall, the magnitude of simulated wind speeds tended to be slightly lower than that of observed wind speeds. For seasonal and diurnal cycles, the comparison between the simulated data and observed data was also consistent with 3TIER’s validations of other simulations. With respect to direction, 3TIER did a wind rose analysis that compares simulated and observed wind direction at the 22 NCEP ADP stations on an annual and monthly basis. At IFC’s request, 3TIER has included that analysis in Appendix III at the end of this report.

Future comparisons between simulated data and other data collected by anemometers installed at tall towers within the Bolivia model domain probably will confirm the slightly low bias that 3TIER identified during validation. Grid point values for each corner of each 2 km x 2 km box represent spatial averages. Wind speed observations collected at locations other than grid points themselves will, on average, tend to exceed values for simulated data, especially at data collection locations selected because they are particularly windy. For example, an anemometer sited at the top of a ridgeline between two points on the simulation grid probably will record higher wind speeds than the simulated values at adjacent grid points.


!" #$%&'

Online, clickable, interactive maps of annual average wind speed at 20, 50, and 80 meters are available for free at http://firstlook.3tier.com. IFC has also funded new FirstLook functionality that permits users to access additional data useful for preliminary site analysis at specific locations using an expanded popup window. Below is a screen shot of the wind resource over Bolivia at 80 meters above ground level:

!

Figure 7: Bolivia's wind resource at 80 meters as shown at FirstLook

To access the expanded popup window mentioned above, FirstLook visitors should click the words “Click Here” at the bottom right-hand corner of the popup window that appears above the arrow pointing to a site selected in Bolivia. This expanded popup is only available for locations within Bolivia and IFC has arranged for 3TIER to make it available for 3 years after this Final Report has been published. Appendix II contains the following 23 maps:

(1) Annual Mean Wind Speed at 20m



(4) Values for Weibull A parameter at 20m


(5) Values for Weibull k parameter at 20m





(10) Annual Capacity Factor at 80 m (GE 1.5 sle)14

(11) Annual Mean Wind Power Density at 80 m

(12) January Mean Wind Power Density at 80 m

(13) February Mean Wind Power Density at 80 m

(14) March Mean Wind Power Density at 80 m

(15) April Mean Wind Power Density at 80 m

(16) May Mean Wind Power Density at 80 m

(17) June Mean Wind Power Density at 80 m

(18) July Mean Wind Power Density at 80 m

(19) August Mean Wind Power Density at 80 m

(20) September Mean Wind Power Density at 80 m

(21) October Mean Wind Power Density at 80 m

(22) November Mean Wind Power Density at 80 m

(23) December Mean Wind Power Density at 80 m

The data for the maps has a pixel resolution of 0.016666 degrees; however, because

the earth!s curved surface is being displayed on a flat plane for the purpose of these

maps, the linear measurement of each pixel changes according to latitude. In the range

of latitudes covered by the Bolivia Wind Atlas (between 9.7 degrees south and 22.9

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!14

Annual Capacity Factor is labeled Annual Power Capacity in the upper left of the expanded popup

window at FirstLook. Expressed as a percentage, this variable is a ratio of actual (or in this case simulated) energy generation divided by theoretical generation during an entire year. For this example, a GE 1.5 sle turbine operating at full capacity year round could generate 13,140 megawatt hours of energy per year (1.5 megawatts x 24 hours x 365 days); however, wind turbines do not always operate at full capacity due to the variability of the wind resource. An annual capacity factor of 35% at a given location means that averaged over a year, a GE 1.5 sle turbine could be expected to generate 4,599 megawatt hours of energy. Rather than using power curves available from turbine manufacturer GE for each air density, 3TIER used a power curve for a standard atmosphere at 15 degrees Celsius and 10% factor for turbulence intensity, which in turn was multiplied by the result of the change in air density. Therefore, the capacity factor in map 10 will tend to have an upward bias for high air density and a downward bias for low air density.


degrees south), 0.016666 degrees is equivalent to east-west distances between

approximately 1.8270 km and 1.7088 km and north-south distances between

approximately 1.8415 km and 2.2135 km.15

At IFC!s request, 3TIER used nearest-neighbor interpolation to color the printed maps

and their downloadable counterparts, the latter available from TDE. Through this

method, 3TIER did not smooth color transitions between grid points. Abrupt color

changes identified by zooming in on a specific location should alert Bolivia Wind Atlas

users to some condition (possibly a pronounced topographical feature) that might cause

such changes. As noted in their respective legends, all maps include Bolivia’s transmission network, protected areas, and roads as well as international boundaries, city names, and the names of and boundaries between Bolivia’s 9 Departments (administrative divisions within the country). TDE assisted with the provision of this important information to 3TIER for the Bolivia Wind Atlas. The digital version of this report has been formatted to facilitate download, but printed versions of this report include each of the maps in 11” x 17” format and digital maps containing each of the 23 above-noted data layers are also available for download from TDE in two file formats: PNG and GeoPDF. Appendix II describes how to use the maps in GeoPDF format, which requires Version 9.1 or later of the free Adobe Reader

software, available for download at http://get.adobe.com/reader/.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!15

