Trans RINA, Vol 151, Part B2, Intl J Small Craft Tech, 2009 Jul-Dec ©2009: The Royal Institution of Naval Architects FIFTY YEARS OF THE GAWN-BURRILL KCA PROPELLER SERIES D Radojčić, A Simić and M Kalajdžić, University of Belgrade, Serbia (DOI No: 10.3940/rina.ijsct.2009.b2.92) SUMMARY The first extensive systematic tests of flat-faced segmental-section propellers were those performed by Gawn in 1953 (open-water tests) and Gawn and Burrill in 1957 (cavitating environment). Since then several attempts to develop mathematical representations of propeller hydrodynamic characteristics (thrust coefficient K T and torque coefficient K Q ) have been made in order to improve computer capabilities in predicting propeller performance. The first mathematical model was that of Blount and Hubble (1981), which was soon followed by Kozhukharov’s (1986) and then Radojcic’s (1988). These models were developed through application of multiple regression analysis. Koushan (2007) challenged more than 20 years of the regression approach, for representing the highly non-linear Gawn-Burrill KCA propeller characteristics, and suggested application of the artificial neural network technique. This paper compares the four mathematical models mentioned above. NOMENCLATURE DAR developed area ratio EAR expanded area ratio J advance coefficient K T thrust coefficient K Q torque coefficient ΔK T K T reduction for cavitating conditions ΔK Q K Q reduction for cavitating conditions P/D pitch ratio Q c torque load coefficient z number of propeller blades η 0 propeller efficiency η atm propeller efficiency non-cavitating conditions η cav propeller efficiency cavitating conditions σ cavitation number based on advance velocity σ 0.7R cavitation number based on resultant water velocity at 0.7 radius τ c thrust load coefficient 1. INTRODUCTION Flat-faced, segmental section propellers are relatively simple to manufacture, easy to repair and have respectable open-water and cavitation characteristics. These are the main reasons they are still widely used for small, high-speed craft. Amongst the first systematic tests of these propellers were those performed by Gawn in 1953 [1] (open-water tests) and Gawn and Burrill in 1957 [2] (cavitating environment). Several attempts to develop mathematical representations of propeller hydrodynamic characteristics (thrust coefficient K T and torque coefficient K Q ) have been made in order to improve computer capabilities in predicting propeller performance. The first mathematical model, denoted here as Model 1, was that of Blount and Hubble [3]. They derived separate equations for open- water characteristics, based on Gawn’s Admiralty experiment works (AEW) data (actually polynomials of the B series propeller form were used with 39 and 47 terms for K T and K Q respectively, but new regression coefficients were evaluated), and for the cavitating regime (represented by a relatively simple set of equations). This model was followed by Model 2 (Kozhukharov [4]) with single equations for both regimes which had 121 and 116 polynomial terms for K T and K Q respectively. In order to reduce the number of terms and increase the accuracy of previous models, Radojcic [5] published separate equations for non- cavitating and cavitating regimes, Model 3, the first having only 16 and 17 and second 20 and 18 polynomial terms respectively. The mathematical models mentioned above were developed using multiple regression techniques. Recently, Koushan [6] challenged more than 20 years of the regression approach for representing highly non-linear Gawn-Burrill KCA propeller characteristics, and suggested application of the artificial neural networks (ANN) technique. Koushan derived separate equations for the non-cavitating and cavitating conditions, Model 4, with 34 equation terms for both K T and K Q for the non-cavitating regime, and 89 and 107 terms respectively for the cavitating regime. It should be noted that only Model 1 was based on AEW data (for non-cavitating, i.e. open-water conditions), while Models 2, 3 and 4 were based on KCA data (for atmospheric and cavitating conditions). AEW propellers are geometrically close to KCA propellers – see [1] and [2]. Data for cavitating conditions of Model 1 are based on various sources, see Blount & Hubble [3]. Moreover, it is often assumed that open-water and atmospheric conditions are the same, although that is not the case (this discussion, however, is beyond the scope of the paper). Consequently, Model 1 should inherently produce slightly different results from Models 2 to 4. Common for all models, however, is that they are applicable for 3- bladed segmental section propellers under cavitating and non-cavitating conditions. The primary purpose of this paper is to compare the mathematical models mentioned above. In addition, an analysis of results of two decades old regression techniques and those of the more modern artificial neural network technique is given.