A MONTE CARLO APPROACH FOR ESTIMATING EXTREME CURRENTS IN THE SINGAPORE STRAITS

摘要

Utilizing the independency of tide, through-flow, surge and high-frequency currents in the Singapore Straits, a Monte Carlo simulation method of combining the different components is proposed, expanding the horizon of available measured and modelled data and facilitating the definition of design current speeds. The statistical model proceeds by, first, making N number of random picks from the non-exceedence probability distributions of the surge, through-flow and high-frequency components. The number of random picks made in a given year for each component, N, is defined by assuming its occurrence rate is Poisson-distributed around a known annual mean value. N number of random start times are then chosen from each year and the maximum value of tidal current predicted over an ensuring 3-day window is combined with the randomly sampled component (either surge, through-flow or high-frequency current). Assuming an intended design life of 50 years, this process is repeated N number of times in each of the 50 years and for each current component, yielding 50 annual maximum values. For random 3-day windows that overlap, the model takes the vector sum of the maximum tidal current and the 2 (or 3) concurrent components. The process is repeated 1000 times, producing 1000 * 50 values of annual maxima which are then assigned non-exceedence probabilities. Return Period levels are obtained directly from the non-exceedence probabilities. The method provides a reduction in design current when compared to values derived by multiplying the exceedence probabilities of the varying independent contributions directly.