Predictive MaintenancePredictive Maintenance (PdM)

Predictive Maintenance
Skip Navigation LinksITR Home > Predictive Maintenance

PdM is the practice of using asset diagnostics and prognostics to facilitate condition-based preventive maintenance planning. This includes using analyzed data from predictive technologies to identify potential failures modes before they occur or adversely affect an asset and taking action to avoid the failure mode. In a world of ever-shrinking resources, doing the right work at the right time has never been more important. While avoiding all unanticipated machine failures is not yet possible, asset analytics using predictive technologies has shifted modern maintenance practices forever from reactive to proactive

Successful PdM processes follow the common PDCA cycle:

Predictive Maintenance Cycle

This PdM process specification highlights the critical controls and resources ITR has identified over 30+ years that determine whether or not a PdM process is effective.

Process analysis

Like any key business process, the PdM process must be analyzed to determine measurable objectives and critical resources, controls, and inputs. Many organizations focus on the predictive technologies but overlook how predictive information becomes effective preventive and corrective action using limited resources. Process management begins with process analysis to identify the needs of stakeholders, resources necessary to support the process, controls necessary for ensuring fulfillment of requirements, and materials and information that drive process. The PdM process is no different.

Asset identification and criticality and risk assessment

PdM is simply risk management. Risk management begins with identifying critical items at risk, severity of failures, likelihood of occurrence, and ability to stop the failure from occurring. Critical assets are identified and prioritized by criticality towards fulfilling measureable objectives and risk of failure. Reliability professionals manage risk by applying limited resources according to risk priority.

Asset and technology alignment

Understanding critical asset potential failure modes enable reliability professionals to align the right predictive technology to each critical asset. Alignment also includes identifying the appropriate test methods and frequencies for each critical asset. Limited resources mean successful asset and technology alignment often determines whether or not a PdM process fulfills its objectives. Redundant or ineffectual applications of predictive technologies results in inefficiencies and wasted resources.

Database development

For acquired data, analyzed data, and reported information to be useful over time, reliability professions must setup their databases with this goal prior to beginning work. Database setup often includes establishing test parameters and considerations for tracking and classifying potential failure mode severity, asset component performance, analytical effectiveness, and state and effectiveness of preventive and corrective actions.

Scheduling

For the use of predictive technologies to be truly predictive maintenance, data analysis and information reporting must occur more frequently than monitored potential failures typically occur. Using the risk assessment results, reliability professionals plan testing intervals accordingly. Complex, high value, high risk assets often require frequent or continuous monitoring. Conversely, simple, lower risk critical assets often require less frequent monitoring to make best use of resources.

Data acquisition

A common expression is “garbage in, garbage out”. When acquiring data, reliability professionals use data collection methods that ensure both repeatable and reproducible data. Accuracy and precision of data are very important. Consistency of data of over time is essential.

Data processing

Acquired data may be immediately ready for analysis, or it may require conditioning. Rich data sets, such as waveform data, usually require signal conversion from analogue to digital, filtering, and data compression before they are in a form the reliability professional can analyze. Careful concern for data processing and the needs of the analyst are critical to an effective PdM process.

Exception notifications

When working with large data sets, particularly for continuous waveform data, reliability professionals identify indicators within the data to initiate planned actions. Sometimes the data drives hard controls, such as asset shutdown, independent of people. In most cases, algorithms drive the notification of competent individuals to perform detailed data analysis to determine if further action is required. A successful condition monitoring process ensures data analysis always occurs when needed and without wasted effort.

Scalar data analysis

In asset analytics, quantities expressed as a single number are scalar. Advantages of these data include simplicity and the possibility of discrete control limits. Conversely, individual scalar data usually does not provide insight into underlying meaning. For example, if a damaged electrical connection has increased to 200 degrees F, the reliability professional may know to initiate action to correct the problem but no additional information is possible regarding root cause without further investigation.

