1. EXAKT is software for performing Reliability Analysis (RA). RA is a general set of software enabled procedures used by Reliability/Maintenance Engineers to develop and deploy rules (called models). These models support, monitor, and systematize day-to-day decision making in maintenance.
2. However, RA, if it is to generate realistic decision models, requires data of a specific type and quality. LRCM is the work order and condition monitoring (CM) based process for ensuring that such data is automatically available for RA. Without LRCM RA tools and software are unusable because the needed data is unavailable. Poor quality age data at the failure mode level is the key obstacle to improved management of day-to-day maintenance.
3. What distinguishes EXAKT from other forms of RA is that it extends the dimensionality of conventional “age based” RA (e.g. Weibull, Pareto, Monte Carlo Simulation) to include (CM) data.
4. When RA accounts for both age and CM data it will support realistic decisions in maintenance. Decisions based on age alone are usually non-optimal leading to maintaining too soon or too late. Good age/CBM models, by systematically including the equipment’s current state in the decision process, enable maintainers to intervene at the right moment and on the right components.
5. It is necessary that the effectiveness of decision rules in maintenance be verifiable and reported so as to be continually improved.
6. Using EXAKT and LRCM the Reliability Engineer (RE) builds and deploys optimized models. An optimized model is one that supports the organizational maintenance objectives, specifically those related to high availability at low cost.
7. Using EXAKT / LRCM procedures and software the RE tracks and improves the effectiveness of those models

© 2011, Murray Wiseman. All rights reserved.

Yet mishaps keep occurring at an alarming rate in large and small organizations. Postmortem investigations invariably demonstrate that avoidable factors precipitated or could have predicted the event. That experienced operating personnel failed to mitigate the disaster, in spite of extensive standards and procedures, strains the organization’s credibility in the mind of the public. Surely, countless incidents should have, by now, provided a reliable model for the prevention of catastrophic events.

James Reason^[1] models the anatomy of industrial accidents as an unfortunate alignment of  organizational influences, unsafe supervision, preconditions for unsafe acts, and the unsafe acts themselves. In this model, an organization’s defenses against failure are a series of barriers, with individual weaknesses in individual parts of the system. Those weak points vary continually in size and position. The system as a whole fails when all individual barrier weaknesses align, permitting “a trajectory of accident opportunity”, so that a hazard passes through all of the holes in all of the defenses, leading to a failure.

Root cause analysis after the fact often exposes a series of foretelling incidents preceding the main event, by minutes, hours, and often by weeks or months. The ubiquitous maintenance department finds itself involved in most such “pre-events”. And these are recorded as work orders in the Maintenance Management System (CMMS). Analysis of the work order database would, seemingly, provide ample opportunity for preemptive action to “plug the holes” as they are discovered. However the typical estrangement between the CMMS and RCM knowledge base usually precludes recognizing opportunities in the existing preventive maintenance strategy. Likewise the conclusions and maintenance changes recorded in accident follow-up software applications seldom find their way back to the RCM knowledge base which loses synchronization with the plan.

ISO 14001:2004 has been conjectured to provide a framework for a holistic, strategic approach to the organization’s environmental policy, plans and actions by enabling it to:

1. Identify and control the environmental impact of its activities, products or services, and to
2. Improve its environmental performance continually, and to
3. Implement a systematic approach to setting environmental objectives and targets, to achieving these and to demonstrating that they have been achieved.

LRCM fulfills the requirements of ISO 14001 with regard to human-machine interaction implicated in virtually all man-made disasters. The most difficult and important aspect is contained in the third requirement, “…demonstrating that they [targets] have been achieved”. The RCM analytical technique applied day-to-day  as a “living” process promotes the routine examination of each maintenance event with respect to the consequences of the observed failure.  In Living RCM work orders are considered to be instances of “knowledge records”. Each record of the referenced RCM “knowledge base” consists of the basic elements that describe  failure and its causal event (called a “failure mode”). The RCM knowledge elements are responses to the seven questions:

1. What system function was compromised?
2. In what way? Was it a partial or complete or potential failure?
3. Why? What event caused the failure?
4. What happened or could have happened (at the component, system, organizational, and societal levels)?
5. Why does the failure matter?
6. What maintenance can be done to mitigate or avoid the consequences of the failure? If none, then
7. Should the failure cause be designed out or should the failure be permitted to occur?

