DEVELOPMENT OF AN EXPERT SYSTEM FOR THE OPTIMUM DESIGN, CALIBRATION AND COST ESTIMATION OF OIL STORAGE TANKS

Abstract

As a result of expansion in oil sector, there is need for indigenous design and installation of storage tanks in refineries and depots to adequately store the petroleum products instead of depending on turnkey projects from the foreign engineers, so as to preserve the country’s foreign exchange. The specialists in oil tank design, calibration and costing were consulted in acquiring domain knowledge manually for developing integrated Expert system incorporated with multicriteria decision tools. Basic knowledge such as design equations and other considerations were elicited from reference materials (API 650, API MPMS 2.2A). In tank design, height – diameter (H – D) ratio was considered as a major factor and it ranged from 0.5 to 1.5. The optimum design was established within this range based on the preliminary design considerations (criteria) such as: Area occupied (m2), Weight (kg), Wind Moment (Nm), Seismic Ringwall Moment (Nm), Base shear (N) and Estimated Cost (₦). In order to establish the height – diameter ratio that results in the optimal design, Multi Criteria Decision Method (MCDM) was adopted. Objective weight was assigned to each of the criterion using CRITIC (Criteria Importance Through Intercriteria Correlation). Eleven alternatives A1 to A11 (0.5 to 1.5) were identified as possible options for H – D ratio; this is because a minimum of ten data set is required for analysis. Among these alternatives, the best alternative was chosen based on the ranking of alternatives through TOPSIS (Technique for Order Preference by Similarity to Ideal Solution). Modified version of TOPSIS, known as Fuzzy-TOPSIS was used in the selection of best method of tank calibration while Engineering Build-up cost estimation was adopted for the cost estimation. Build Up Engineering (BUE) method of cost estimation was used by considering the bill of engineering for each cost element. The garnered expert information were programmed in Java programming language to develop Expert system package. The basic expert design parameters such as the weight, ix Ringwall moment, estimated cost and Base shear of the Developed Expert System (DES) were compared with that of actual human experts using the Analysis of variance (ANOVA) at 5% significance level. A case study of an on-going project on the design and fabrication of 13,000 m3 storage tank was considered. The objective weights as obtained by CRITIC were 0.2356, 0.2607, 0.1238, 0.1249, 0.1196, 0.1354 for the Area occupied, Weight, Wind Moment, Seismic Ringwall Moment, Base shear and Estimated Cost respectively. Ranking of the alternatives by TOPSIS rated the alternative A4 (0.8 H – D ratio) as the best alternative that gives the optimal design with the closeness coefficient of 0.5801. Electro-Optical Distance Ranging (EODR) method of tank calibration was rated best while Optical Reference Line Method (ORLM) was ranked least by Fuzzy TOPSIS. The results of comparison between the output values of the DES and that of the human experts by ANOVA gave the p-value of 0.941, 0.123, 0.956 and 0.077 for the weight, Ringwall moment, estimated cost and Base shear respectively. This confirmed that there was no significant difference between the mean value of the DES results and that of the human experts. An Expert system has been developed for oil and gas storage tank design, calibration and cost estimation based on the elicited information obtained from respective domain experts, API 650, API MPMS 2.2A-D, with the incorporation of MCDM for decision making. It is therefore recommended that the findings of this research be deployed to oil industry for industrial use and also as a teaching aid for people enrolling for storage tank design and calibration courses.