How manufacturers can improve equipment performance by fixing the data, work systems, and operating discipline around maintenance
Maintenance and reliability programs often become more complicated than the assets they are supposed to protect. Companies collect more data, add more metrics, and invest in new technology while technicians still spend much of their day reacting to failures.
In this episode of MetaPod, Ron Crabtree speaks with reliability engineer Russ Parrish about building a practical maintenance program in the digital age. Parrish has worked in nuclear power, heavy manufacturing, operations, maintenance, reliability, and corporate real estate.
His central point is simple. Start with an honest baseline, reduce the noise in the data, focus on the assets that matter most, and introduce AI only after the organization defines the problem and the outcome.
Know where the operation stands
Most reliability efforts begin after a problem has already become expensive. Leaders need to understand current performance before they choose technology or set a target.
Parrish recommends reviewing the available information, walking the facility, and comparing the organization's perception with what the evidence shows. A plant may consider itself partly proactive while its work history shows that most maintenance remains reactive.
In one example, a manufacturer believed its work was roughly balanced between proactive and reactive maintenance. The analysis showed about 25 percent proactive work and 75 percent reactive work. After more than two years of focused improvement, the mix moved to about 60 percent proactive and 40 percent reactive.
Measure less and use more
One company Parrish supported had metrics for almost everything. Different locations used different definitions, and the volume of information made decisions harder rather than easier.
The team reduced the active measures to roughly one quarter of the original set, standardized the remaining data, and built a roadmap with six, nine, and 12 month milestones inside a three to four year plan. After about 18 months, the organization had increased output by an estimated 25 to 30 percent at the plant Parrish described.
The lesson is not to avoid data. It is to collect information that the team intends to use. A metric should identify a gap, track an improvement, or support a decision.
Standardize the definition of OEE
Overall equipment effectiveness combines availability, performance rate, and first time quality. The calculation is useful only when every site applies the same definitions.
Downtime is a common source of disagreement. Teams may move events into planned downtime to protect a local score. Parrish recommends starting with a strict baseline. If the plant is down, count it. Definitions can become more refined after leaders establish control and stop departments from arguing over ownership.
The manufacturer in Parrish's example began below 50 percent OEE and targeted 80 to 85 percent. That goal forced operations and maintenance to work on the same business result instead of defending separate budgets and metrics.
Fix the system around the maintenance work
Poor reliability is rarely caused by one technical issue. In Parrish's example, the planner and scheduler spent little time planning and scheduling. The storeroom held many parts, but employees struggled to find the right ones. Obsolete items created search time, and work stayed reactive.
The improvement effort clarified roles, cleaned the storeroom, corrected parts availability, reviewed staffing and qualifications, and connected supporting production lines to the same facility goals. These changes gave technicians a better chance of completing planned work when the equipment was available.
Prioritize critical assets
A full reliability centered maintenance analysis can identify every way a component may fail, but the work is expensive. Applying it to an entire plant can consume time without producing a proportionate return.
Start with asset criticality. Identify equipment whose failure would stop production, create a safety risk, damage quality, or impose a high replacement cost. Apply the deepest analysis to that small group and connect the findings to the capital replacement plan.
Design for reliability offers the strongest opportunity because the organization can still influence equipment selection, access, lubrication, spare parts, and maintainability before the asset enters production. Once the machine is installed, many choices become harder and more expensive to change.
Set the AI playing field before buying tools
AI can help triage work requests, improve planning and scheduling, support root cause analysis, search maintenance history, and guide troubleshooting. Those uses depend on good information and clear expectations.
Begin by defining what success looks like and who inside the organization understands the technology well enough to challenge it. Build milestones that can produce an early win, then collect enough history to show that the result is repeatable rather than an anomaly.
AI can also capture the path an experienced technician follows during troubleshooting. That knowledge can become training for the person working the overnight shift who has not seen the same failure before. The tool supports the workforce, but it does not remove the need to build technical skill.
Discipline matters more than novelty
Parrish's advice to leaders is not primarily technical. Processes and procedures only help when people have the discipline to keep using them after the project team leaves.
Invest in employee training and certification. Bring maintenance, operations, safety, procurement, HR, and engineering into the same reliability program. Give technicians and operators access to the same trustworthy information at the point of work.
The technology will keep changing. The fundamentals will not. Reliable assets depend on useful data, clear work practices, capable people, and management attention sustained over several years.
Connect and learn more
A common sense reliability program starts with the plant as it is, not the plant leaders wish they had. The organization can then improve the ending through better data, proactive work, asset priorities, and steady follow-through.
To learn more, connect with Russ Parrish on LinkedIn or through Russpec Consulting. You can also connect with Ron Crabtree on LinkedIn or contact the MetaPod team at MetaExperts.com.