Reliability Revolution

Is a new shift in paradigm occurring?

Author : Manou Hosseini B.Sc., Ph.D.

Canadian Mining Journal, April 1999

In Optimising Production Scheduling for Maximum Plant Utilisation and Minimum Downtime¹, Sandy Dunn raises the important issue that Availability and Utilization, which are traditionally considered as being performance measures by many mining organizations, are insufficient for decision-making on equipment strategies. He emphasizes that we ought to focus on the Equipment Reliability, instead, as it has a significant impact on equipment performance: "The focus on reliability is revolutionizing the way that mining companies look at improving short-interval scheduling and equipment performance."

The purpose of this article is to discuss the equipment reliability issue and address the need for the analysis of various performance measures through "what-if" examinations of various business conditions and operation scenarios. This can be achieved by developing comprehensive reliability and maintenance optimization models, which are integrated with the operational characteristics of operating systems.

Defining Reliability

First of all, let us define reliability. The reliability of an item is the probability that the item will perform a specified function under specified operational and environmental conditions, at and throughout a specified time.² The first important issue to notice is that reliability is a probability. This means that we are dealing with the law of random chances as they occur in nature. For example, the occurrence of untimely interruptions in production as a result of shovel failures, are random events, the expected frequencies of which we aim to reduce. The next thing to notice is that demand time may be of four kinds:

an instant in time,
a time interval (a mission),
continuous (all the time),
continuous in a pair of modes (standby and active)

Thus the best way to specify the reliability of an item depends upon how the item is expected to function. This is shown in Table 1.

Table 1 Reliability in various demand times

Demand Time Reliability Measure

Instant / a Series of Instants Success Probability

Interval Mission Reliability

Continuous Mean Time Between Failures (MTBF)/ Availability

Standby Dependability

Here, our focus among the above four demand times is on the "interval" and "continuous" time demand cases. In the interval case, we are concerned with mission reliability or simply reliability. This is defined as the probability that an item will operate without failure throughout a specified interval. For example, where we are scheduling the next week's production, the equipment reliability or probability that the equipment will operate throughout the week is our concern.

However, if we want to evaluate the performance of a piece of equipment with a continuous demand, for instance, within the last two years, the focus should be on the expected mean time between the failure events (MTBF) that cause the equipment to go down. In this case we may also focus on the availability of the equipment, which can be defined as the fraction of time that the equipment was actually operating.

Considering the definition of reliability, it can be seen that in Reference 1, enough attention has been paid to the interpretation of reliability within the context of appropriate demand time. Obviously, in the case of a short interval production-scheduling focus should be placed on the probability that the production system performs without failure. Therefore, how can one say that this is a "Reliability Revolution"? This is simply the definition of reliability within an interval time demand. The merit of Reference 1, however, is its attention to the need for more complicated performance measures, such as so-called Production Efficiency and the Overall Equipment Effectiveness. Production Efficiency is defined as the ratio of actual output from a machine (which meets the required quality standards) to its rated output, during the time that it is operating. The Overall Equipment Effectiveness is defined as the product of availability, utilization, and production efficiency.

Availability and Utilization

The following two examples are mentioned in Reference 1 to demonstrate how availability and utilization are insufficient as performance measures. Consider two situations:

A haul truck is operating but because of a problem with the engine it can only haul at 80% of its normal speed. Apparently, while the truck is available and being utilized, according to the definition of utilization, the maximum output has not been achieved.

A shovel happens to trip, causing a 15-minute delay while it is reset. During this time the haul trucks start to queue at the shovel. One can see that while trucks are available, the maximum output is not achieved.

In the cases cited above, production efficiency and overall equipment effectiveness are more appropriate measures of performance.

A more complicated financial measure, which represents the value of equipment effectiveness, is called economic value added (EVA). John S. Mitchell³ proposes an EVA model for the smallest identifiable producer within an enterprise. The producer is defined as an entity or subsystem for which the cost of materials and the price of finished goods can be calculated. Examples include: each line in a manufacturing facility or each production unit in a mining site, as well as each unit in a multi-unit power station, chemical plant or oil refinery. EVA is basically after-tax operating profit minus the cost of capital. The larger the value of EVA the more value being created.

Maintenance Process Maturity

It is interesting to note that equipment management process maturity, like many other processes, may evolve from an ad hoc level up to an optimized level (Figure 2). The process maturity level does not necessarily indicate the sophistication of maintenance tactics and technologies that an organization may deploy. Some organizations may use various kinds of modern condition monitoring tools but seldom measure the performance of their process and associated costs. On the other hand some organizations may not have any condition-based maintenance systems in place, but conduct very sophisticated performance measurements. When the performance is measured and tracked, one will concomitantly realize the need for optimization, for instance:

Minimizing maintenance costs by finding optimal maintenance/inspection intervals (frequencies)
Minimizing the number of resources while maintaining the same amount of output
Measurement of the sensitivity of a producer to a maintenance policy
Quantitative justification of the optimal diagnostics and condition monitoring
Determination of the optimal amount of condition based maintenance, preventive maintenance and run-to-failure maintenance

reliability_revolution.gif - 7664 Bytes
Figure 2 Maintenance process maturity levels

Finding the Best Strategy

A practitioner of equipment management may now ask: What is the best equipment management strategy for my case? What is the best performance measure that I should track? One may indeed ask the same questions but adding the phrase "in today's slow market". Another manager may want to know how to optimize short interval scheduling and a third one may want a long-run optimization. A fourth manager may not even want to know about any optimal processes since he or she is still struggling to take control of the processes themselves.

In fact, an ideal optimization model should contain provisions for "what-if" analyses under variations in business and operating conditions. In an optimization model, the incorporation of the operating characteristics of the so-called producer to reliability and maintenance functions makes it possible to investigate the impact of the market situation and production scheduling on optimal policies. The model will then be solved for various performance measures. This leads to creation of a knowledge base for the decision-makers, providing them with the "what" to each "if".

A Paradigm Shift

Referring to the definition of reliability, in an optimization model we should deal with probabilities and random events. Failures occur randomly. Repair and maintenance durations are not fixed periods. The fluctuations in the market happen randomly. Here we are dealing with a family of random variables and complex models. There has been tremendous effort by both maintenance practitioners and academics to address these issues which are in the area of reliability, maintenance optimization and operational research. However, the focus has mostly been on performance and reliability of the components of a system individually and in such a way that they perform in isolation. In fact mathematical tractability of complex models is a big issue and the traditional mathematical approach to maintenance optimization cannot effectively deal with complicated models.

The challenge of real world problems requires a new paradigm shift in maintenance and reliability modeling and optimization. Recent developments in modern analytical and simulation modeling tools, however, make it possible to develop and analyze complicated reliability and maintenance models and examine "what-if" questions that mining organizations will be faced with, in the more competitive market of the next century.

As a final word the issues discussed in this article are supported by nearly half a century of research in maintenance and optimization, although mostly at the component level. Nevertheless, development of new and modern modeling tools synthesizes these achievements in favor of complicated situations and complex models, bringing these achievements from academic journals to the shop-floor.

References
¹ Sandy Dunn, Optimising Production Scheduling for Maximum Plant Utilisation and Minimum Downtime, The Reliability Revolution. Dollar Driven Mining Conference. Perth, Western Australia, July 1997.
² Paul Kales, Reliability for Technology, Engineering, and Management, Prentice Hall, 1998.
³ John S. Mitchell, Producer Value- A Proposed Economic Model for Optimizing Equipment (Asset) Management and Utilization. MARCON 1998