A basic tool to calculate the linear distance between two points for which one has the latitude and

longitude is available online at the website of the Northern California Earthquake Data Center, which has the following URL: http://www.ncedc.org/convert/distance.html.


!" #$%&'&()*+,&-.+/$012+/$31-3&(%+

!" #$%&'()*+,-.)'/0*$.*'

No single formula defines “the best wind resource” because many variables affect the potential success of a wind energy project. The amount of energy that can be generated by a wind turbine varies with the wind speed, the air density, the diameter of the wind turbine rotor, and the efficiencies of the turbine and electrical system. Wind energy developers should select the largest diameter rotor and most efficient turbine and electrical system that makes economic sense for their target wind energy application. Wind speed remains the single biggest factor in determining the success of your project and there is little one can do to affect it other than to select the windiest location to install a turbine. Because the power varies as the cube of the wind speed, a 25% increase in the wind speed at a given moment results in a potential doubling of power output from a wind turbine at that same moment. When considering mean wind speeds, the relationship between wind speed and power output is more complex and depends on wind speed distribution during the period for which the mean was calculated.

Unfortunately, high wind speeds alone are not enough. Timing is also crucial to determine the suitability of the wind resource in a given location. High wind speeds during the winter provide little economic benefit if peak demand is for cooling needs during the summer. Similarly, an off grid residential system will require more battery storage if the highest wind speeds occur during the night rather than during the day. The best wind resource is the one that provides high wind speeds at times when the generated electricity has the highest value.

Wind turbines function best within the range of wind speeds for which their manufacturer designed them, which varies according to the size and desired use of the equipment. At the bottom of that range is the cut-in speed, at which a turbine begins to generate power. The turbine’s cut-out speed, when it will stop generating power, is at the top of its operational range. For example, the GE 1.5 sle turbine, which 3TIER uses at the 80 m hub height in FirstLook, cuts in at 3.5 m/s and cuts out at 25 m/s. Within the range is each turbine’s “rated wind speed,” the lowest wind speed at which it will generate its rated power (maximum generating capacity). At its rated wind speed of 14 m/s, the GE 1.5 sle turbine generates its rated power (1.5 MW).

While studying the data collected by an anemometer will provide the greatest precision about the type of timing information mentioned above, a FirstLook report can provide a preliminary approximation designed to facilitate the decision of whether or not installing an anemometer would be advisable. IFC has made arrangements for the issuance of discount coupons to qualified Bolivia Wind Atlas users from the public sector and NGOs

for online purchase of a FirstLook report. Interested parties should visit the “Atlas Eólico” hyperlink at www.tde.com.bo to request FirstLook coupons for Bolivia.


!" #$%&'%()*'+&,'-.&'/0)/1)2'03)45+/.,65)

3TIER reviewed the Annual Average Wind Speed maps at the three hub heights of 20, 50 and 80 meters using the color bar in the legend that extends between less than three and greater than nine m/s. On the color bar, cool colors (purple, blue, moving to green) represent low wind speeds while warm colors (yellow, orange, moving to red) represent progressively higher wind speeds. While searching for orange and red sections of the maps at the three hub heights, 3TIER noted that the most robust wind resource in the Bolivia appears to be concentrated in four sectors:

1) Around Santa Cruz de la Sierra, largely south and west of the city center;

2) At Bolivia’s southwestern border with Chile and Argentina in the Department of Potosi;

3) In a roughly east-west “corridor” between the cities of Santa Cruz and La Paz that runs south of the 230 kV transmission line between Santa Cruz and Cochabamba and slightly north of it between Cochabamba and La Paz; and

4) In a roughly north-south “corridor” between the area just east of the town of Oruro and west of the town of Potosi

