Canada has aggressive targets for introducing wind energy across the country, but also faces challenges in achieving these goals due to the harsh Canadian climate. One issue which has received little attention in other countries not experiencing these extremes is the behaviour of composite blades in winter conditions. The scope of the work presented is to analyze the static stresses and fatigue response in cold climates using finite element models of the blade.

The work opens with a quantification of the extremes of cold experienced in candidate Canadian wind turbine deployment locations. The thesis then narrows its focus to a consideration of the stresses in the root of the composite blades, specifically two common blade-hub connection methods:​ embedded root carrots and T-bolts. Finite element models of the root are proposed to properly simulate boundary conditions, applied loading and thermal stresses for a 1.5MW wind turbine. It is shown that the blade root is strongly affected by the thermal stresses caused by the mismatch and orthotrophy of the coefficients of thermal expansion of the blade root constituents.

Fatigue analysis of a blade is then presented using temperature dependent material properties including estimated fatigue coefficients.It was found that the natural frequencies of a 1.5MW wind turbine blade are not significantly altered at cold temperatures. Additionally, cold temperatures slightly increase stresses in the composite blade skin when the blade is loaded, due to an increase in stiffness. Cold temperatures also lead to higher cyclic flapwise bending moments acting on the blade. However, this increase was found not to affect the lifetime fatigue damage. Finally, it was found that the cold climate as seen in Canada improves the fatigue strength of the saturated composite materials used in the blade. The predicted fatigue damage of the triaxial fabric and the spar cap layers in cold climates was therefore predicted to be half that of the fatigue damage at room temperature. This is caused solely by the temperature dependence of the fatigue coefficient b which requires further experimental verification to validate the numerical results of the current study.