The True Danger of Colors: Injury Probability during Turnaround Phases in a Refinery

Introduction Safety incidents during turnaround phases represent 50% of overall incidents occurring at a refinery. This study aims to understand the different phases, the incident occurrence during such phases, and determine where to have more observations happen during such phases (Ludwig & Laske, 2022). Methods Our team took data from the turnaround phases of an oil refinery in between the years of 2022 to 2023. This involved taking data from the plant and each individual subdivision and pulling only the ones that went through a turnaround phase in the timeframe provided. Out of the fifty-four departments that we gathered information on, there were only five that were in an active turnaround. We had to manually parse the data out per turnaround phase in each department and then process it. A logistic regression was then used to pull results and make conclusions based on what we knew before and what we got from the data. Results The analysis from the two-year sample (2022-2023) resulted in a 9.38% incident probability occurring over any seven-day span across the five departments. One additional observation per shift over a 3-day period was shown to reduce incident probability by 1.21%. This was predicted to lower the number of incidents from 25 to 23 per year, resulting in two fewer incidents annually for the five departments studied. When extrapolated to the entire facility, with over 50 departments, one additional observation per shift could prevent at least 20 incidents annually. Discussion With how many incidents occur during turnarounds, we hope that our findings of adding another observation per shift can mean less overall incidents throughout the turnarounds (Granowsky, 2023). We hope this research can inform further studies because we know that we can go deeper into the research on how to reduce incident probability during these phases, and consider increased safety intervention during the turnaround periods (Ludwig, Leslie, 2023). One limitation that I can see is that we did our analysis on a manufacturing plant during a specific period of time, so it may not transfer that well into other environments that may not have a strict cleaning schedule.