(76d) Model Based Online Analysis and Monitoring of Tank Qualities

A typical refinery or chemical plant has a tank farm consisting of 50-300+ tanks used to receive, store and dispatch material in and out of it. The manufacturing process dictates that the amount and quality of tanks contents be known to all concerned plant departments as soon as possible for their business decisions.

In an automated plant, the tank inventory is obtained online by using an Automatic Tank Gauging (ATG) system whereas the tank quality information relies only on manual samples and their lab analysis. Typically, a plant spends 2-8 MUS$/year on lab analysis of tank contents. The quality information of tank contents is not possible in near real-time even if there are analyzers at the tank inlet or outlet as they indicate qualities only of inlet / outlet streams and not of tanks while loading or unloading a tank. Consequently, any decision or action by refinery planners and operations management, based on tank quality, may be outdated soon if the tank has been active during the decision process.

This paper discusses successful implementation of a sophisticated and state-of-the-art Tanks Quality Tracking System to provide open-loop online tracking qualities of run-down tanks of three Diesel and two Kero blenders and gasoline blender components tanks in a refinery in Asia-pacific region. The system has helped the refinery to reduce their qualities giveaways of 90%, Sulphur, flash and freeze points to acceptable limits and minimize reblends of off spec products. The system has been fully operational at the refinery since August, 2002 and has flexibility, robustness and features to incorporate any of the built-in 100+ linear / non-linear models for the qualities of crude, gasoline, diesel, fuel and other products' tanks considering all possible configuration of inlet / outlet flows, lab analysis of tank samples, inlet / outlet analyzers and other factors. The system has client-server architecture resident on Windows NT/2000 platform with SQL Server /Oracle DBMS and interfaces with any DCS system and field signals. Furthermore, ROI of such a system is less than one month based on the estimate of realizable tangible benefits and was customized and implemented within two months. The system is also plug N' Play upgradeable with either already installed or third party's feedback control or optimizer module.