Heinz P. Bloch, P.E., owner of Process Machinery Consulting, will present “Breaking the Cycle of Pump Repairs” at the 43rd Turbomachinery/30th Pump Symposia in Houston on Sept. 23 at 2 PM and on Sept. 24 at 10:30 AM. The tutorial will describe how achieving the lowest possible life cycle cost (LCC) or lowest cost of ownership has been an undisputed goal of most pump users. From his 52 years of solid engineering practice, Bloch knows that pump hydraulics and fluid sealing details have received due attention over the years. However, bearing protection and lubricant delivery components are now often decades old, and many no longer reflect the best available technology. Additionally, "lean and mean" has often morphed into "cheap and risky." He will explain why a few truly reliability-focused users enjoy pump MTBRs that are four times greater than others’ and why their per-pump maintenance expenditures are often only one-fourth of those forced upon their struggling competition. This tutorial will delineate at least 10 little-known but highly important details of major interest to pump users seeking to improve pump life and minimize maintenance cost.. Learn more about the 43rd Turbomachinery/30th Pump Symposia at http://pumpturbo.tamu.edu.
Of the numerous process centrifugal pumps undergoing repair right this very minute, an estimated 90% have failed randomly before. Some have run just fine until the very first repair two or three years after startup and were never quite the same since after the first repair. Other pumps failed frequently or randomly, perhaps once per year, from the time they were originally commissioned. That brings up some questions. Could it be we don’t really know why many process pumps are failing? Could it be we just don’t give pumps the attention they deserve? Is it because everybody’s priorities are elsewhere? Or are there perhaps elusive failure reasons — factors overlooked by all parties?
Fortunately, improvement is both possible and cost-justified. Allowing repeat failures on process pumps rarely makes economic sense. Simple benefit-to-cost or lifecycle analyses will easily demonstrate that the pursuit of remedial action greatly benefits users.
The cost of failures
One way of exploring the value of extending pump mean time between failures (MTBF) is to examine the likely savings if we could improve the MTBF from presently 4.5 years to a projected 5.5 years. Say a facility has 1,000 pumps; that’s 1,000/4.5 = 222 repairs before and 1000/5.5 = 182 repairs after understanding and solving the problem. Avoiding 40 repairs at $6,000 each is actually a very low estimate, but would be worth $240,000. Avoiding repairs frees up manpower for other tasks: At 20 man-hours times 40 incidents times $100/hr, reassigning these professionals to other repair avoidance tasks would be worth at least $80,000.
There is also about one $3 million fire per 1,000 pump failures. An engineer at a U.S. Gulf petrochemical plant thought it might more likely be 1 fire per 1,000 pump failures; then, out of 10 fires, he figures seven are less than $50,000, two are between $50,000 and $500,000 and one is more than $500,000. We use the number $3 million/1,000 failures based on 52 years of experience and two recent updates. Data obtained in 2012 at an oil refinery in Indiana considered my number low, whereas a facility in Minnesota in 2009 thought it was spot-on.
It should be noted that data and contributory details of catastrophic incidents are often closely guarded secrets. Virtually all consulting done today by qualified independent professional engineers is linked to a legally binding nondisclosure agreement. The client is often compelled to file reports with local and federal regulatory agencies. These reports might differ from the findings of consulting engineers who understand the true root causes of failures or whose sense of priorities is tuned to higher standards. Diverging statements or findings might feed a bureaucratic machine that will busy itself with issues of this type.
Meanwhile, we consider these estimates for the value of fire damage restoration as reasonable. Therefore, avoiding 40 repairs would be worth 40/1,000 x $3 million = $120,000. Together, the three items — $240,000, $80,000, and $120,000 — add up to $440,000.
Although $ 6,000 was used for repair cost avoidance calculations earlier, an average API pump repair at a Texas refinery costs slightly over $10,200; a refinery in Mississippi reported $11,000. If the incremental cost of upgrading during the next repair adds $2,000 to the repair bill and avoids even a single failure every 3 or 4 years over the 30-year total life of a pump, the payback will have been quite substantial. It would be reasonable to assume eight avoided repairs at $6,000 give payback of $48,000/$2,000 = 24:1.
The estimated repair cost numbers of close to $11,000 reflect what needs to be considered in a pump repair cost calculation: Direct labor, direct materials, employee benefits at roughly 50% of direct labor, refinery administration and services costs at close to 10% of direct labor, mechanical-technical service personnel overhead costs amounting to around 115% of direct labor, and materials procurement costs from 7% to 8% of materials outlay. Disregarding the true cost of failures or repairs is likely to deprive some users of seeing the true benefit-to-cost ratio associated with pump upgrades.
We could examine other ways to calculate, as well. It would be reasonable to assume that implementing a component upgrade, generally the elimination of a weak link, extends pump uptime by 10%. Implementing five upgrade items yields 1.1^5 = 1.61 — a 61% mean-time-between-repair (MTBR) increase. Or, say, we gave up 10% each by not implementing six reasonable improvement items. In that instance, 0.9^6 = 0.53, meaning that the MTBR is only 53% of what it might otherwise be. That might explain industry’s widely diverging MTBR. The MTBR gap is quite conservatively assumed to range from 3.6 years to 9.0 years in U.S. oil refineries, and, as of 2014, no well-informed pump professional has disagreed with this range of MTBR numbers.