Experimental study of the effect of Q and R on linear quadratic estimator performance in estimating pipeline dynamics

Pipeline health monitoring in the oil and gas industry is critical, where any damage or loss of its integrity causes massive cost. Therefore, early prediction and estimation of failure is of a particular interest to related industries. This paper discusses the tuning of pipeline integrity monitoring technique which is based on Linear Quadratic Estimator (LQE). The presented method requires minimal feedback measurement from the physical pipeline, while vibration can be predicted at any location along the pipeline using both a structural dynamic model and measurements from sensors placed at specific locations on the pipeline. In this approach, Eigen values and Eigen functions are obtained from a finite element model (FEM) of the pipeline. A Kalman estimator is constructed based on the FEM model while corrected estimates of pipeline vibration are generated using DSpace in conjunction with Matlab as well as actual measurements obtained from the pipeline at specific locations for both horizontal (x) and vertical (y) pipe directions. This paper also discusses the effect of weighting matrices Q and R on LQE performance in estimating pipeline dynamics. Our experimental results show that careful and fine tuning of LQE by varying the values of Q and R has generated good agreement between the actual and estimated vibration signature at any point along the pipeline span.