This paper deals with the analysis of recent full-scale strain measurements in the hull of a large container carrier covering several months of operation. The focus is on the real-time prediction accuracy of responses 5-15 seconds ahead of the measurements. Such results are less applicable in the operation of container carriers but are important in e.g. loading/unloading operations at sea or helicopter landings. Three different procedures are discussed: Conditional processes with analytical estimates of the mean values and standard deviations, the autoregressive predictor method and a method based on superposition of sinus-oidal components. The conditional processes do not need offline training and will be applied to measured time series in order to evaluate the accuracy of response predictions within the next 1-30 seconds. The number of measured points and the time distances between them are varied to determine the best solutions. A procedure based on 11 measured points spaced 1 sec, covering the last 10 sec of the instantaneous measured signal seems generally able to give fair predictions up to 5-10 sec ahead of the current time. The full-scale data is provided through the EU FP7 project Tools for Ultra Large Container Ships (TULCS) project no. 234146.
Proceedings of the Prads 2013, 2013
Main Research Area:
12th International Symposium on Practical Design of Ships and Other Floating Structures, 2013