The accelerating evolution of condition monitoring tools has made a structured, reliability-centered motor management framework essential for guiding CBM technology decisions. In this article, we provide a comprehensive reliability engineering-level framework for effectively selecting Condition-Based Maintenance (CBM) technologies in modern industrial environments.
Without a structured framework, advanced monitoring becomes noise instead of insight.
With rapid advancements in electrical and mechanical condition-monitoring tools and technologies, such as wired and wireless systems, AI, and rules-based systems, companies face increasing complexity in determining which technologies best support their reliability, safety, and asset management objectives.
This article offers a structured, reliability-centered approach that aligns CBM technology selection with organizational risk, failure behavior, lifecycle cost, and ISO 55000 asset management principles. A multi-technology strategy is emphasized to eliminate blind spots and strengthen diagnostic confidence.
Real-world considerations, financial justification, and integration into maintenance strategy are included. The result is a defensible path to improving reliability, reducing total cost of ownership, and ensuring informed corporate decision-making.
The Gap Between Advanced Technology and Real-World Maintenance Outcomes
Even with the introduction of many advanced technologies, including AI and continuous monitoring systems, the mix of reactive to condition-based maintenance has not changed since surveys in the 1990s (Figure 1).
Figure 1: Existing maintenance practices (FEMP/PNNL 2021 and NIST ‘The Costs and Benefits of Advanced Maintenance in Manufacturing,’ updated Feb 2025)
The opportunity, per the NIST document cited in Figure 1, exceeds $22 billion from the application of additional predictive technologies alone. The opportunities also include reducing energy consumption by 5-20% or more and reducing maintenance costs by 15-98%, with maintenance estimated to be between 15-70% of the cost of goods sold.
Additional impacts include reductions in safety incidents exceeding 50%, such as injury and death, and productivity improvements of 69%. To achieve these goals, the organization is shown in Figure 2.
Figure 2: World-class maintenance targets for manufacturing per FEMP and NIST studies.
To achieve the targets, based on Figure 2, the studies recommend the inclusion of advanced maintenance practices, including traditional predictive maintenance technologies and AI/ML, when applied effectively. However, as with many PdM technologies, AI/ML has been marketed as a one-size-fits-all solution, as shown in Figure 3.
Figure 3: Studies performed in 2024 and 2025 related to the success of AI/ML in maintenance, with over 95% of applications cited as failing.
Independent studies have shown that the application of AI/ML-based technologies has not resulted in the expected return on investment for a significant value of over 95% of applications. Most of the responses appear to be related to the selection of specific technologies, driven by marketing hype and overpromised capabilities.
Additional issues relate to improper placement of technologies, leading to poor results ranging from extremely late detection to false positives and negatives, which increase maintenance costs.
These types of technologies often also include the suggestion of removing the ‘human-in-the-loop’ through AI, which contradicts recommendations from standards such as the IEEE 7000 series (“Standards for Ethically Aligned and Responsible Development of Autonomous and Intelligent Systems”), which require human-in-the-loop.
AI can automate analysis, but it cannot replace engineered human judgment.
A majority of the issues, particularly where CBM is involved, stem from the contradiction between the goals of equity-investor-owned organizations and technologies and the requirements of the systems to be monitored. The next is the excessive amount of information that is gathered on any given system and the resulting ‘noise’ and distractions.
Industrial operations today operate within a landscape of increasing complexity, extended supply-chain pressures, rising production demands, and heightened expectations for safety and regulatory compliance. As assets become more interconnected, failures in one system increasingly propagate across others, creating greater operational and financial impact than ever before.
Why Failure Modes Must Dictate the Technology – Not the Reverse
In this environment, CBM has evolved into a strategic necessity rather than an optional enhancement. Yet the challenges for companies and reliability leaders are no longer simply acquiring CBM technologies; it is selecting the right combinations of technologies that align with business needs, failure modes, and risk exposure.
Effective CBM technology selection begins with a core engineering principle: failure modes, not the technology, must drive maintenance decisions. Reliability-Centered Maintenance (RCM) remains the gold standard for aligning maintenance tasks with actual failure behavior.