The wind resource is limited in northern Bolivia. Other than the protected areas in central La Paz Department, there appears to be very little wind resource north of the town of Trinidad (14.84° S, 60.93° W). Another area with an apparently robust wind resource is the area adjacent to the famous lake Titicaca northwest of the city of La Paz. For whatever reason, the extent to which areas with a strong wind resource are “off limits” for development will be a very important factor in assessing individual sites. The Bolivia Wind Atlas identifies protected areas in or near the first three sectors identified above. For example, in Sector 2, the Eduardo Avaroa National Reserve of Andean Fauna masks the southwestern corner of Bolivia. Sector 3 includes 2 national parks (Carrasco and Amboro). Close coordination with the governmental authorities responsible for maintaining these protected areas will be an important element of the due diligence process for anyone wishing to harness wind energy within those areas. As implied above, when and how the wind blows is often more relevant than annual average wind speeds when selecting project sites for further analysis. The wind speed frequency distribution (also known as the Weibull distribution) is a probability density function with an asymmetric bell shape that shows the relationship between each wind speed (x-axis) and the percentage of time that it occurs at a given location (y-axis). The Weibull distribution has two parameters that engineers and developers often study: A (scale) and k (shape).16

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!16

Appendix 1 describes how these parameters were calculated for this project. Because 3TIER derived

the Weibull parameters presented in this report from a simulation, they will only approximate those encountered in the real world. Only long term physical measurement with an anemometer or other


!" #$%&'()*+,'-+%.$/0'

The results of 3TIER!s Wind Power Density calculations offer an additional filter for

project sites beyond a review of average annual wind speeds. Wind Power Density

quantifies how many watts of power are available for each square meter of area within

the radius of the blades of a wind turbine (otherwise known as the swept area). It takes

into account wind speed, air density, and wind frequency distribution. 3TIER used the following formula to calculate Wind Power Density (WPD):

!

WPD =1

2n(

i=1

n

" #)(vi

3) in which

n = the number of records considered for the calculation

!"#the air density (kg/m3)

!

vi

3

"#the cube of the ith wind speed (m/s)

The unit of measure for Wind Power Density is Watts / square meter (W/m2). The FAQ section of the American Wind Energy Association homepage summarizes the relevance of wind power density at: http://www.awea.org/faq/basicwr.html Upon reviewing maps of annual mean wind power density in Bolivia, 3TIER noted a reduction of the extent of sectors 1-4 versus that evident on annual mean wind speed maps. As evident from the above equation, wind power density varies directly in proportion to air density. Therefore, at locations with low air densities, low wind power density should be expected. Many of the places in sectors 2-4 described above are located at altitudes high above sea level, where thinner (less dense) air exerts less force on wind turbine blades at any given wind speed than it would at that same speed if the same turbine were located at sea level. An additional way to check elevation and wind power density at any location in Bolivia is by using the expanded popup window functionality secured by IFC for display at the FirstLook site. As well as obviating the need for visual interpolation by permitting the selection of an exact location, that method permits checking the relative wind power density at multiple hub heights (20, 50, and 80 meters).

###################################################################################################################################################################################properly calibrated device can ensure precise calculation of wind speed frequency distribution and the associated probability density function and Weibull parameters.


!"# $%&'()*+,+-'.+/0'12*-3'+/'4)/2&52'

!" #$%&$'%()

!

The Bolivia Wind Atlas is the first in a series of steps necessary to develop a wind energy project successfully. A detailed guide to wind energy project development exceeds the scope of this project, but the following graphic suggests a logical progression of steps toward the successful operation of a wind energy facility:

The graphic above denotes with traffic light clarity the future activities necessary for successful project development. The Bolivia Wind Atlas is a suitable tool for the prospecting phase of wind energy project development and thus that block is green. Using this report and the resources online at FirstLook, some of the initial resource analysis can be completed and thus that block is yellow. All further blocks are red because the Bolivia Wind Atlas alone is not sufficient to make the decisions necessary complete them. The results or 3TIER’s analysis are based on model simulations and observations. Model simulations were made using a Numerical Weather Prediction (NWP) model similar to the one used by government meteorological services to produce local weather


forecasts. The NWP model simulates atmospheric conditions and accounts for the effects of land use and terrain. 3TIER made model simulations for Bolivia at a spatial resolution of 2km. Values obtained for any location between points on the 2km x 2km model grid are interpolated. Unfortunately, models only provide an approximation of the real world. Publicly available observations from meteorological towers and other wind speed measurements were used to validate the simulation results and can be used estimate the level of uncertainty associated with the wind speeds displayed at FirstLook. The power produced at a particular wind speed depends on the wind turbine installed at a location. Different wind turbines have different rated wind speeds. The rated speed of a wind turbine is the speed at which that turbine produces its rated power. If two wind turbines are both rated for the same power, but one has a much lower rated speed than the other, then the turbine with the lower rated wind speed will typically produce more power when averaged over a long time. The function relating power to wind speed is called the power curve and is different for each turbine. To generate FirstLook reports, 3TIER uses one of three standard turbines depending on hub height as follows:

