Vendor representatives and industry consultants are never short on suggestions for improving asset maintenance and reliability (M&R), but it’s the end users who decide which ones actually make the grade. Plant Services invited several women leading M&R in their organization to describe the tools and processes they’re using to improve reliability, performance, and cost management.
Condition prediction technologies
Data has always been and will remain the M&R engineer’s most valuable tool, observes Maretha Price, reliability team lead for Sasol Synfuels in South Africa, because “the more information you have, the better you are equipped to rectify, improve, and optimize.” The use of mobile devices makes it easier for any person, regardless of job level, time of day, or discipline to record and digitally transmit real-time information instantaneously to data lakes, she notes.
“Thanks to this technology, more information is available that translates to better data, and better data guides optimal decision-making to manage risk, safety, and cost in a proactive and efficient manner,” explains Price.
Heavily regulated industries such as biopharmaceutical manufacturing have a heightened focus on quality, safety, and reliability. “We implemented predictive maintenance technology (vibration analysis, thermography, and ultrasonic) in early 2018 and have already seen huge paybacks from this investment,” says Shannon Ostendorff, senior manager of maintenance at Lonza – Bend. “We have found misaligned shafts, N2 leaks, and have detected bearing failures on critical equipment prior to catastrophic failure.”
For corrugated packaging company WestRock, utilizing airborne ultrasound has improved electrical reliability, says condition monitoring engineer Charlee Lipham. “Electrical faults such as corona, tracking, and arcing have successfully been detected and repaired on electrical assets,” she notes.
Central Arizona Project (CAP) is using new technology for its canal and pipeline inspections, including mobile tablets with GIS-based applications to collect and analyze data. “This new technology improves the efficiency and accuracy of the assessment significantly,” observes Jennifer Jia, senior civil reliability engineer at Central Arizona Project. It has allowed reliability/maintenance engineers to perform condition assessments in a digital format where all data are georeferenced to the asset locations. The data can also be accessed and shared easily.
For Amazon, the reliability of fulfillment center equipment is vital. Gail Edgar, director of North America reliability and maintenance engineering at Amazon, says her team is implementing several predictive strategies including heat detection, bed alignment, belt stress detection, oil analysis, and temperature and vibration sensory systems to help fulfill a mission to be “Earth’s most customer-centric reliability and maintenance engineering service provider.”
These technologies guide the technicians to the failure modes in a timely manner, allowing enough time to troubleshoot and fix an issue before it has an operational impact, says Edgar. “Currently, we are building an ecosystem that will bring all these predictive technology data points together onto one platform for a comprehensive interface for our technicians and to support data-driven decisions based on correlation.”
Core foundational strategies
Senior maintenance planner Gloria Alorchie-Apetor at Newmont Ghana Gold says Newmont Ahafo relies considerably on root-cause analysis (RCA) to drive continuous improvement and reliability in the plant. “We have efficiently narrowed down downtime factors within our control to achieve average availability of 92%, and we are gravitating toward failure mode and effects analysis (FMEA) using an availability workbench tool, which will help optimize our maintenance strategies and maintenance spares and further decrease downtime factors within our control,” she says.
Owens Corning is working to improve overall performance with total productive maintenance (TPM), says reliability project manager Gina Kittle. A strong foundation is required to support TPM’s pillars, such as planned maintenance, autonomous maintenance, and focused improvement. Key to this foundation is a good master equipment list with the appropriate hierarchy as well as robust reliability work processes. It drives granularity in the data to drill into the root cause of the problem, she explains.
“Once we know these problem areas, we can then determine the best tool to use in order to improve reliability, whether it be a predictive technology, an online system, or a design change,” says Kittle. Owens Corning presently employs predictive technologies in varying stages at multiple plants, helping to catch failures early in the P-F curve and enable precision maintenance. Vibration and ultrasonics are two of the most widespread technologies applied.
Preventing parlysis by analysis
Because the prevalence of condition monitoring technologies is creating an overabundance of data from which to make decisions, being able to turn this data as fast and reliably as possible into information for prioritization of corrective actions and decision making is essential. It is here where machine learning (ML) and artificial intelligence (AI) add value, says Rosana Reyes, asset performance management lead engineer at GE Renewable Energy.
“All the acquired data from condition monitoring, asset performance, environmental, process, and many other operational conditions can be correlated to past failure and survival data and, using mathematical models, forecast the probability of failures of equipment,” explains Reyes. The models can further help by determining the probability of success of maintenance and corrective actions and adjust over time to improve accuracy.
“In my opinion, ML will not substitute the job of condition monitoring experts in the short or medium term, but it will (empower) them with higher certainty on proposed corrective and maintenance actions when trying to obtain buy-in from stakeholders and key decision-makers,” adds Reyes.