CMMS packages today do an excellent job of collecting, sorting, filtering, summarizing, and even trending data, using standard reports or internal/external visual analytical tools (e.g., dashboards). For instance, equipment performance reports can be derived from condition-based maintenance data. However, CMMS packages are just beginning to scratch the surface in the analytics arena. Few packages, for example, provide sophisticated cost/benefit analysis and modeling of equipment repair/replace decisions, in light of production volumes, vendor and model differences, equipment reliability statistics, and other such variables.
Real maturation of the CMMS industry will come when built-in analysis tools such as machine learning and artificial intelligence are developed to assist maintenance and, more importantly, operations by clearly presenting productivity improvement opportunities to managers and frontline workers.
Collecting and analyzing equipment history data
A successful CMMS implementation produces savings and benefits that stem ultimately from the analysis of equipment history data. Typically, equipment history data is used to report on actual versus plan for labor, material, and other costs. More advanced features to look for include:
- tracking maintenance costs by user-defined statistics (e.g., cost per volume produced)
- equipment status tracking and analysis
- problem, cause, action, and delay code analysis
- analysis of static information (e.g., tombstone data)
- analysis of production versus machine downtime
- determination of mean-time-between-failure (MTBF)and other reliability-related analytics,
- drill-down on summary reports to determine the root cause of downtime
- tracking or even optimization of the ratio of fail-based : use-based : condition-based maintenance, based on cost/benefit.
It is not fair to solely blame the vendors for the dearth of highly advanced analysis tools, as users are not yet demanding sophistication in this area. Many operations managers are quite content to leave the analysis of all asset-related data to maintenance management. In turn, the typical maintenance department is satisfied with being able to pump out reports that summarize actual versus planned hours and expenditures. Although these reports are important, there is a general lack of understanding as to the full potential of a CMMS, and just how powerful a tool it can be.
This article is part of our monthly Asset Manager column. Read more from David Berger.
Additionally, few packages specifically provide analytics for the maintenance worker. CMMS vendors offer mobile solutions primarily for downloading work orders, accessing asset history, and documenting the results of work completed. When the frontline analytics gap is addressed, and companies better understand the maintainer’s potential role in fully exploiting the system, these packages will have a far greater impact on maximization of equipment uptime and maintainer utilization. For example, if maintainers were encouraged to access component history, diagnostic data, and analytical tools prior to and/or during the servicing of a piece of equipment, they may more quickly and effectively address the root cause of the problem.
The next-generation analysis tools
There are many reasons why operations and maintenance management are not using the sophisticated analysis tools mentioned above. One reason might be that they do not trust the accuracy of the data collected, which in turn, renders the analytics suspect. A second reason may be that too much work is required to understand and work with the analysis tool. A third reason is that there are so many choices—what data to collect when, what models to use, what parameters are required, and so on.
Some of these problems may go away with the advent of new technology and new tools. The problem of accurate data may be resolved by eliminating the human through condition monitoring (i.e., using electronic devices to collect data on the condition of equipment and surrounding environment). As well, artificial intelligence and machine learning may address the problem of too much work and too many choices in using CMMS analysis tools.
Most analysis tools today use statistics to make predictions about the future based on rules governing the past. These rules are already defined, such as the MTBF algorithm, or are induced by the software as with simple condition-based monitoring software.
Analysis tools that use machine learning and artificial intelligence are far more powerful and easier to use than traditional means. These tools look at a large database and find patterns and relationships between attributes. The computer then determines how a given pattern leads to a predictable outcome.
This is especially useful in enhancing reliability management systems, one of the most important uses of equipment history data. Artificial intelligence can be used to forecast how best to repair equipment by matching attribute patterns of problem, cause, delay, and action codes for a given equipment class. These codes, how best to use them, and the benefits of doing so are explained below.
Getting the most out of reliability management data
Reliability management attempts to minimize equipment downtime through analysis of failure data. The three basic steps involved are described below.
1. Define and codify failures
Problem (symptom) code—When an employee from any department requests work from the maintenance department, the presenting problem that needs to be solved must be classified by choosing from a list of problem or symptom codes.
Cause (failure) code—When the maintenance worker goes out to fix the problem, the root cause is determined. It is important to isolate which component or part has failed. Knowing to what level of detail to go is difficult. Use simple cost/benefit analysis to help decide. If this is an infrequent failure, inexpensive part, or quick and inexpensive fix, then there is no value in drilling down to a greater level of detail.
Action code—Once the cause of the problem is determined, the maintainer will take action which can be codified as well.
Delay code—It is important to understand the nature and cause of delays to operations. Some of the major reasons for delays are setups, changeovers, machine adjustments, lack of parts, lack of labor, machine breakdown, and operator breaks.
2. Build the data warehouse. Use automated techniques such as online condition monitoring, or manual means to record the data above. As well, the CMMS stores static data describing the attributes of the equipment itself. Stored or “warehoused” data can then be sliced, diced, analyzed, and modelled.
3. Determine appropriate tasks. Once the failures have been identified, the work program can be evaluated in order to eliminate or at least minimize the frequency and/or impact of failure. This can be done through traditional or sophisticated analysis techniques described above.
Various condition-based maintenance tasks should be explored to prevent a problem from occurring in the first place. Automated tools can also be used for predicting failures in similar parts, components and equipment, once a pattern is determined. This would lead to monitoring the condition of key components that had not yet failed but were deemed likely to do so, in order to catch a problem before it happens.
Mining and analyzing reliability management data
Many people wonder whether all the coding and diagnostic analysis is worth it. For a large company there is no question that the effort pays off, especially when features such as coded fields, condition-monitoring, and sophisticated data mining and analysis tools are seamlessly integrated into the CMMS.
Maintenance, in partnership with operations, must better analyze the equipment history in order to reduce maintenance and operations budgets, improve maintenance service quality, and improve the quality of any output running through the equipment. For many large companies, reliability management is one of those few remaining areas that holds huge, as yet untapped potential for realizing significant gains to bottom-line results.
This story originally appeared in the June 2021 issue of Plant Services. Subscribe to Plant Services here.