07 July, 2017

Background

  • Onshore wind turbines provide key opportunity for decarbonisation
  • High resource availability
  • Low land usage
  • Cost competitive against many conventional technologies
  • Technology highly site specific

What makes a good site?

Different actors consider different parameters:

Site developers

  • Largely concerned about Economic parameters
  • Site endowment
  • Easy to calculate

Local parties

  • Social and Environmental concerns
  • Visual intrusion of sites
  • Difficult to assess impacts

Wind Turbine Site Selection

Existing Research

Global research into identifying suitable turbine sites. 15 key studies identified since 2000.

Existing Research

  • Extensive use of Geospatial Information Systems (GIS) to identify suitable sites
  • Multi-criteria decision analysis (MCDA) used extensively to determine best options
  • Primarily based on geospatial parameters

Example Studies

[1] J. R. Janke, "Multicriteria GIS modeling of wind and solar farms in Colorado," Renew. Energy, vol. 35, pp. 2228-2234, 2010.
[2] J. J. W. Watson and M. D. Hudson, "Regional Scale wind farm and solar farm suitability assessment using GIS-assisted multi-criteria evaluation," Landsc. Urban Plan., vol. 138, pp. 20-31, 2015.

Development Patterns

Research Gap

Two key challenges were identified:

  1. Are we using the right parameters to assess site suitability?
  2. Can we accurately predict the likelihood success rate based on geospatial parameters?

Research

Research Approach

  • Inverse/Retrospective GIS approach used to develop model parameters
  • Logistic regression analysis used to assess influence of parameters

Turbine Data

Location of Wind Turbines used within the analysis

Location of Wind Turbines used within the analysis

Model Parameters

  • 30 parameters gathered
  • Parameters mapped to wind turbine sites.

Findings

  • 9 parameters were found to be significant
  • Optimisation of model to identify key parameters

Segmented Model

Second set of models built to explore whether there were regional differences in parameters.

Model Fit

Variable Full Reduced Nested: England Nested: Scotland Nested: Wales
Observations 1476 1476 646 698 132
Parameters 27 9 10 11 10
Nagelkirke R2 0.11 0.1 0.11 0.16 0.24
Pearson Chi-squared 113.7 111.9 51.5 89.3 25.9
Residual deviance 1932 1932 833 895 157
Model Accuracy 60% 62% 60% 65% 57%

Model Generalisation

Implications

Outcomes

  • Range of siginficant parameters identified as significant within the analysis
  • Quantitiatvely linked social characteristics to acceptance rates.
  • Relatively poor model fit demonstrates issue of purely geospatial analysis

Future Work

Model refinement

  • Additional parameters added to the model
  • More detailed geospatial variance

Applying findings

  • Sensitivity integrated into a geospatial model for predicting sites
  • Can we understand where sites are best located?

Dissemenation

  • Academic
  • Local planners
  • Developers

Future Work

Thank You