These questions are revisited in the light of current observations at the moment of closing the work order. Requiring the technician, planner, or engineer to link the executed work order to the appropriate knowledge record invokes the above seven-question RCM thought process. As a result, should the facts warrant a clarification or change, the knowledge record will be updated, particularly Question 4 “What happened or could have happened” as a result of the failure. Questions 5, 6, and 7 will be reconsidered. The key advantage of revising knowledge on-the-fly is the responsiveness to vivid, relevant facts fresh in the minds of all involved.

LRCM, pioneered at Cerrejón Coal, provides an audit trail of evolving knowledge. LRCM is the responsible approach to HSE. Not only does the process ensure continuous growth of relevant knowledge as required by ISO-14001, but also that the state of organizational knowledge at any given moment in time prior to an incident will be available for investigation. Continuous review and refinement of knowledge as prescribed by LRCM improves the probability that an accident will be prevented. The ability to assess an organization’s preventive measures given the state of its knowledge at any point in time ensures accountability.

© 2011, Murray Wiseman. All rights reserved.

1. ^[1] Reason, J. (1990). Human error. Cambridge University Press. ↩

Maintenance software and systems including the CMMS, RCM software, Reliability Analysis (RA) tools that deliver Weibull Analysis and Simulation, Condition Based Maintenance (CBM) systems that capture and analyze condition monitoring and process data, and Asset Performance Management (APM) software that filters and groups data from multiple systems and databases are high quality products that have outstanding technical characteristics and performance. They are indispensable in today’s high tech competitive environment. Yet their contribution toward improved maintenance, reliability, and cost performance has eluded measurement by managers, engineers, and CFOs.

Maintenance departments in all industries seek out and implement these excellent products. They are all well supported and make use of the latest software engineering technology. Whether a company chooses one brand or another has little influence on its long term success in achieving Continuous Process Improvement (CPI) in maintenance. The software brand selected is hardly the determining factor. In other words, after the project is implemented and the novelty has subsided, something is still missing.

This web site explores this important subject. It describes the fit between a “living” RCM (LRCM) process for achieving reliability from data and the variety of data centric tools that organizations are required to evaluate and compare in preparation for a reliability improvement initiative . Feel free to browse the various categories and tags in the sidebar.

I’ll appreciate your thoughts and comments on this material. And will gladly discuss any questions.

© 2011, Murray Wiseman. All rights reserved.

© 2011, Murray Wiseman. All rights reserved.

A Failure Mode is an event that causes a change in a part resulting in the loss of a required system function. Failure modes of a major component (for example, an engine, or a final drive)  can be identified accurately only upon disassembly. We must not presume that when an equipment in the fleet exhibits a symptom, for example an anomolous vital sign or a high oil analysis result, that this alone determines a failure sample point for an EXAKT model. The failure must be confirmed visually by the rebuilders or failure analysts. Only then should it be used as a point in a sample for any kind of reliability analysis.

Conversely, an engine that is rebuilt as a result of having reached its predetermined useful life could still, neverthless, present damage or advanced wear to the extent that it is judged to have “potentially” failed. That is, failure is considered, in the judgment of the technicians or failure analyst, imminent within hours or days. Obviously, to make such a judgment, the organization must, through consensus among technicians and failure analysts, gradually develop standards for distinguishing between suspension and failure. Photographic records and other rich information are helpful.

The scheduled disassembly of an Engine at age 18988 hours detected advanced wear of the connecting rod bearings at positions 7 and 8. The photos show the condition of the crankshift at position B7-A7 and the connecting rod bearings at positions B7-A7, B8-A8. At the other crankshaft postions wear was much less pronunced and considered ”normal”.

