This page contains a Flash digital edition of a book.
Feature 2 | CAD/CAM UPDATE


HSVA develops the empirical prediction method


Today prediction of resistance and propulsion in ice covered waters is usually carried out by using well established semi empirical methods. Hamburgische Schifau-Versuchsanstalt GmbH (HSVA) looks at the development of empiric prediction methods for ice-going ships based on artificial neural networks.


F


or navigation in level ice, an approach based on certain ship main dimensions as well as


hull shape angles and ice properties is used [1]. Te total resistance in ice encountered by the ship is split up into components that each include different physical phenomena like initial crushing, breaking, as well as submersion and sliding of ice floes along the hull. For prediction of resistance and


power in the broken channel, the rules calculation according to Finnish-Swedish Administration is used. Te same is also applicable for the approach for level ice formulas which include certain hull form and ice perameter. Another similarity is the superposition of single components to determine the total resistance [2]. Te disadvantage of these procedures


is that the interactions of simultaneous effects are hardly taken into account. Results of model tests and full scale trial on the other hand show high correlation between single parameter influences, for example the dependence of total resistance to ice thickness and ship speed. Another aspect is the validity of the existing methods, which are based on a certain range of ships that are comparable. Since establishing these methods, ice breaking hull shape has been further developed and the number of ships with a conventional hull form operating in ice covered waters has increased. To offer a prediction including both, a


preferable realistic parameter basis and a larger range of validity concerning hull shape and main dimensions, a prediction method based on artificial neural networks (ANN) was developed in scope of a master thesis at Hamburg Ship Model Basin [3]. Neural networks offer the possibility of learning multiple relations


52 Figure 1: Gradient descent method.


requested delivered power). Te training itself is performed with gradient descent methods (see figure 1). If the training set includes enough


information, in a second step, the network should be able to generalise the dependencies to predict the target values by using an unknown input data set. To avoid the memorisation of


presented data during the learning phase, the network has to be validated continuously by using unknown data sets. The optimum training stage is reached, if both training and validation error have reached their maximum (see figure 2). To enable the networks to learn


Figure 2 : Error for training and validation.


the relvant parameter relations, data collected during model tests at Hamburg Ship Model Basin were used. Te input vector included main ship dimensions, hull shape and ice parameter. Te results produced, presenting unknown input vectors to the network showed acceptable accuracy and plausible dependencies. Besides prediction in the early design


stages, the networks may be used to interpolate a parameter range (see figure 3) and can therefore be used to confirm or amend results gained by model tests, as there are usually not enough data to cover each single parameter range. NA


Figure 3: Results of model tests and ANN.


of a physical problem without requesting an explicit approach. Assumptions about superposition or interaction of single components or effects are, therefore, unnecessary. Te networks are trained on a certain parameter set including both, input and target qualities (resistance,


References [1] Lindqvist, G.: A straight forward method for calculation of ice resistance of ships, In POAC, June 1989


[2] Riska, Kaj: Performance of merchant vessels in ice in the Baltic, Helsinki University of Technology, December 1997


[3] Reimer, Nils: Develoment of empiric prediction methods for ice going ships, Hamburg University of Technology, May 2010


The Naval Architect January 2011


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68  |  Page 69  |  Page 70  |  Page 71  |  Page 72