• 80 m: Model GE 1.5 SLE • 50 m: Model Vestas V52 • 20 m: Model Bergey XL.1

Users of the Bolivia Wind Atlas should not make decisions about the design or construction of a wind energy facility based exclusively on the information contained in this report or available in FirstLook. The Bolivia Wind Atlas was designed to provide a realistic assessment of the wind resource at a specific location. It should orient decision-makers about whether it is worthwhile to further pursue wind power development at that location, but variances between simulated and real life wind speeds and turbine function are to be expected due in part to the averaging of terrain and land use data to model grid resolution. In practice local roughness, obstacles, array and orography effects will mean that the actual production for each turbine in real world conditions and landscape will differ from the simulated production for that turbine based on the NWP model. Users should remain mindful of the limitations of data analysis for a set of grid points between each of which there is a horizontal distance of 2 km.


Figure 8: The graphic above depicts the horizontal resolution for the NWP simulation used to create the Bolivia Wind Atlas. The WRF model calculates discrete meteorological values (wind speed, temperature, atmospheric pressure, etc.) for each corner of each cell in the 2 km x 2 km grid. At the beginning of the simulation, 3TIER paired each data point with land use, elevation, and roughness values.

Given the 2km resolution of the NWP model simulations, the wind resource at a location may differ from the estimate available at FirstLook. While the model simulations incorporated the effects of terrain on atmospheric conditions, some local terrain features (especially in mountainous parts of Bolivia) may be too small to be represented in the model. These limitations are an inherent part of all estimates of wind speed and power. Only carefully maintained, long-term observations at the future location for a wind turbine can provide an unequivocal assessment of the wind resource at that site. Because such observations are time-consuming and expensive, collecting them should be considered only for locations at which the wind resource is sufficient to justify the required investment.

!" #$%&'(%)*+,-./0/)

!

After identifying project locations from the areas deemed to be of interest for further evaluation, 3TIER recommends seeking the assistance of qualified professionals to conduct complementary studies to guide decision-making about wind power development, which has been keyed to the numbered steps in the graphic provided in Section VI B. above:

• 3. Selecting a site for data collection - Selecting precisely where to collect observational data requires a site visit. During that visit, particular attention should be paid to local orography17 and wind obstacles and the completion of a proper roughness assessment of the site.18

• 4. Collecting relevant on site data (wind speed, wind direction, temperature, and atmospheric pressure) to corroborate the model results. The measurement device (anemometer or other) selected should be: (i) properly calibrated19, (ii)

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!17

Available online at http://amsglossary.allenpress.com/glossary, the glossary of the American

Meteorological Society defines orography as “1. The nature of a region with respect to its elevated terrain. 2. That branch of geomorphology that deals with the disposition and character of hills and mountains."

18 Available online at http://amsglossary.allenpress.com/glossary, the glossary of the American

Meteorological Society defines aerodynamic roughness length—(also known as roughness length or z0) as, “The height above the displacement plane at which the mean wind becomes zero when extrapolating the logarithmic wind- speed profile downward through the surface layer.”

19 Anemometer calibration is done at enclosed facilities (wind tunnels) where each device may be

exposed to wind blowing at a known uniform velocity with minimal turbulence. Adjustments are made to the anemometer to ensure that it measures the known wind velocity precisely. Equipment


installed on a tall mast with a proper design, (iii) and mounted along with any booms in accordance with wind industry standards.

• 5. Extrapolating from on site data collected – There are various methodologies for extrapolating in space and time from observed data collected on site at meteorological towers. These include: (i) micro-scale modeling using software tools such as WAsP or WindPro (spatial extrapolation) and (ii) the procurement of regional reference wind speed data from adjacent met stations for normalization to a mean wind year (temporal extrapolation).

• 5. Analyzing the medium to long-term variability of the wind resource – Another

method for temporal extrapolation involves the further use of NWP models. Such an analysis is very important to facilitate dispatch planning because wind is the “fuel” for a wind energy project and therefore one of the primary drivers of project revenues. Since the wind resource varies from year to year as it does month to month and hour to hour, putting a one or two year set of observations into a 15-20-year context will facilitate more precise estimation of project revenues, taking into account years which have a weak, average, and strong wind resource in relative terms. While overestimation is the most problematic error, underestimation of the expected wind resource during the early years of a project can lead financial entities to design a loan payback structure that is inconsistent with a wind farm’s actual production.