The failure analysts can usually identify the failure mode directly responsible for failure. In addition they may also identify potential failures occurring elsewhere in the engine. A standardized investigative process provides the detailed information about the engine’s actual state at the moment of disassembly.  The true states of each relevant failure mode constitute  sample points for the development of a decision model. The more accurate the identification of the failure mode and its ending event (Failure or Suspension), the more confidence can be entrusted to the decisions flowing from the model.

A disassembled engine, whether it has failed functionally or not,  is a rich source of information vital to reliability analysis. The engine reveals its secrets of failure mode behavior completely for us. Typically however, Engines which have completed their scheduled life prior to rebuild are not subjected to failure analysis.  Consequently it is assumed, mistakenly that all failure modes in the engine were suspended rather than to have potentially failed. The error of this assumption is clear in the case depicted above. Ignoring or mistaking potential failure for suspension and will result in CBM decsion models having poor prediction performance.

© 2011, Javier Custode. All rights reserved.

© 2011, Murray Wiseman. All rights reserved.

Download Video: Closed Format: MP4



© 2011, Murray Wiseman. All rights reserved.

It depends on how well the monitored (CBM) data reflects internal damage or external stress on the asset. If the intrinsic relationship between the condition data and the developing failure mode is strong, a relatively small amount of data (at least four life-cycles ending in functional or potential failure) is required. If the relationship is weak, more historical data is required to achieve a good  model. The software tells you how “good” your model is and whether and to what degree it is acceptable for predictive use, meaning how much confidence can you ascribe to  decisions made using the model. The software measures this by indicating confidence bounds in the Remaining Useful Life Estimate (RULE) and by providing a standard deviation indicating the amount scatter around the RULE. The modeling process guides you in cleaning and augmenting your data management methods to achieve continuously growing confidence in Condition Based Maintenance (CBM) decision making. More information on this subject is given at Confidence in Predictive Maintenance.

It is preferable to build a sample for RA and predictive modeling based upon past failure mode life cycles ending in potential failure rather than life cycles ending in functional failure. This is because functional failures usually have significant consequences. A potential failure, on the other hand, is an imminent failure, confirmed by the technician’s observations at the time of repair. A potential failure has relatively fewer consequences.

If a maintenance organization does not have good historical data (the case in most companies) it means that they haven’t got a good reliability information management process in place. Without such a process, systematic improvement in maintenance and reliability is impossible. With such a process in place, the RCM knowledge base grows with each new significant work order. The (LRCM) process requires linking each work order, at the time of closure, to a RCM knowledge record. The work order should also indicate whether the failure mode failed (a functional or potential failure) or was preventively renewed without having failed (a suspension). Using this approach, good samples for analysis can be obtained automatically. Reliability Analysis software such as EXAKT, can be applied easily to these data samples.

A PM (see the article What is PM) work order is should provide two types of data (See the article What’s the right data).

© 2011, Murray Wiseman. All rights reserved.

Quick fix work orders can be considered CBM observations or inspections of a sort. They are similar to non-rejuvenating repairs. Their rate of occurrence can be used to trigger a diagnostic work order as that described above. Assume that the cuprit is a particular seal. How should the information surrounding this failure be used to build a sample for Reliability Analysis?

Although there were five work orders, only the fifth located and corrected the single failure. Truthfully the failure occurred at the date of the first work order not the fifth where it was finally diagnosed. The offending seal may have been similarly detected in other trucks in the fleet. The consequences (average downtime and failure frequency) would be determined from the various work orders involved. The downtime would be that accumulated over all five work orders. With half a dozen detections of this type a data sample can be extracted and reliability analysis performed. A Weibull and cost analysis could be used to justify failure analysis leading to redesign. An EXAKT analysis could be performed to discover which, if any, monitored variables will allow detection and correction of the problem at the first work order rather than waiting for repetitive confirmation as is the practice.

© 2011, Murray Wiseman. All rights reserved.

© 2011, Murray Wiseman. All rights reserved.