Passive and active compensation of uncertainty in trussed architectures

This thesis numerically investigates dynamic stiffness versus strength trades of imperfect passive and active trusses that fail in local elastic-stability modes and are relevant to space vehicle programs. First, nominal deterministic design scaling laws of trussed architectures are derived as functions of a primary design variable set. The laws describe the dynamic stiffness versus strength trades found as simultaneously optimized structural-controller Pareto fronts. The laws are functions of the primary design variables which are non-dimensional terms forming a design framework. These terms include descriptions of structural systems' slenderness, strength, tensioning, and active control mass fraction properties as well as a key active control power term. The laws are derived again in the presence of member straightness or eccentricity imperfections as well as sensor and actuator noise. These two imperfections are each characterized as a statistical uncertainty. Reliability-based design optimization techniques are employed to propagate the uncertainty distributions. The resulting nondeterministic scaling laws can be compared to the deterministic laws to study the pathology of the architectures. The pathology is how the architectures' non-deterministic performance departs from deterministic performance as a consequence of statistical imperfections.;Unlike previous studies, this design framework finds sensitivities to statistical imperfections in terms of non-dimensional design variables. Also, design decision points are probabilistically understood, e.g., when it is favorable to integrate active control. Member eccentricity statistical characterizations either position the active control favorability decision point at approximately 10% compensator mass fraction or shift the point to many multiples of this fraction. The former decision point occurs if eccentricity scales with truss member length and the latter decision points occur if eccentricity scales with truss member radius. In either case, a roughly 30% probabilistic performance penalty across the design space studied is paid for the statistical uncertainties considered.;The principal contributions of this investigation concern control design for active structures. Dynamically distinct high-authority control designs were discovered to coexist with low-authority control designs at traditionally low-authority compensator power rates. In the high-authority regime power is expended to dampen and, unlike in the low-authority regime, also stiffen the closed-loop structural systems. High-authority designs are feasible irrespective of truss topology and connectivity. The active control power term accurately predicts the pivotal compensator power rates delineating the two control authority regimes within the non-dimensional design framework. These predictions are statistically insensitive to active control mass fraction and imperfection uncertainty.