• 5. Evaluating demand for the wind facility’s future production – Treatment of this

aspect will depend on the target use for wind energy at each selected site. Those installing wind turbines to support a specific activity with intense energy demand (such as a large desalination facility) will have different priorities when evaluating demand than the owner of a remote eco-lodge with no grid connection hoping to supplement generation from his mini-hydro turbine or government planners who use wind data to identify existing locations where stand-alone diesel gen-sets might be used more efficiently in tandem with wind power systems. To help potential clients determine whether using a specific turbine would be suitable for a desired application, some turbine manufacturers or their representatives offer testimonials about those who have successfully used their equipment.

• 6. Assessing potential environmental, social, and economic impacts (geological

and geotechnical characteristics, geomorphology, land use, noise studies, ecosystem (flora, & fauna), access by road, effects on or benefits to nearby communities, etc.).

• 6. Assessing natural hazards to power generating installation (seismic threats, volcanic activities, erosion).

• 6. Studying Wind Energy Integration – This type of analysis should focus on the

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

suppliers use various different calibration standards and should be able to provide details about how their equipment has been calibrated.


continued efficient operation of the entire energy transmission and distribution system. In doing so, it should consider the availability and location of transmission systems within a reasonable distance from the point of wind power generation, as well as regulatory conditions in Bolivia’s energy sector that affect energy consumers’ ability to purchase electricity from renewable sources like wind.

!" #$%%&'%(%$%)*%&+,-%$.,/0&

!

• RETScreen International (www.retscreen.net) offers multiple didactic resources, including a software suite called RETScreen designed to facilitate the analysis of renewable energy projects. The site also features an e-textbook with a chapter dedicated exclusively to wind energy project development. Many of the materials that RETScreen provides have also been translated into multiple foreign languages, including Spanish.

• The Danish Wind Industry Association website has a very useful section entitled “ know how, which contains everything from a children’s guide called “Wind With Miller” to a “guided tour” which includes a Spanish-language version that describes many wind power fundamentals. Please visit http://www.windpower.org for further information.

• The Resources section of the American Wind Energy Association (AWEA) homepage also contains lots of useful reference material. Please visit http://www.awea.org/ AWEA’s Wind Energy Siting Handbook, which contains a list of United States best practices for analyzing wind energy development, is available at http://www.awea.org/sitinghandbook/download_center.html

Bolivia Wind Atlas I-1

!""#$%&'()*(+,-./-,0&1$(12(3#&4/--(",5,6#0#57(

!

At IFC’s request, 3TIER calculated the Weibull parameters for the Bolivia Wind Atlas according to the standards defined by National Laboratory for Sustainable Energy at the Technical University of Denmark – DTU (RISØ) for the elaboration of the European

Wind Atlas, and in an appropriate manner for mesoscale numerical weather prediction

(NWP) simulations. The Weibull distribution, which displays in graphical format the probability that the wind will blow at each given speed, is described by Equation #1:

!

f (v) =k

A

"

# $

%

& ' v

A

"

# $

%

& ' k(1

exp (v

A

"

# $

%

& ' k)

* +

,

- . EQUATION #1

The variables used in this formula and presented in the Bolivia Wind Atlas are:

• A: the scale factor with units of wind speed. (Note: Some equations for the Weibull distribution refer to this parameter as c.)

• k: a dimensionless shape factor. At increased k values, the distribution becomes narrower and taller. When k=2, the so-called “Rayleigh distribution” occurs. For wind energy applications, the value of the k parameter has a key significance. As noted by the Danish Wind Industry Association on their website, many turbine manufacturers publish performance statistics for the equipment that they produce according to a Rayleigh distribution.

The variable v refers to wind speed (velocity) and is measured in meters per second (m/s). Equation #2 defines the value of A according to mean wind speed (V) and the shape factor (k).

!

A =V

" 1+1

k

#

$ %

&

' (

EQUATION #2

Equation #3 defines the wind power density E for a Weibull distribution with mean speed V and shape factor (k).

!