RCM requires defining asset functions, identifying functional failures, determining failure modes, and analyzing the consequences of each failure mode. Only after understanding these drivers should technology selection occur.
This ensures that CBM resources are allocated based on engineering logic rather than vendor influence, convenience, or trend adoption. For the C-suite, using RCM as the decision filter transforms CBM selection from a technology experiment into a strategic reliability investment.
The whole process is outlined, based upon “Physical Asset for the Executive,” 2008, Penrose, as shown in Figure 4. This process requires identifying assets through an asset census, then determining their criticality, which we will discuss before the seven steps of RCM.
Figure 4: A complete CBM/Maintenance practice outline (Penrose, 2008, ‘Physical Asset Management for the Executive’)
Asset criticality defines the required depth and frequency of monitoring. Not all assets require the same level of maintenance sophistication, and deploying advanced technologies on low-criticality equipment can waste capital and labor. The selection of technologies, including continuous or periodic monitoring, can then be determined based on logical assessment. The risk assessment process can be as simple as Table 1 and applied to Figure 5.
Table 1: Risk Assessment
Figure 5: Criticality Scorecard
For Table 1 and Figure 5, the assessment is based on Risk, calculated as Pf * Sf, where P is the probability, and S is the severity of the failure. Given the risk, it is important to adopt the mindset that not all failures can or should be prevented, and that risk management is good maintenance.
The process also requires that Functional Failure is when a condition results in the device, or system, not performing to expectations or intended functions. A well-designed maintenance program uses the concept of risk to:
- Assess the risk of functional failure on individual systems and equipment levels.
- Develop tasks to prevent failure based on that risk assessment.
- Allocate resources where they provide the most significant benefit.
The purpose of the maintenance program is to ensure that the proper maintenance is performed on the right equipment at the right time and for the right reasons. This means that maintenance is performed to ensure that the equipment or system operates in line with the expectations or needs of the application.
So, for instance, if a pump is capable of pumping 1,000 gallons per minute (gpm), but only requires 500 gpm, maintenance must be performed to 500 gpm, not excessive maintenance to meet the design.
Moving from Criticality to RCM: A Structured Seven-Step Process
Once we have the equipment’s criticality determined, we need to proceed through the RCM process. For our purposes, we will describe the basic seven steps, but recommend any number of RCM documentation and training to achieve a complete understanding. For technology, we will identify common technologies and explain how to determine the frequency of performance, whether it is periodic or continuous.
Figure 6: The seven steps of classical RCM (“Reliability-Centered Maintenance,” 1978, Nowlan and Heap)
The RCM question process looks like this:
- What are the functions and associated desired standards of performance of the asset in its present operating context (functions)?
- In what ways can it fail to fulfill its functions (functional failures)?
- What causes each functional failure (failure modes)?
- What happens when each failure occurs (failure effects)?
- In what way does each failure matter (failure consequences)?
- What should be done to predict or prevent each failure (proactive tasks and task intervals)?
- What should be done if a suitable proactive task cannot be found (default action)?
The first step in RCM is to select the system and components to be analyzed. This is usually limited to inputs and outputs from the blocked selection. For instance, with a pump system such as Figure 7, we may have input requirements of 100 kw at 460 volts(+/-10%), and ~150 Amps, with expected outputs of 1790 RPM to accomplish 500 gpm from the pump.
We would then sort through all of the potential functional failures, which, in this case, would be ‘not providing 500 gpm.’ The failure modes would then be a list of conditions that would impact the system in a way that we could not provide the 500 gpm required. We then list, for each failure mode, the effects.
These effects would be local, where the failure mode occurs, at the subsystem or system, and the end effect. For instance, a failed bearing would be: local – increased temperature/friction; increased power requirement; and a reduction in speed resulting in loss of output from the pump.
Figure 7: P-F Chart (RCM, Nowlan & Heap, 1978)
The P-F chart, shown in Figure 7, contrary to a number of published articles, is solely intended to determine the frequency of testing based on the technologies, training, and systems available. This is also balanced against costs, in which you would see all the applications and then determine the optimal approach.
