Could statistical analysis streamline offshore wind site characterization?
A Study on Applying Cluster Analysis to Geophysical Data
When it comes to offshore wind, we know what it takes to put a plan into action and get the blades turning. Broadly, it involves years of careful planning, data collection, data processing, and construction.
What if we could increase the efficiency of this process?
In this blog, I will discuss the use of statistical analysis on geophysical data to streamline the crucial planning step of site characterization.
This project was a master’s thesis that I completed in May of 2020, while studying at the University of Delaware. It focuses on geophysical data collected within the Maryland Wind Energy Area (WEA). The objective was to perform a more efficient site characterization by using cluster analysis on bathymetry data and processing select subsurface geophysical lines within those clusters. If successful, the characterization could be considered when making a preliminary foundation recommendation.
Figure 1. Maryland Wind Energy Area (WEA).
During the study,1,084 lines of Chirp subsurface data were collected throughout the WEA. Chirp geophysical data is similar to how a bat uses echolocation. A sound wave reflects off subsurface layers back to a receiver, creating an image we can analyze. The initial characterization approach was to analyze and identify features on each line. To streamline characterization, I collected bathymetry data within the WEA and processed this data using K-Means cluster analysis to create five unique bathymetry “clusters”. A select number of Chirp lines from each cluster were then analyzed to provide a first order characterization.
What were the results?
When compared to more in-depth geological studies of the Maryland WEA, this method provided an accurate characterization at a fraction of the processing time. The originally proposed processing method of all 1,084 Chirp lines would take approximately 360 hours. The processing level of this study, statistical analysis included, took approximately 60 hours. This method was successful in part due to prior knowledge of paleochannels present within the WEA. The characterization highlighted four distinct subsurface units, paleochannels in the western portion of the WEA, and parallel layers of sands and muds in the eastern portion.
After evaluating monopile, gravity base, jacket, and suction caissons and their feasibility within the Maryland WEA, the preliminary foundation recommendation is a jacket or tripod with suction caissons instead of piles. Jacket foundations use less steel than monopiles and caissons are easily produced, installed, and decommissioned without the use of pile driving.
Figure 2. Jacket foundation with suction caissons.
From this study, I found that this foundation type, as seen in the figure above, would be most successful in the eastern portion of the WEA due to the sediment type present and absence of paleochannels. The cluster area method effectively streamlined the characterization process.
If you are interested in learning more about how this methodology can help narrow your project design envelope before geotechnical data surveys, please contact me for more information.