Probabilistic Design of Midship Panel based on Model scale compressive Ice Test

• Main Author: Neumann, Karoline Mali
• Other Authors: Leira, Bernt Johan, Ehlers, Sören, Kujala, Pentti, Norges teknisk-naturvitenskapelige universitet, Fakultet for ingeniørvitenskap og teknologi, Institutt for marin teknikk
• Format: Master Thesis
• Language: English
• Published: Institutt for marin teknikk 2013
• Subjects: Sea ice
• Online Access: http://hdl.handle.net/11250/238745

This thesis is investigating compressive ice loads acting on the mid-ship using a model experiment performed in Aalto ice tank where atactile sensor was mounted on the side. In order to get insight in thevery complex behavior of ice; the sea ice growth, mechanical prop-erties and failure mechanisms are presented. Further some previouswork on the topic is presented, with discussion, in addition to regula-tions on local ice loads and structure requirements. The execution ofthe experiment is described, and the method for processing the dataand based on this data using probabilistic design to design a midshippanel. For datasets where the interaction area is semi-continuous anew event denition is proposed, based on temporal events consistingof spatial events. The maximum event method developed by [Jordaanet al., 1993] is applied in sampling data from the new temporal events.Data is sampled for increasing area sizes corresponding to number ofconnected triggered censor cells. The data is adjusted for exposure inorder to have a standardized curve corresponding to the area in ques-tion. An exponential distribution is tted to the tail of the data, andpresented in a Weibull probability plot. The parameters of the distri-butions, x0 and are functions of area. Assuming the area is acting ina line corresponding to the span of a longitudinally framed panel, andgiven a return period and a scenario, a design load is predicted. Basedon this load, scantlings of a panel is recommended which also complieswith Finnish-Swedish ice class rules. The integrity of the structure ischecked using Monte Carlo analysis. To get more insight into the dataset, a local pressure area curve, and an average pressure over totalmeasured contact area relationship is presented for the entire data set,as well as spatial and process pressure area curves and pressure historyfor the biggest load event and the biggest pressure event.