ANALYTICAL IMPROVEMENT OF THE STATISTICAL EXPERIMENTS METHOD FOR THE PURPOSES OF PORTS DESIGN

Abstract

The planning of business processes in sea transport is the foundation of transportation companies and technological projects development forecast. The calculations of business processes parameters are mainly complicated by the fact that most of the input values are stochastic ones. This is due to the specifications of technological processes and influence of a wide specter of external environment events (e. g., hydrometeorological and ice conditions) on the one hand, and market conditions (e. g., competition conditions, market agents behavior, exchange rate) on another hand. To provide the necessary level of results reliability, the calculations, that are implemented during research of specifical sea transport operations tasks (such as technological design of seaports), require to pay a great effort in experiments planning. Planning the experiments means the process of choosing the conditions, procedures and methods of experiments running. It also includes the definition of how many experiments should be provided. In other words, the purpose of experiments planning is providing the results maximum precision and reliability with minimum number of experiments. This element of research could take a whole volume in the scientific research. A method for implementing mathematical calculations with stochastic values, which allows you to define the form of the output probability distribution without Monte-Carlo method utilization, is suggested in the paper. It is proved in the paper that the output distribution is close enough to the results, which are calculated with conventional methods. At the same time, when the number of variables is small, the method requires less computational resources.