Boost availability and reduce maintenance costs with field instrumentation data
The oil & gas industry operates in tough environments with challenging KPIs and costly maintenance when things go wrong. Plant managers and engineers facing increasing complexity in their systems and constant pressure to reduce operating costs, improve reliability (uptime) and increase process efficiency. You need better visibility of the condition of your process equipment to avoid costly unscheduled maintenance and reduce downtime
According to Kimberlite research, just 3.65 days of unplanned downtime a year can cost an oil and gas company $5.037 million. An average offshore platform experiences about 27 days of unplanned downtime a year, which can amount to $38 million in losses. There are also cases where the losses are as high as $88 million.
Also, many operations are dangerous, remote locations that aren’t healthy and safe for employees. According to the Centers for Disease Control (CDC), between January 2015 and January 2017 oil and gas extraction workers were involved in 602 incidents, 481 hospitalizations, and 166 amputations.
Despite the large financial impact unplanned downtime can inflict, few oil and gas organizations utilize optimized maintenance strategies. Three out of four organizations take either a time-based or reactive approach to maintenance. Less than 24 percent report their maintenance strategy as predictive and focused on data or analytics. In the Kimberlite study, 42% of the offshore plants were over 15 years old. Data is the new oil According to the U.S. Department of Energy, predictive maintenance saves 8% to 12% over preventive maintenance and upward to 40% over reactive maintenance. A study by McKinsey & Co. found that predictive maintenance could reduce unplanned downtime by 30% to 50% and also increase the life of the machine by 20% to 40%.
How does it work?
In a nutshell, all analyzers or other i/o devices generate data for years, but we don’t use it only the process data. The other data like the historical, diagnostic, and operational state is very useful to predict when an instrument will fail. The simplified process looks as follows:
Step 1:Collecting real-time data predictive maintenance starts with collecting the data from types of equipment potential failure points (e.g., process analyzers, fire, detectors, monitoring systems) with the help of sensors. It’s good to have a data set that illustrates the health of all the equipment in the field and shows identifiable failures. Operators can use this data set as the base for creating predictive models.
Step 2:Adding context for better reliability and accuracy of future predictive models, real-time data is combined with equipment metadata, equipment usage history, maintenance data, diagnostics, and operational state. This data can be fetched from ERP (enterprise resource planning, EAM (enterprise asset management), EMS (enterprise management system), and other enterprise systems.
Step 3:Searching for patterns operators examine the combined data set and context data to identify dependencies and make technical assumptions regarding the potential failure signals and usage patterns leading to failures.
Step 4:Creating predictive models The essence of the stage boils down to running the combined data set through machine learning algorithms to identify equipment failure patterns and, based on them, build predictive models. After all the tests they know that all the data is accurate, representative, and reliable.
Oil & Gas, Petrochemical, and Renewable organizations should move their maintenance approach from reactive to predictive, as it’s shown to reduce unplanned downtime by 20% to 40%. With only a 1% rate of unplanned downtime costing more than $5 million annually, even small changes to optimize maintenance can materialize into large savings for these organizations. This does not even include other savings such as sampling, reduction of man-hours, and perhaps the most important the safety of all the employees.
Book a demo with one of our experts if you want more personalized advice on what is the best way to implement a predictive maintenance strategy in your organization.