Traffic signal structures undergo wind-induced vibrations that result in fatigue damage accumulation and reduced service life. Mast arms have failed and required removal while in service. A dual experimental and analytical modeling approach is taken to mitigate fatigue and fracture in steel traffic supporting structures. A full-scale prototype structure is instrumented to study natural wind response. Excitation mechanisms are identified, and response is characterized statistically by a lognormal distribution. Helical strakes are found to reduce the vortex-induced vibration of cantilevered traffic signal structures, however are not a panacea for fatigue mitigation as marginal service life gains occur in severe wind environments. A probabilistic framework is extended to assess the risk of wind-induced fatigue and estimate service life while considering uncertainties in fatigue demand and capacity. The framework is successfully demonstrated against compiled inspection records. Locations with higher prevailing winds are susceptible to wind-induced fatigue, but the prevalence of low-speed vortex-induced response is primarily responsible for the early fatigue failures in more mild environments. A low-cost damage avoidance system is proposed to mitigate fatigue and fracture in steel traffic supporting structures. Applied prestress introduces a fail-safe, supplementary load path to balance dead load moment, eradicating the detrimental tensile mean stress found in traffic signal structure connections. Field observations are made without and with the proposed system installed. The benefit of applied prestress is quantified by determining service life without and with the system based on changes in response and fatigue resistance using: (i) a code-based technique; and (ii) the proposed probabilistic framework. Fatigue performance is modeled as mean stress-dependent by modifying nominal stress-life relationships. Service life is shown to increase by an order of magnitude, regardless of wind environment. The concept shows potential to reduce the detrimental effects of non-redundancy for a variety of similar, fatigue-critical infrastructure components. The validity of simplified mean stress-dependent connection modeling is explored. A fracture mechanics-based, total life (initiation-propagation) model is used to demonstrate the detrimental effect mean stress has on tube-to-transverse base plate fatigue resistance. Using fatigue strength curves derived from total life analyses, probabilistic analyses are repeated to justify the use of simplified models.
