Predictive maintenance is more possible than ever before using a combination of IoT, cloud, and analytic technology to monitor machine conditions, guide the maintenance schedule and reduce downtime, an engineering expert told Petrochemical Update.

×

Historically, predictive maintenance was just an educated guess in many plants and resulted in reactive maintenance instead of the proactive approach. Today, the guesswork is taken out of predictive maintenance using data, analysis and digitalization. 

Digitalization

Digitalization includes systems, software and algorithms that convert data into valuable insight, enabling optimal plant operations.

Avoiding failures that impact production is time-sensitive and critical, and yet there were 2,200 unscheduled shutdowns in the U.S. alone between 2009 and 2013, an average of 1.3 incidents a day, costing roughly $20 billion per year, according to Deloitte Insights.

By adopting a digitalization strategy and utilizing artificial intelligence (AI) and machine learning, it is possible to improve plant profitability, according to plant engineers.

Predictive maintenance’s key feature is the use of statistical tools to predict future events by making use of data from the past and present.

The first step to a predictive maintenance strategy is obtaining all the data that will help make decisions.
The next step is to narrow the data down to the most relevant information. Artificial Intelligence (AI) and machine learning have become powerful tools for these tasks.

With tools such as AI and machine learning, an engineer can better understand what goes on in with process equipment. This knowledge enables improved prediction and optimal production levels.

Case study

A BASF plant in Louisiana is exploring technological advances to shift its plants from reactive to predictive maintenance. In order to solve the challenge of capturing all the data from operator rounds, the engineering team will implement electronic rounds and focus on operator recognition of conditions that require attention.

“This will allow our maintenance to react to potential equipment issues before they reach the point of requiring emergency work or cause downtime in the plants,” Amy Odom, Asset Effectiveness Engineer at BASF said. “Since we are a batch operation, it is possible for opportunity downtime between batches or at certain points in the process to do reactive maintenance.”

This strategy is also planned to integrate equipment health evaluation tools, such as infrared scanning, oil analysis and vibration data into a single system so that the “whole picture” is visible to operations, reliability and maintenance.

Real-time devices have become the biggest gains made in the requesting, planning, scheduling, executing and documenting of maintenance, Odom said.

Too many times, maintenance comes to work a job that is already completed, or a job for which all the needed parts aren’t available. Communication is paramount in the maintenance process.

“The use of real time devices, such as tablets, to order plan, schedule, order parts and document work done as it happens is key,” Odom said. “Allowing real-time updates and having the expectation that these updates be done at every step of the process will allow better documentation of failure causes and improve maintenance efficiency.”

Management support

Lack of support from top-level management and failure to follow-through are the most common mistakes that prevent new processes and procedures taking hold within the workforce, Odom said.

“New processes and procedures must be embraced by top level management and the people impacted by the changes must see a behavior change before they commit to changing the way things are done.,” Odom said.

Time is also a factor; if processes and procedures change often, the workforce becomes almost immune and most certainly complacent to changes.

“Many companies change processes and expectations before allowing the last changes to be fully implemented, so they never really determine if the changes are successful. If changes are made “optional,” this is a sure way to prevent them from taking hold,” Odom said. “This is also true if work-arounds and exceptions are allowed for changes.”

New processes and procedures need to also be fully vetted, perhaps through a small trial, before widespread implementation so that any unintended consequences can be identified and rectified early in the implementation phase, Odom said.

“If exceptions are needed, they should be few and be well defined before the company’s wholesale adoption of the changes,” Odom said.

People and processes

No matter how much technology improves a predictive maintenance strategy, strategies will fail without the right people and processes in place, according to Odom.

“Technological advances make data capture easier, but it requires dedicated people and a robust process to sustain,” Odom said. “The goal of predictive maintenance is to improve equipment health and reduce failures. For this process to be successful, it must be robust, and people must believe that the process works so it is sustainable.”

Turnover among crafts remains high in the petrochemical industry, so the more that can be done to automate this process, the more sustainable it can be.

“Setting up a PdM plan in SAP or other system is not enough. The plan needs to be fully executed and periodically evaluated so that improvements can be made,” Odom said.

By Heather Doyle