Fault Detection And Data Validation In Refinery Power Production

CATEGORY: COMPRESSORS
THESIS
Fault Detection And Data Validation In Refinery Power Production
A Thesis Submitted To The Graduate School Of Natural And Applied Sciences Of Middle East Technical University
By Tuba Kahya
In Partial Fulfillment Of The Requirements For The Degree Of Master Of Science In Chemical Engineering (February 2014)
Abstract
In refinery utility generation, boilers, steam and gas turbines are the primary equipment, producing steam as well as electricity. In boilers, high pressure steam is produced by combustion of a mixture of waste gases from the production line and natural gas – called fuel gas. The steam is at relatively high pressures and is then reduced to lower pressures in steam turbines. The resulting medium and low pressure steam are sent to the refinery for use in processes. In a gas turbine, electricity is produced by compressed air using energy released from combustion of natural gas. In this study, the aim is to model the efficiencies of the steam production equipment, tracking their performance in an effort to optimize the equipment utilization.
For this purpose, historical data from a refinery are verified. Material and energy balances are checked on individual units using ASPEN software to verify the data as a reference. Then, MATLAB software is used to adapt our situation in the plant. In these analyses, clustering methods are used. Results of the model and analysis of historical data allows us to make recommendations on which sensors need to be calibrated and how data integrity issues can be resolved.
In the examined case, it was postulated that oxygen sensor at the stack gas line was accurate and then combustion air flow calculation was set forth to check mass balance. To check energy balance, some imaginary heat exchangers are defined at critical locations for which there should be no net heat loss/gain. Based on these restrictions, the dataset of the plant over a period of 3 months was corrected. After ensuring data integrity, efficiency values of multiple boilers in production are analyzed. These efficiency values can be used reliably in an overall optimization scheme to optimize equipment usage.
Introduction
As is the case with many plants petrochemical plants use steam boilers to produce energy in addition to electrical power which is purchased from electrical power company. In this study, İzmit TÜPRAŞ refinery is taken as basis. In power production plants, boilers, steam turbines and gas turbines are the main equipment to produce steam and obtain electricity. In boilers, steam is produced using the energy released by combustion of fuel coming from the production line. These produced steams at relatively high pressures are also reduced to lower pressures, to obtain electricity in steam turbine. In gas turbine, electricity is produced by compressed air using energy released from fuel combustion. Moreover, the high temperature of air is used to produce steam from water.
In this study, the aim is to comprehensively model a steam boiler with a MATLAB based code, confirmed by ASPEN as a reference. By modeling the process in MATLAB, the overall efficiency of the system can be tracked easily. When the process starts to run regularly and systematically, not only the bad data (outliers of the system) can be eliminated but also missing data can be estimated in the real system.
Free Full Text Source: http://etd.lib.metu.edu.tr/upload/12616933/index.pdf