A Simplified Method for Justifying Condition Monitoring Projects

It required only a couple notes of the haunting soundtrack to signal an imminent shark attack in the movie Jaws. As the theme began to swell, it became obvious that someone was going to “get it.” It would be nice to have a sixth sense that allowed us to hear "warning" music in daily situations. Condition monitoring systems provide a means of sensing when sharks are skulking toward your critical machines.

I have often been asked by users of rotating equipment, as well as vendors of condition monitoring (CM) equipment, "How should condition monitoring projects be sold to plant management?" While this is a complex question, a highly simplified economic justification approach may be the answer. Instead of starting with the CM hardware provider telling the users what equipment is required, I have turned the reliability assessment around by first asking the question, "How much CM investment will the potential reduction in machinery risk support?"

Before I present the simplified approach, we need to understand the concept of risk. Risk is a product of two quantities: consequence and frequency (or probability). For justification, consequence is usually presented in monetary units. If a pump fails catastrophically, a potential consequence is $50,000. If this event occurs at a frequency of once every ten years, this represents an annual risk of $5,000/yr, but if it occurs every two years, the annual risk is $25,000/yr.

How Insurers View Risk

When insurers set premiums they first want to know the following levels of consequence¹:

• Normal Loss Expectancy (NLE)-For a pump, this would the cost of a normal failure.
• Probable Maximum Loss (PML)-For a pump, this would be the cost of a catastrophic failure. This level of event sets the insurance policy pricing (see Figure 1).
• Maximum Possible Loss (MPL)-For example, a pump case ruptures with significant product release and fire. This level of event sets the policy limit (see Figure 1).

We will concentrate on the PML consequence level as a means of justifying a CM project. In my experience, it has been easy to get management to visualize the PML event for a given piece of equipment.  Here is a typical description of PML for a centrifugal pump: Imagine an unmonitored pump ingests water in the bearing housing, causing a rapid bearing failure and subsequent seal failure. As a result of this primary failure, the rotating element rubs and eventually breaks in two pieces, resulting in extensive casing damage. Because this pump is unspared, you will lose $250,000 in production and $50,000 in pump repair costs, for a total of $300,000. If this event occurs every ten years, an annual risk of $30,000 is experienced.

How Much CM Equipment Can I Justify?

Economic justification requires knowledge of four things:

• Annualized risk of the "do nothing" case
• Annualized risk of the improvement case
• Payback period requirements or IRR requirements. To keep things simple, I will only consider a simple payback model.
• Probability of detecting the machine fault in its early stages

Assume the annualized risk can be lowered from $30,000 to $5,000-with a certainty of 80 percent-by installing condition monitoring hardware. This represents a project income of $20,000/yr. If management requires a two year simple payback, then we can justify $20,000 x 2 ($40,000) in a condition monitoring project. If we go on to assume the installed cost is twice the cost of the basic hardware, we know we can only justify the purchase of $20,000 in hardware.

Simplified Method

Now that some basic concepts have been covered, let's explore the simplified justification formula. Assume that all of the annualized risk is represented by PML events. Based on this assumption, I propose the maximum amount of condition monitoring equipment justified (CM) is determined by the following formula:

Where:

1. CM is the maximum amount of condition monitoring equipment justified.
2. PML is the probable maximum loss predicted without condition monitoring. (Include probable losses due to fire and environmental releases, etc.) The equipment OEM is one good source for this information.
3. PML_CM is the probable maximum loss predicted with condition monitoring. Again, the equipment OEM is one good source for this information.
4. D is the probability of detecting the machine fault in the primary state.
5. t_PB is the payback period required by management
6. T_PML is the predicted time between PML events. Use five years for unproven, isolated or severe service machines, ten years for typical industrial duty machines and 20 years for proven, light-duty service machines.
7. Sum these terms for all components in the train or general area to determine the total amount CM equipment justified. 

Consider a spared pump example involving two 500-hp spared pumps. No production loss is associated with pump failures. If we only consider secondary damage due to an undetected primary failure, we decide PML-PML_CM is $75,000. Now assume this is an isolated installation, and therefore select T_PML of five years, a D of 75 percent, and a management-required payback of two years. This analysis will lead us to the conclusion that an investment of up to $75,000 x 0.75 x 2 / 10 = $11,250 in CM hardware is justified.

Additional Analysis Notes:

• In cases where multiple machines are in close proximity to one another, calculate this value for all the rotating equipment-drivers and driven-in a given area and add them to determine the total amount of condition monitoring equipment justified.
• If a permanent system cannot be justified, consider a walk around monitoring approach.
• More CM hardware may be justified if a more detailed risk assessment is performed.

Once the maximum amount of condition monitoring equipment justified is determined, work with the CM hardware provider and determine how to best spend this money. The types of CM hardware will depend on the answers to the following questions:

1. What types of machines are going to be monitored? Compressors, pumps, gearbox, etc.?
2. What are most predominant machine failure modes? Bearings, seals, etc.
3. What bearing types do you have? REB or hydrodynamic?
4. What is the transmissibility ratio at the bearings, i.e. the ratio of casing vibration to shaft vibration?
5. What are the machine/shaft operating speeds?
6. How will the failure of the critical components manifest themselves? Vibration, temperature, acoustic energy?
7. How do these components usually fail? Gradually or suddenly?

Finally, determine if the proposed CM installation will actually yield the reduction in annualized risk assumed in the original analysis.  If not, the user must consider other CM design options or rerun the analysis considering the new assumptions.

I am curious to know what readers think about this simplified approach to justification. Is this method too conservative or too liberal? If there is enough interest, I will cover several illustrative justification examples in the next column.

References

1. Jones, R. B., Risk-Based Management, A Reliability-Centered Approach, Gulf Publishing Company, 1995, pp. 197 to 201.

Pumps & Systems, March 2008