E =1

2"V 3

#(1+3

k)

#3(1+1

k)

$

%

& & &

'

(

) ) )

=1

2"A3#(1+

3

k)

EQUATION #3

Bolivia Wind Atlas I-2

The specifications for calculating the Weibull shape factor from the European Wind Atlas (Risø, 1989 – see http://www.windatlas.dk/Europe/About.html for more information) call for the use of an iterative process to identify the value of k for each Weibull distribution that:

1. Best fits with the frequency at which wind speeds greater than the average speed occur, and

2. Maintains the total amount of wind energy E identified through the observed wind speed probability distribution.

The European Wind Atlas methodology does not include wind speeds that occur less than 1% of the time (i.e. the highest wind speed values at the far tail of the distribution) when calculating Weibull parameters. This is a reasonable practice when using observational data as extreme wind events may strongly affect the distribution of wind speed values and the resulting Weibull parameters. However, Numerical Weather Prediction (NWP) models generally underestimate the magnitude of extreme wind events, and therefore underestimate the true, observed variability. Therefore, excluding wind speed values that occur less than 1% of the time is not necessary, and potentially detrimental, when computing Weibull parameters from a mesoscale NWP model data set.

Additional References:

Interested users of the Bolivia Wind Atlas may obtain a straightforward description of the Weibull distribution as applied to wind energy and the use of Microsoft Excel to perform Weibull analysis at the following locations: Description of the Weibull distribution at the webpage of the Danish Wind Industry Association: http://www.windpower.org/en/tour/wres/weibull.htm Dorner, W.K., “Using Microsoft Excel for Weibull Analysis”, http://www.qualitydigest.com/jan99/html/body_weibull.html

Bolivia Wind Atlas II-1

!""#$%&'())*(+,"-(

!" #$%&'%&()

!

Along with instructions for using digital maps available from TDE for download, this Appendix contains the following 23 maps created by 3TIER for this project:










(10) Annual Capacity Factor at 80 m (GE 1.5 sle)20

(11) Annual Mean Wind Power Density at 80 m

(12) January Mean Wind Power Density at 80 m

(13) February Mean Wind Power Density at 80 m

(14) March Mean Wind Power Density at 80 m

(15) April Mean Wind Power Density at 80 m

(16) May Mean Wind Power Density at 80 m

(17) June Mean Wind Power Density at 80 m

(18) July Mean Wind Power Density at 80 m

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!20

Annual Capacity Factor is labeled Annual Power Capacity in the upper left of the expanded popup

window at FirstLook. Expressed as a percentage, this variable is a ratio of actual (or in this case simulated) energy generation divided by theoretical generation during an entire year. For this example, a GE 1.5 sle turbine operating at full capacity year round could generate 13,140 megawatt hours of energy per year (1.5 megawatts x 24 hours x 365 days); however, wind turbines do not always operate at full capacity due to the variability of the wind resource. An annual capacity factor of 35% at a given location means that averaged over a year, a GE 1.5 sle turbine could be expected to generate 4,599 megawatt hours of energy. Rather than using power curves available from turbine manufacturer GE for each air density, 3TIER used a power curve for a standard atmosphere at 15 degrees Celsius and 10% factor for turbulence intensity, which in turn was multiplied by the result of the change in air density. Therefore, the capacity factor in map 10 will tend to have an upward bias for high air density and a downward bias for low air density.


(19) August Mean Wind Power Density at 80 m

(20) September Mean Wind Power Density at 80 m

(21) October Mean Wind Power Density at 80 m

(22) November Mean Wind Power Density at 80 m

(23) December Mean Wind Power Density at 80 m


!" #$%&'(')*(+,-.(

All coordinates in this report and the digital maps are based on WGS 84 datum and

displayed in decimal degree format. The data for the maps has a pixel resolution of

0.016666 degrees; however, because the earth!s curved surface is being displayed on a

flat plane for the purpose of these maps, the linear measurement of each pixel changes

according to latitude. In the range of latitudes covered by the Bolivia Wind Atlas

(between 9.7 degrees south and 22.9 degrees south), 0.016666 degrees is equivalent

to east-west distances between approximately 1.8270 km and 1.7088 km and north-

south distances between approximately 1.8415 km and 2.2135 km.21

At IFC!s request, 3TIER used nearest-neighbor interpolation to color the printed maps

and their downloadable counterparts, the latter available from TDE. Through this

method, 3TIER did not smooth color transitions between grid points. Abrupt color

changes identified by zooming in on a specific location should alert Bolivia Wind Atlas

users to some condition (possibly a pronounced topographical feature) that might cause

such changes. As noted in their respective legends, all maps include Bolivia’s transmission network, protected areas, and roads as well as international boundaries, city names, and the names of and boundaries between Bolivia’s 9 Departments (administrative divisions within the country). TDE assisted in the provision of this important information to 3TIER for the Bolivia Wind Atlas. The digital version of this report has been formatted to facilitate download, but printed versions of this report include each of the maps in 11” x 17” format. Digital maps in PNG and GeoPDF format are also available for download from TDE. Each of the maps in PNG format includes the map image of a single data layer (identified by numbers keyed to the descriptions at the beginning of this Appendix) as well as coordinate hash marks at the map borders to permit alignment of multiple layers.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!21

A basic tool to calculate the linear distance between two points for which one has the latitude and

longitude is available online at the website of the Northern California Earthquake Data Center, which has the following URL: http://www.ncedc.org/convert/distance.html.


