Papers »

Multi-scale stress usage monitoring from global acceleration measurements: an experimental validation

Usage monitoring is the process estimating the loads/stresses in a particular member or structure. Stress usage monitoring is an essential component of fatigue damage diagnosis and prognosis. In this paper we investigate the accuracy and computational efficiency of two different algorithms for stress usage monitoring, namely, the Kalman filter and a recently developed finite element model-based observer. The comparison is carried out in the context of an experimental rectangular frame structure instrumented with 4 accelerometers and 8 dynamic strain sensors. The frame is made of steel plates and has overall dimensions of 48” x 12” x 6” consistent of two continuous columns and four beams. The thickness of the members is 0.125” and are connected using steel bolted angle connections. The structure is excited laterally, along the weak direction, by realizations of a non-white stationary random process delivered through an electrodynamics shaker. Four different load locations were considered, each corresponding to the elevation of each beam. For each location of the load, four acceleration measurement scenarios were considered. In each test, the time histories of acceleration measurements are used as inputs to the state estimation algorithms. Each algorithm estimates mean and variance of displacements at all degrees of freedom of the model. The estimated displacements, and its uncertainty, are converted into strains at critical locations along beams, columns and angle connections. This is done by coupling a global structural model of the frame with a localized finite element model of the various connections. It is found that the finite element model-based observer outperforms the Kalman filter in estimation accuracy and computational efficiency.

Author(s):

Néstor Polanco    
The University of Vermont
United States

Eric Hernandez    
The University of Vermont
United States