Waveform data analysis

Analog data is often referred to as waveform data since it is collected as sine waves. The analog data may be processed as an analog signal or converted to a digital signal for micro-processing or both. Waveform data provide reliability professionals with a richer data set than scalar data. Waveform data often describe not only asset condition but also provide insight into the sources of the condition. Challenges related to waveform data include handling large volumes and parsing meaning from their complexity. Reliability professionals balance the use of scalar and waveform data when designing the PdM process to ensure all critical potential failure modes are identifiable and actionable.

Statistical analysis and trending

Monitored asset operating parameters are often referred to as features. Feature data is analyzed over time to identify potential failure modes, potential root causes, and their severity. Absolute data is useful, but reliability professionals learn more from statistical data analysis and trending showing how feature data changes over time. Reliability professionals account for statistical analysis when identifying asset features and planning their analysis processes.

Pattern recognition

Whether expert systems or reliability professionals (or some combination) are performing data analysis, the goal is to recognize patterns within data and make it meaningful and actionable. As assets increase in complexity, pattern recognition becomes more difficult and organizations lean more from humans than computers. This may seem counterintuitive, but as pattern recognition becomes more complex, the cost of complex algorithm development versus the risk of missed failure modes or false alarms becomes prohibitive. The most effective analytical processes find the optimal balance between automated expert systems and highly skilled engineers performing analysis.
Asset Analysis Complexity

Reporting standards

Data analysis is only as good as the associated information reporting, and effective information reporting relies upon established reporting standards. Even automated reporting eventually results in people receiving actionable information. Actionable information is clear, concise, correct, and suitable for the intended recipient. Particularly in PdM processes with many stakeholders, reliability professionals design and establish reporting standards that ensure PdM information is both understandable and actionable by the people who need the information.

Communication processes

Communication technology has improved dramatically over the past 20 years making many new communication methods possible. But simple communication process failures still occur, and despite advanced monitoring, analysis, and reporting capabilities, PdM processes may still fail. Reliability professionals ensure communication processes are established to ensure the right people have the right information at the right time. And since organizations continually evolve, communication process management is an ongoing task for the modern reliability professional.

Information management

Some by-products of the PdM process are (a) effectiveness and outcomes related to the use of predictive technologies, (b) records of asset and component condition and performance over time, (c) maintenance actions taken and their effectiveness, and (d) asset failures (and avoidance of failures). A well-designed PdM process anticipates needing these records for analysis and implementing information management processes to ensure their availability. This makes it possible for reliability professionals to use these records to drive proactive maintenance actions and continual PdM process improvement.

CMMS integration

Communication and information management processes often involve computerized maintenance management system (CMMS) integration. For many organizations, CMMSs are integral to maintenance and reliability work. Reliability professionals plan and ensure effective CMMS and PdM technology integration so organizations make use of existing processes instead of creating competing or conflicting ones.

Avoiding potential failure modes

Most asset failure modes occur nonlinearly, so despite advanced analytical tools, predicting precisely when a failure will occur is usually impossible. Reliability professionals design PdM processes to provide as early warning as possible so managers have adequate information for making decisions and taking preventive action. Avoiding potential failure modes is a balancing act between maximizing an asset’s life cycle and minimizing the costs associated with repair and downtime. By taking condition-based action on minor issues, organizations usually save much more over time.

Removing actual failure modes

Sometimes partial or complete component failures occur before catastrophic asset failure. In these cases, predictive technologies are still useful for providing additional information regarding root cause of the failures. Reliability professionals ensure the PdM process uses this information to ensure root causes are removed when taking corrective action.

Verifying and recording effectiveness

The final step in the core PdM process is verifying and recording the effectiveness of preventive and corrective actions initiated by condition monitoring. Closing out planned work is certainly critical to process success. But additionally, complete and easily accessible recordkeeping allows reliability professionals to monitor overall process effectiveness, trend process performance data, and identify opportunities to improve maintenance and reliability practices. This drives improved PdM process planning. A popular business process axiom states: “only that which is measured is improved”.