!" #$%&'$&()*&+,-,'./&0.1)&,2&3*$456&7$80.'&

The maps in GeoPDF format permit offline interaction with variable map layers (wind speed, wind power density, capacity factor, etc.) and base layers (roads, transmission lines, city names, etc.) through the following two basic capabilities of the free Adobe Reader software (Version 9.1 or higher, available for Windows and Mac operating

systems at http://get.adobe.com/reader/):

1) Switching layers on and off – The layer tool is located in the navigation panel at the left of the Adobe Reader window. If it is not shown in their default view, users may access the layer tool through the View Menu as shown in the following Figure II-1:

!

Figure II-1: How to access the layer tool from the Adobe Reader View menu

!

Once the layer tool is open, clicking the eye-shaped icons to the left of each layer name allows users to turn layers on and off. If the eye-shaped icon is visible, the layer with which it is associated will be turned on (visible). If the eye-shaped icon is not visible, the layer will be turned off (invisible). Figure II-2 below is a screen shot of the layer tool when the menu of layers has been fully expanded.!


!

Figure II-2: Screen shot of layer tool when fully expanded

2) Identifying coordinates (latitude, longitude) of map locations – With the Geospatial Location Tool, Bolivia Wind Atlas users may see the coordinates of any location within Bolivia over which they position their mouse pointer.

!

Figure II-3: How to access the Geospatial Location Tool from the Tools menu

Once the Geospatial Location Tool has been activated, a rectangle at the lower right-hand corner of the screen will display the latitude and longitude of the current mouse pointer location. In Figure II-4 below, the pointer is southeast of the city of Santa Cruz.


!

Figure II-4: Screen shot of Geospatial Location Tool function

Bolivia Wind Atlas III-1

!""#$%&'()))*(!%%&+&,$-.(/-.&%-+&,$(01&$%(2,3#34(

3TIER’s Validation Report included a comparison of wind roses at the NCEP ADP observation stations between simulated data and observed data on a monthly and annual basis. At IFC’s request, that comparison has been included as an Appendix to this report.

III-2

SBCR (Corumba, Brazil)

Figure III-1: Annual Wind Rose Comparison: SBCR

SBRB (Rio Branco, Brazil)

Figure III-2: Annual Wind Rose Comparison: SBRB

III-3

SBVH (Vilhena Aeroporto, Brazil)

Figure III-3: Annual Wind Rose Comparison: SBVH

SLCB (Cochabamba, Bolivia)

Figure III-4: Annual Wind Rose Comparison: SLCB

III-4

SLCO (Cobija, Bolivia)

Figure III-5: Annual Wind Rose Comparison: SLCO

SLCP (Concepción, Bolivia)

Figure III-6: Annual Wind Rose Comparison: SLCP

III-5

SLET (Santa Cruz / El Trompillo, Bolivia)

Figure III-7: Annual Wind Rose Comparison: SLET

SLJE (San Jose de Chiquitos, Bolivia)

Figure III-8: Annual Wind Rose Comparison: SLJE

III-6

SLJO (San Joaquin, Bolivia)

Figure III-9: Annual Wind Rose Comparison: SLJO

SLLP (La Paz / Alto, Bolivia)

Figure III-10: Annual Average Wind Rose Comparison: SLLP

III-7

SLOR (Oruro, Bolivia)

Figure III-11: Annual Average Wind Rose Comparison: SLOR

SLPO (Potosi, Bolivia)

Figure III-12: Annual Wind Rose Comparison: SLPO

III-8

SLPS (Puerto Suarez, Bolivia)

Figure III-13: Annual Wind Rose Comparison: SLPS

SLRB (Robore, Bolivia)

Figure III-14: Annual Wind Rose Comparison: SLRB

III-9

SLRI (Riberalta, Bolivia)

Figure III-15: Annual Wind Rose Comparison: SLRI

SLRY (Reyes, Bolivia)

Figure III-16: Annual Wind Rose Comparison: SLRY

III-10

SLSA (Santa Ana, Bolivia)

