|Dick Ghiselin • Houston|
We have heard it a million times. "If it ain't broke, don't fix it!" This is the argument for reactive maintenance. And in some cases, it makes sense. Run things until they break; then fix them. We are all guilty of performing reactive maintenance. For example, we typically wait until light bulbs burn out before we replace them. The downside is a minor inconvenience; the upside is we get maximum life out of the light bulb. Imagine what it would cost over time if we systematically replaced our light bulbs every year. Or if we kept track of the "on" hours and replaced them when they reached some arbitrary limit. Since we have little chance of predicting with accuracy when a lamp is going to burn out, it is unreasonable to implement some kind of preventive maintenance plan for light bulbs.
Naturally, most folks minimize the inconvenience factor by keeping a few spare bulbs on hand. The added cost is hardly worth counting, and only amounts to the cost of maintaining an inventory and storage. For light bulbs, this is infinitesimal.
Another example is your car. If you run your car until it quits, the cost and inconvenience can be considerable. And we all know about Murphy's Law. The chance your car will break down across from the dealership is also infinitesimal. The breakdown is more likely to occur on a country road, on dark night, in a blizzard, as you are rushing your wife to the delivery room. This is why most people follow a prescribed preventive maintenance program of systematic checks and oil changes. The cost and risk of an unexpected breakdown far exceed the cost of the maintenance program.
But what about an offshore rig? Maybe we can follow the wait-and-see program for the light bulbs, but there are dozens of items onboard where failure is not an option. For example, most rigs have an extra mud pump, so they can activate it when one of the operating mud pumps indicates it needs refurbishing, usually by a reduction of output pressure. By performing repairs offline, downtime is eliminated, and the risk that both operating pumps will fail at once is very low. Even so, the risk can be minimized by systematically taking a pump out of service and rebuilding the fluid ends after a certain number of operating hours.
The engines that drive the generators are another story. Maintaining electrical power on a rig is vital, and tearing down a diesel engine is a non-trivial activity. Can we afford to wait until they fail before doing something about it? What should be done? Change the oil and filters at regular intervals like we do on our car? The answer might surprise you.
Time is not the answer
For items where failure is an unacceptable risk, performing preventive maintenance on an arbitrary time schedule - miles, operating hours, etc. - is both costly and risky. Do engines have some sort of built-in weak point that we know will fail after a specified time? Of course not. The engine manufacturers have run exhaustive tests of their products and have determined the least common denominator of acceptable risk. This established the manufacturers' recommended maintenance interval. For most cars these days, the number is 5,000 mi (8,000 km). This means that almost all engines are expected to last longer, and if the oil and filter are changed at 5,000 mi, the engine should not fail. However, one major manufacturer determined that 80% of its engines could last 10,000 mi (16,093 km) or more without maintenance. But they were unwilling to risk the bad reputation they would get if the remaining 20% failed prematurely. So changing oil at 5,000 mi costs a majority their customers a lot of money they did not need to spend.
Anything can cause premature failure. Operating in dusty or sandy desert areas shortens engine life. Operating at ambient temperature extremes of heat or cold for long intervals shortens life. Salt spray, vibration, or shock loading can all contribute to shorter lifespans. How can we be sure we are really preventing failure?
Help from above
The solution to the maintenance problem comes from the airline industry and the military air forces. In a situation where failure is not an option, these entities have adopted condition-based maintenance (CBM). Rather than depending upon arbitrary time or distance schedules, these people systematically measure the condition of their engines and vital sub-assemblies. Samples of crankcase oil, transmission fluid, gear oil, and exhaust gases are captured at regular intervals and subjected to detailed chemical analyses to the molecular level. Not only is the oil condition determined, which tells if it is losing its lubricity or mobility, but also trace elements of engine metals can be recovered from the oil that indicate wear of critical components. Cylinder walls, piston rings, valve stems, turbine bearings, and blade erosion leave tell-tale residue in the oil that signals when it is time to take the component out of service for maintenance. These tattle-tales are valid regardless of the operating environment, so they are not time-dependent.
For non-lubricated components, other parameters can be measured that indicate when they are about to fail. For electrical submersible pumps, shaft vibration, and increased motor-winding temperatures warn that the pump is in failure mode. Spares can be staged on-site and a switch can be performed with minimal downtime. In many cases the pump can be retrieved before permanent damage occurs. These can be refurbished to like-new condition and re-deployed.
Rigs of all types have one-of-a-kind components that are critical to the drilling operation. Top drives are an example. Using CBM techniques can give early warning that they are in incipient failure mode in time to schedule a change-out at the most convenient time. Repairs conducted offline eliminate costly field failures, rig downtime, or critical loss of production.
There is a better way
The "If it ain't broke" mantra still works for some things, but adopting a time-based maintenance program amounts to overkill in many cases and can be quite costly. Even so, it does not prevent all breakdowns. Monitoring the condition of all critical components has the highest likelihood of identifying failure modes before failure occurs. CBM is relatively inexpensive, both from the cost of the analysis to the time it takes rig personnel to catch the samples. By identifying components before they fail, collateral damage is avoided and often the component can be refurbished and returned to service. Then switch-outs can be scheduled for the most convenient time from an operational standpoint.