【正文】 than the wave acting width, the slightly nonhomogeneous of wave action is responsible to the decrease of the dimensionless total forces. Third, as the deck width is in excess of the wave acting width, the total forces increase little and the pressure distribution length remains unchanged , therefore the dimensionless forces seem almost constant. It should be noted that, at a large clearance level, the deck width is mostly larger than the wave acting width, and thus the dimensionless uplift forces slightly decrease with the increasing deck width.3. 2. 2 ?? Prediction Model of Uplift LoadsThe dominant influence factors for the loading process on the deck of the exposed highpile jetty were found to be the impact angle of wave surface, air layer and the wave dynamics. Based on the analysis of measured data, new prediction method was developed by utilizing the envelope for all tests in order to make sure of safety in engineering application, as shown in Eq. ( 3) . The effect of the relative width of deck is included and the coefficient 1. 1 is introduced in calculating the crest elevation to represent the wave reflection from the deck and the downstanding beams. Fig. 7. Dimensionless maximum uplift load versus Ls / B.3. 2. 3 ?? Comparison of the Measured Dimensionless Uplift Force with the Prediction by New MethodPredictions of waveindeck uplift loads by Eq. ( 3) are pared with the measured data in Fig. 8 and Fig. 9. The parison shows that the new prediction method gives a good result on uplift forces with large magnitude, while deviations in this model are mostly for the forces with small magnitude. The main trend is that it underestimates the forces at high clearance cases, the corresponding forces are small and not the critical situation for design. Generally, the model gives conservative results for design.3. 2. 4 ?? Comparison of the New Prediction Method with the Existing Prediction MethodsFour examples of parison between the new prediction model and the existing three prediction models are given in Fig. 10. The distribution lengths of pressure are taken as L s/ 4, x 1% ( without wave reflection coefficient) , L s/ 6, x 1% ( with wave reflection coefficient ) respectively for the Goda model ( 1967) , the existing guidance model ( 1994) , the Guo and Cai model ( 1980) , and the new prediction method. For case that x 1% is larger than B, it is taken as B .The parison shows the following details:As a result without considering the effect of the deck width, the existing guidance model overestimates the forces on the wide deck (with width B = Ls ) , but underestimates the forces on the narrow deck (with width B = L s/ 8) . The existing guidance model predicts that the maximum uplift force appears at the still water level, which disagrees with the trend explored in the test that the peak force occurs at the relative clearance above the still water level ??h/ ??1% = 0. 4~ 0. 8. The force reduces sharply with the clearance increase due to the linear relationship, and thus the predicted uplift force is paratively small in large clearance. The range of clearance corresponding to the effective uplift force is narrow without considering the effect of wave reflection in calculating the wave contact length and the crest elevation. The Goda model gives highly large prediction for uplift load, as Goda model is fitted for the open trestle bridge in deepwater, the wave reflection from structure is large and the resulted slamming of interference standing wave is responsible to the great impact pressure on deck.The Guo and Cai model estimates the uplift loads by the impulsive pressure, thus it gives relatively large result. The distribution length Ls / 6 is adopted and proportional to wave length, but is independent of the clearance, which leads to overestimation of wave force with large wave length and for deck with high clearance. Compared with the predictions by other models in Fig. 10, the equations derived here show significant improvement, with consideration of the effect of deck configuration and the effect of wave reflection. Moreover, it should be noted that the forces on downstanding beams and berthing members contribute to large forces on the superstructure of jetty at high clearance level, though in that situation the wave can not reach the deck and the force on deck is zero. Therefore, for large clearance cases, the forces on deck, downstanding beams and berthing members should be calculated separately, and the sum is then the desired result. Applications in case study confirm that it provides a more consistent predict ion with the practical results. Fig. 8. Dimensionless maximum uplift loads and the envelope of the maximum.Fig. 9. Comparison of the measured dimensionless uplift forces with the proposed new prediction method.Fig. 10. Comparison of new prediction model with existing prediction models.3. 3 ?? Probability Distribution of Uplift LoadsWaveindeck uplift load induced by irregular waves is random. The parameterization of the data is in normal distribution, threeparameter Gamma distribution, Weibull distribution and Rayleigh distribution. The KolmogorovSmirnov test is advocated to test the acceptance of the given four models. Conclusion from the study is that it seems to accept the Weibull distribution for 75% of test data sets, mainly those poorly fitted sets are relevant to large clearance case. Deviation at higher clearance is understandable, because for higher level deck, small waves in a random wave sequence can not reach the deck and the values in the corresponding force data set are zero. This results in poor fitting of the Weibull model to measured uplift loads. The analyzed results of the real data confirm that the load distribution obeys Weibull distribution and the conversion ratios of wave loads with different exceedance probabilities are suggested in Table 1.Table 1 Conversion rati