ABSTRACT

Statistical methods suitable for analysing wave force data from a fixed cylinder in the sea have been applied to regular and random wave data from large scale laboratory tests on a smooth cylinder In-line force data from periodic tests were fitted to Monson's equation on a wave by wave basis to obtain Cm and Cd, using two different statistical assumptions u and u were uncorrelated or ulul and u were uncorrelated The latter, i e =O, was seen to hold more accurately by almost an order of magnitude, but identical Cm randomly slightly different Cd values were obtained Predicted and residual forces were also compared The random data were also analysed In this way Results for Cm, Cd with KC>4, and for CFrms were seen to agree well with the regular results A comparison with random wave results for a similar, slightly rough cylinder suggested that wave randomness reduces the effect of the roughness compared to the case in regular waves Finally, the random data were analyse using the mean square approach developed for the Christchurch Bay Tower experiments Cm was slightly higher and, when it could be determined, Cd was considerably smaller than from the wave by wave approach, results were found for only a narrow range of KC A number of difficulties in the application of the mean square method were also noted

INTRODUCTION

The hydrodynamic loading from waves is a major factor in the design of offshore and coastal structures The determination of these loads is still largely based on Monson's equation with coefficients derived from experimental results But there are uncertainties associated with extrapolating these data to full scale, post-critical conditions in random seas

This paper evaluates and compares statistical methods that are suitable for analysing force data from a fixed cylinder in the sea The data used originated from regular and random wave experiments carried out in the De Voorst (DHL) wave channel in the Netherlands

Research using harmonically oscillating water flows in U-tubes (Sarpkaya, 1976, Bearman et al, 1985b and many others) has greatly increased our understanding of wave loading Problems of scale, however, make it difficult to simultaneously obtain high Reynolds number (Re) and high Keulegan-Carpenter number (KC) in such tests In the 1980s several research programmes were initiated to study wave loading at scales that it was hoped would enable truly post-critical Re to be achieved at high KC There is, however, no one generally accepted way of analysing data from these experiments This is because the method of Fourier analysis used for planar oscillating flow is not so obviously applicable in non-hear or multidirectional waves, where many superimposed components may contribute to the velocity signal A further consequence IS that care must be taken when comparing results from different methods

In this paper analysis methods used by two of the large scale research programmes are compared

This content is only available via PDF.
You can access this article if you purchase or spend a download.