As the API 754 RP states, process safety incidents should be categorized in to four different key performance indicators according to the recommended practice definitions analyzed in this article. To normalize the data and allow comparison between business units and operations, the PSI rate is calculated based on the total worked hours (employees and contractors). The classification of the process incidents is based on the Loss of primary containment quantification and comparison to the API 754 thresholds of LOPC and its consequence on people, environment or property.

Through this document statistics and examples are shown, supporting how the implementation of API 754 safety performances is been improved for a Colombia oil field operation, having a trend of less Tier 1 and Tier 2 events each year. The API 754 process safety incidents methodology was implemented as part of the process safety program, improving and strengthening it.

From the recommended practice, Tier 1 and 2 correspond to lagging indicators and Tier 3 and 4 are leading indicators. The objective of a mature process safety program is to have a low number of Tier 1 and 2 incidents and a larger number of Tier 3 and 4 events that can be managed and resolved, this strategy allows to develop a robust barrier program. In addition, PSI performance indicators are now included as annual KPIs and analyzing the incident investigations, allows to stablish year-to-year action plans. Additionally having the statistics with normalized and uniform criteria let the managers compare facilities and business units, having a better general view about process safety and allowing them to have better data based decision-making.

The API 754 allows the standardization of process incident classification across operations. This methodology seeks to minimize near misses than can lead to a significant incident. It allow us to detect common failures, generate action plans, and measure our performance and define annual goals.

You can access this article if you purchase or spend a download.