Balancing Detection Windows, Testing Frequency, and Total CBM Cost
For instance, if we are using vibration analysis in this scenario, on a critical system, and we can detect a particular bearing defect 12 weeks before estimated functional failure, we would have to check half the time between the point of detection (P) and functional failure (F), or 6 weeks. As the risk increases, we may adjust the ½ P-F to evaluate at 1/3, ¼, etc. The related maintenance, or CBM, costs (total) can then be balanced, as well, to determine the optimized approach. These costs would include:
- Technology/equipment
- Training
- Personnel
- Fees/Costs for such things as SaaS
- Other fees and expenses
This would then be balanced against the costs and priorities of the criticality analysis. For instance:
- Safety, which is defined as human safety only; there are usually no limits to cost, but one should consider the associated safety risk assessment.
- Regulatory/Law, which would be such items as emissions, etc., in which the balance of losses would be considered.
- Production/Mission, in which the related production losses (total) are balanced against the cost to maintain.
- Other, which includes other conditions, which don’t have to include cost effectiveness, but can also include reputation, management requirements, etc.
As we consider the above, some of these maintenance aspects may include inspections, lubrication, and other tasks. However, in this article, we focus on CBM/PdM technologies.
In Table 2, we identify the capabilities of some technologies (not exhaustive) for random wound electric motors, followed by Table 3, which identifies some estimated management considerations. These should be balanced against the actual considerations based upon technology and consensus standards.
Table 2: Motor System Diagnostic Technology Comparison (pump application)
Table 3: Management Considerations
Some of the definitions for Tables 2 and 3 include:
- PQ: Power quality, including harmonics.
- Cntrl: Control faults including contacts, loose connections, VFDs, soft starts.
- Conn: Loose or open connections.
- Cable: Shorted or degraded cable, including insulation breakdown.
- Stator: Winding faults, including shorts.
- Rotor: Open rotor bars and related defects.
- Air Gap: Eccentricity, static and dynamic, mixed.
- Brgs: Bearing degradation.
- Ins: Insulation to ground conditions.
- Vibe: Faults that impact machine and/or structural vibration.
- Align: Misalignment, soft foot, other.
- Cplg: Coupling defects.
- Pmp Brg: Pump bearing degradation.
- Imp: Impeller conditions, including cavitation.
- Seal: Seal defects and wear.
Determining when something is detected depends on several factors, including the type of technology and manufacturer, location, sensitivity, and other considerations.
If we look at an example of a critical pump that has implications for production, and we set the block to include the transformer, MCC bucket, motor, coupling, and pump, with the transformer taking 12,460 Vac, 3-phase, and converting it to 480 Volts. The motor is a 100 kW, 1785 RPM, 460 V, 150 Amp machine, and the pump is a 1,000 gpm machine, direct-coupled to the motor (Figure 8).
Figure 8: Motor and pump application example. The abilities of ESA and MCSA are included for demonstration.
For brevity, we assume we have performed the process and obtained the results shown in Table 4, which summarizes the available technologies.
Table 4: Evaluation Example
When reviewing the results and comparing them to the equipment’s criticality, personnel availability, and related costs. We find that an applied ESA continuous monitoring system costs $7k plus installation ($1,000), and infrared runs $500, for a total of $8,500. For the selected ESA technology, several other systems are available, dropping the cost to ~$3k for a cost of $3,500 for the tech.
Training for personnel runs ~$12k for everything and, when applied, results in a much lower cost. Maintenance on the technology drops it to an estimate of $4k investment and annual costs of ~$2,500 for monitoring, not including verification.
Turning Insights Into Action: Closing the Loop on Repetitive Failures
When a planned RCM-based process is followed, then we can keep our associated costs down while decreasing the risk of unplanned failures. It must be remembered that monitoring does not correct problems, especially repetitive issues.
Included with monitoring must be the use of the applied information to eliminate repetitive issues through corrective actions or re-engineering. Periodically, the RCM process should be reviewed, which can be kicked off either by calendar or if some KPI or trigger occurs, including replacing technology, monitoring, or inspections, also based upon technology changes.
