(41j) Management of Change: Perhaps the Most Challenging Aspect of Process Safety
AIChE Spring Meeting and Global Congress on Process Safety
2023
2023 Spring Meeting and 19th Global Congress on Process Safety
Global Congress on Process Safety
GCPS Alternate Presentations
If weâre honest with ourselves, Process Safety has a lack of data problem. Nowhere does this show up more than in the types of calculations we perform for Layer of Protection Analysis (LOPA) and Safety Integrity Level (SIL) calculations, for example. Sure, we have generic failure data. But do we have the confidence that this generic data is right for our specific application? In addition, many LOPA scenarios contain âone-offâ equipment parameters (either initiating event frequency or probability of failure) for which there is no generic data, leaving teams guessing at what value to use. Worse, LOPA targets are getting smaller (i.e., 1e-5 or 1e-6 per yr) which often leaves gaps, requiring decisions to be made regarding capital spending. Sticking with generic data in these cases can leave us feeling that we are being too conservative. Do we want to be correct or be conservative? On the Operations and Maintenance side of the LOPA equation, we face similar problems when attempting to verify the installed performance of an IPL (Independent Protection Layer). A multitude of assumed parameters (e.g., failure rates, test and inspection intervals, time in bypass, etc.) for which we would like a method to incorporate actual site data into the values used during design. And ideally this method could optimize these parameters for potential cost savings (for example, extending maintenance intervals).
This paper will present a straightforward and easy to use method for feeding operational data back into process safety calculations, using commercial software that is already running on your computer. The paper will explore how much data is needed to confidently claim a parameter value, starting with an assumed or generic value, and periodically updating that value with small data, as evidence (from testing, maintenance, actual demands, etc.) is collected over time. The authors have been using these methods successfully on real process safety applications for several years now, that were all triggered by difficulties and shortcomings in LOPA. These application case studies will be discussed as well.