Pageflex Server [document: D-Aalto-2A28D96F_00001]

Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Kamarizan Kidam Name of the doctoral dissertation Process Safety Enhancement in Chemical Plant Design by Exploiting Accident Knowledge Publisher Aalto University School of Chemical Technology Unit Department of Biotechnology and Chemical Technology Series Aalto University publication series DOCTORAL DISSERTATIONS 168/2012 Field of research Plant Design Manuscript submitted 28 August 2012 Date of the defence 14 December 2012 Permission to publish granted (date) 30 October 2012 Language English Monograph Article dissertation (summary + original articles) Abstract The accident rate in the chemical industry has not been decreasing although they could be prevented by using the existing knowledge. The aim of this thesis is to enhance the utilization of knowledge from earlier accidents especially in the designing of chemical plants. The experience feedback on accidents is improved by analyzing and disseminating knowledge on accident contributors to design activities. The research was done by analyzing the 364 chemical process accident reports available in the Failure Knowledge Database (FKD). It was found that the technical contributors (79%) dominated the accidents in the CPI. Deeper analyses were carried out to identify the accident contributors, and design and operation errors for the six most common equipment types of accidents. The other indicators of accidents included in the study were; the contributors share as main contributor (SMC), equipment specific contributors, and the combination of high SMC and frequency. In design and operation errors analyses, the study found that about 80% of the accident cases were contributed by at least one design error with an average of 2.3 errors per accident. The timing of the errors was analyzed and it shows that about half (47%) of the design and operation errors were made during the process design-oriented stages. Thus, more focus should be given in the making of fundamental decisions such as process conditions, chemicals and reactions during the early phases of the design. The corrective actions proposed in accident reports employed typically the outer layers of protection such as procedural changes (53% of cases) even though the design errors are generally dominant. The inherently safer design proposed was only 18% of cases; and these were based on the most used principles which were ’error tolerance’ and ‘moderate’. Current design oriented safety methods do not fully utilize knowledge from earlier accidents and therefore do not facilitate learning. For example, HAZOP is often employed only as a final check and do not support the designer during the work. Therefore the thesis proposed a method for identification of accident contributors and design errors throughout the design stages by utilizing knowledge from earlier accidents. The method is based on information obtained from accident contributors and design errors discovered which will be presented in the first part of this thesis. The aim is to show also their mechanisms and time of creation. The proposed method would support the design process by having an early design error detection and elimination through design changes. Therefore, cost and safety benefits can be achieved by undergoing changes in the earlier stages of plant design. The Bhopal tragedy is used as the case study to demonstrate and test the method. The proposed method could be used to predict an average of up to 85% of accident contributors.