Impact/Purpose:

Wind generated electricity can provide an environmentally beneficial alternative to conventional energy sources. The use of wind power can reduce demand for fossil fuels, decrease local air and water pollution, and reduce greenhouse gas emissions. Humboldt County has an appreciable wind resource and great potential for wind development on both the small and large scale. For large wind farms site assessment is a small portion of the overall cost, but the small to mid-sized wind turbine market is hindered by a lack of reliable, accurate, and low cost wind resource data. Current practices require at least one year of measured wind data or the use of expensive software to estimate a site’s resource. We propose to develop and evaluate a low cost method for site assessment based on relatively short monitoring periods.

Description:

Our Phase I successes involve the installation of a wind monitoring station in Humboldt County, the evaluation of four different measure-correlate-predict methods for wind site assessment, and the creation of SWEET, an open source software package implementing the prediction methods.

Tower Installation
After ten months of planning, site preparation, and construction, the Renewable Energy Student Union (RESU) successfully installed an 80ft wind monitoring station on a ridge east of Humboldt Bay. This site is in a region with a high potential for wind power development. Based on the data collected at the site, as much as 4,230 lbs CO₂/year could be displaced if the landowners choose to install wind turbines.

From the installation, we learned that the building codes in Humboldt County are rigorously defined and strictly enforced. To satisfy these requirements, we paid for a full structural analysis of our traditional tilt-up tower, as well as 5 cubic yards of concrete to anchor the tower. Ultimately, the labor and cost of materials and services necessary for permitting our tower amounted to 250 person hours and $2700, which is $2000 more than we anticipated.

Comparison of Measure-Correlate-Predict Methods
Despite the difficulties we encountered installing the monitoring tower, we achieved notable success in our implementation of a software tool to perform the correlation and prediction. Four statistical correlation methods were utilized and their predictive capabilities were evaluated using a variety of data sets. These methods are referred to as the Variance Ratio Method, Mortimer’s Method, the Multi-model, and Artificial Neural Network. The four methods were also compared against a “no correlation” control scenario.

Overall, the Variance Ratio method produced the most reliable predictions, with a standard error of 2%, in the estimation of energy production at a given site. It has the additional benefit of being simple to use. The Multi-Model also performed well, but the model is much more complex and time intensive to use. Mortimer’s method systematically overestimated average wind speed and the Neural Network was the least consistent of any of the methods.

We were able to make predictions that substantially improved upon the no correlation method, using “well behaved” data from buoys on the Pacific Coast and monitoring stations in North Dakota. These data sets are considered “well behaved” because they come from high quality monitoring stations in locations with little obstruction from topography or terrain roughness, namely, the Pacific Ocean and the great plains of North Dakota. The performance of the statistical methods demonstrates that our methodology is valuable in assessing wind power potential based on a small period of monitoring.

However, due to the relatively low quality sources of data in Humboldt County, the correlate and predict methods did not improve upon the no correlation method. We believe a long term source of high quality data is therefore needed to apply the methods locally.

Software Tool
We have created a web-based software application, SWEET, which uses statistical methods to predict long term wind behavior at a potential wind site, based on historical data from nearby weather stations. Currently, SWEET is a prototype application based on open source programming languages.

URLs/Downloads:

Record Details:

Related Organizations:

Project Information:

Our study will include two components: A) We will use long term wind data from local monitoring stations to develop and verify a correlate and predict software tool using a variety of statistical analysis techniques. B) We will install a wind monitoring station in the Humboldt Bay region to further verify the software tool and to provide the experience of installing and managing a wind monitoring system to the student team members.

The long term goal of this project is to provide a service that will decrease the overhead associated with wind site assessment, thereby promoting wind power development in Humboldt County.

Our study will include two components: A) We will use long term wind data from local monitoring stations to develop and verify a correlate and predict software tool using a variety of statistical analysis techniques. B) We will install a wind monitoring station in the Humboldt Bay region to further verify the software tool and to provide the experience of installing and managing a wind monitoring system to the student team members.

The long term goal of this project is to provide a service that will decrease the overhead associated with wind site assessment, thereby promoting wind power development in Humboldt County.

Project IDs: