COLUMN | Focus on Fundamentals
Revisiting MTBF: Establish Focused MTBF Indices
Sotirios Christofi | Industrial Machinery Consultant
Mean time between failure (MTBF) is generally seen as a maintenance index to measure maintenance personnel performance. However, an unplanned pump breakdown usually causes considerable production losses at much higher costs when compared to the usual maintenance costs. This article suggests a MTBF approach focused on the pumps—which can cause this kind of loss—as a way to measure the impact of pump reliability.
| IMAGE 1: Reliability-centered maintenance general bathtub curve (Image courtesy of Hydraulic Institute)
Definition
MTBF represents the average time between two failures for a repairable system. It is one of the most widely recognized yet least understood indicators in the maintenance and reliability world.1
It is a major industrial index for calculating the performance of maintenance personnel, and this calculation is often used to compare the pump performance at different plants or plant units or from different manufacturers. It is also used to measure the performance of different maintenance teams or for management to track key performance indices (KPIs).
MTBF formula
MTBF = (total number of pumps/number of pump failures within the considered time) x length of considered time
Example
- Total number of pumps = 500
- Length of considered time = 12 months
- Number of pump failures within the considered time = 50, or 10% of the pumps
- MTBF = (500 x 12)/50 = 120 months = 10 years (a reasonable result, because if 10% of the pumps fail every year, then in 10 years, all pumps will have failed once)
Equation 1
What Is a Failure?
In short, a failure is any condition where a pump can no longer perform its intended function effectively or safely. Still, in some industries there is confusion about what should count as failure.
Small maintenance activities during pump operation that prolong the pump’s useful life after a predictive maintenance finding—such as retightening of packing after visual inspection or oil refilling after oil analysis—should not count as failures.
Activities that necessitate the pump stoppage, however brief they might be, such as shaft realignment after vibration analysis or pump partial or overall overhaul, should count as failures. The main criterion here is pump stoppage. If a critical pump has to stop, it is not fit for operation. This would count as a failure. Preventative maintenance on an operating pump is not a failure.
Should All Failures Count the Same?
Pump failures are grouped in three categories (Image 1).
- Early failures – This includes those at the beginning of operation mainly due to construction errors, incorrect misalignment or balancing, defective or faulty parts, improper lubrication, pipe-induced strain, seal installation errors and more.
- Constant random failures – This includes clearance widening due to corrosion/erosion; unbalance due to deposits; misalignment due to part wear; bearing failure due to foreign objects; unbalance and misalignment; leakages; process upsets; operator errors; cavitation; foreign objects; etc.
- Wear-out failures – This includes wear-out of corrosion allowance, end of operating life for bearings, elastomer seal degradation, coupling insert degradation and more.
Attention: MTBF is applicable to constant failure rate situations, i.e., only those during the normal operating period.
Spared vs. Unspared
A spared pump is a pump configuration where one or more backup pumps (standby) are installed in parallel with the operating pump. If the primary pump fails or requires maintenance, the "spare" can be manually or automatically started to keep the plant running without production loss. An unspared pump is a single pump operating alone in a specific process loop with no permanently installed backup. If an unspared pump fails, the process stream stops, which usually forces a partial or full shutdown of the operating unit.
Should the failures of spared vs. unspared or critical vs. noncritical pumps count the same? The general approach is that they do count the same, but this does not take into account the overall impact of the pump failure. However, the failure of an unspared pump can affect the plant’s bottom line far more than the failure of a spared pump.
The cost of production losses, along with lost profits, lost market share and lost customers or market image, is generally accepted to be at least one order higher than the maintenance costs. A medium-sized refinery is generally estimated to have a cost of production loss of $500,000 to $2 million a day—depending on the technology and profit margins compared to the cost of a pump revamp. This usually does not exceed $100,000, even for the bigger ones.
| IMAGE 2: Engineer performing a pump inspection (Image credit: SAHARAT - stock.adobe.com)
Preventative Maintenance
Should preventative maintenance activities count as failures? In plants where they do not, maintenance managers might be tempted to perform unnecessary and excessive preventative maintenance. This reduces the probability of a failure but increases the maintenance costs and effort.
In plants where preventative maintenance activities do count as failures, the maintenance manager may feel their team should perform as little maintenance as possible or no preventative maintenance at all, thereby increasing MTBF but also increasing the probability of a failure during operation with correspondingly high production losses. Unsurprisingly, sometimes MTBF increase is reported despite repeated unplanned shutdowns.
So, should the failures of spared vs. unspared or critical vs. noncritical pumps count the same? Should preventative maintenance activities count as failures or not? The answer is the introduction of “focused” MTBF critical failures (MTBFCF) and MTBF all failures ( MTBFAF).
MTBFCF indicates the impact of pump reliability as it relates to the plant service factor. This could be considered an “operations” MTBF.
The only failures that count are those of critical pumps, i.e., those that cause production losses. This includes:
- Unspared pumps, which can cause production losses
- Spared pumps, which cannot or historically are not repaired within the allowable time frame
Preventative maintenance activities do not count. These pumps have to be adequately pinpointed, the importance of prevantative maintenance should be made clear to all parties involved and MTBFCF should be introduced as a major management KPI.
MTBFAF indicates the overall maintenance effectivity. This could be considered a “maintenance performance” MTBF. With MTBFAF, the failures of all pumps should count. Preventative maintenance activities do not count.
The conventional MTBF can be used as an indication of overall maintenance effort. For this, the failures of all pumps should count, including all maintenance activities. This could be considered a “maintenance cost” MTBF.
Alternative MTBFs can be established depending on the plant’s needs, according to the following:
- Pump construction code
- Different plant units
- Different manufacturers
Given the access to digital maintenance management systems, these indices are now easier to produce than ever. However, what is the allowable time frame for repairing a failed pump? How should it be calculated? This will be discussed in my next article, “Introduce Maximum Allowable Repair Time of Spare Pumps as a KPI,” in Pumps & Systems.
References
- reliabilityacademy.com/articles/defect-elimination/mtbf-mean-time-between-failure
After a career of 38 years in the petrochemical industry, Sotirios Christofi provides consulting and training services for technical personnel on the design, construction, inspection, selection, condition monitoring and maintenance of industrial machinery. He is a mechanical engineer with a master's degree in business administration from the University of Warwick, U.K.
In This Issue