Figure III-17: Annual Wind Rose Comparison: SLSA

SLSI (San Ignacio de Velasco, Bolivia)

Figure III-18: Annual Wind Rose Comparison: SLSI

III-11

SLSU (Sucre, Bolivia)

Figure III-19: Annual Wind Rose Comparison: SLSU

SLTJ (Tarija, Bolivia)

Figure III-20: Annual Wind Rose Comparison: SLTJ

III-12

SLVR (Viru-Viru, Bolivia)

Figure III-21: Annual Wind Rose Comparison: SLVR

SLYA (Yacuiba, Bolivia)

Figure III-22: Annual Wind Rose Comparison: SLYA

III-13

Figure III-23: Observed at SBCR (Corumba, Brazil) Figure III-24: Simulated at SBCR (Corumba, Brazil)

III-14

Figure III-25: Observed at SBRB (Rio Branco, Brazil) Figure III-26: Simulated at SBRB (Rio Branco, Brazil)

III-15

Figure III-27: Observed at SBVH (Vilhena Aeroporto, Brazil) Figure III-28: Simulated at SBVH (Vilhena Aeroporto, Brazil)

III-16

Figure III-29: Observed at SLCB (Cochabamba, Bolivia) Figure III-30: Simulated at SLCB (Cochabamba, Bolivia)

III-17

Figure III-31: Observed at SLCO (Cobija, Bolivia) Figure III-32: Simulated at SLCO (Cobija, Bolivia)

III-18

Figure III-33: Observed at SLCP (Concepción, Bolivia) Figure III-34: Simulated at SLCP (Concepción, Bolivia)

III-19

Figure III-35: Observed at SLET (Santa Cruz / El Trompillo, Bolivia)

Figure III-36: Simulated at SLET (Santa Cruz / El Trompillo, Bolivia)

III-20

Figure III-37: Observed at SLJE (San Jose de Chiquitos, Bolivia)

Figure III-38: Simulated at SLJE (San Jose de Chiquitos, Bolivia)

III-21

Figure III-39: Observed at SLJO (San Joaquin, Bolivia) Figure III-40: Simulated at SLJO (San Joaquin, Bolivia)

III-22

Figure III-41: Observed at SLLP (La Paz / Alto, Bolivia) Figure III-42: Simulated at SLLP (La Paz / Alto, Bolivia)

III-23

Figure III-43: Observed at SLOR (Oruro, Bolivia) Figure III-44: Simulated at SLOR (Oruro, Bolivia)

III-24

Figure III-45: Observed at SLPO (Potosi, Bolivia) Figure III-46: Simulated at SLPO (Potosi, Bolivia)

III-25

Figure III-47: Observed at SLPS (Puerto Suarez, Bolivia) Figure III-48: Simulated at SLPS (Puerto Suarez, Bolivia)

III-26

Figure III-49: Observed at SLRB (Robore, Bolivia) Figure III-50: Simulated at SLRB (Robore, Bolivia)

III-27

Figure III-51: Observed at SLRI (Riberalta, Bolivia) Figure III-52: Simulated at SLRI (Riberalta, Bolivia)

III-28

Figure III-53: Observed at SLRY (Reyes, Bolivia) Figure III-54: Simulated at SLRY (Reyes, Bolivia)

III-29

Figure III-55: Observed at SLSA (Santa Ana, Bolivia) Figure III-56: Simulated at SLSA (Santa Ana, Bolivia)

III-30

Figure III-57: Observed at SLSI (San Ignacio de Velasco, Bolivia)

Figure III-58: Simulated at SLSI (San Ignacio de Velasco, Bolivia)

III-31

Figure III-59: Observed at SLSU (Sucre, Bolivia) Figure III-60: Simulated at SLSU (Sucre, Bolivia)

III-32

Figure III-61: Observed at SLTJ (Tarija, Bolivia) Figure III-62: Simulated at SLTJ (Tarija, Bolivia)

III-33

Figure III-63: Observed at SLVR (Viru-Viru, Bolivia) Figure III-64: Simulated at SLVR (Viru-Viru, Bolivia)

III-34

Figure III-65: Observed at SLYA (Yacuiba, Bolivia) Figure III-66: Simulated at SLYA (Yacuiba, Bolivia)

Bolivia Wind Atlas - FINAL · Headquartered in Cochabamba, TDE is Bolivia’s principal power...

Documents

Transcript of Bolivia Wind Atlas - FINAL · Headquartered in Cochabamba, TDE is Bolivia’s principal power...