Quick wins from operational changes

A monthly column by Daryl Mather, author of “The Maintenance Scorecard”. First published by www.PlantServices.com

As most of us are aware, maintenance is about the cost effective management of assets for a given level of performance and risk. Part of this equation requires failure management policies for equipment that provide the best value for each operational maintenance dollar, a reduction of the risk of critical failures to a tolerable level, and sustainable levels of production or operation.

If only it were that simple! The sad fact is that often many of the failures that maintainers are required to manage are not under their control. Fundamentally, there are three ways that equipment fails; (Discounting failures caused by human error)

1. Failures caused by wear out of components or degradation due to either use, time or the environment they work in. Corrosion, erosion, wear, chemical breakdown and structural degradation are among the causes of these types of equipment stoppages.
   
2. Failures caused by operating equipment out of limits, operating in a way that increases the probability of failure, or with incorrect feed materials. Equipment overload, lack of knowledge of operational parameters, and changes in equipment demands are the chief reasons of failure in this area.
   
3. Failures caused by inadequate or unsuitable components. These are caused by assets being not fit-for-purpose. Poor purchasing practices, inadequate asset selection practices, and poor initial design are the chief reasons for these types of equipment stoppages.
   

Of all these failures it is often those in the first point that are able to be directly influenced by the maintenance department of most companies, yet they also required need to manage and react to the failures caused by operations, purchasing and/or asset design. This highlights one of the regular comments of a friend of mine, Mr. Steve Turner, “Maintenance is a process, not a department”. Maintenance managers need to be the plants diplomats, stitching together agreements on operational and purchasing strategies to ensure cost effective performance.

One area where there are some quick wins available for reducing operationally induced failures is in the area of equipment rotation. Over the years duty and stand-by have become almost redundant terms in many plants, and have been replaced by a 50/50 spilt of operational running.

The thinking behind this appears sound on the surface. It is most common with pumping systems, but can affect any duty/stand-by asset configuration. By regularly changing the pump (or whatever) that is operational at any one time the belief is that the company is getting two advantages:

• It is ensuring that the stand-by pump (say) is going to be ready for operation if called upon, and
  
• Ensuring a longer mean time to repair by interchanging the two pumps. Thus gaining the combined life out of the installation, rather than running one pump only.
  

These are intoxicating arguments, and as stated above, they do appear sound on the surface.

However, they are both wrong and misleading!

If we take the example of pumps, the most rigorous and stress inducing time that they go through is that of start up. This places significant strain on the pump in various areas. Most notably seals and bearings are prone to failures due to frequent starts. So by stopping and starting pumps regularly we are often increasing their probability of failure rather than reducing it.

If we are going to use a true duty/stand-by arrangement then the pumps can be running continuously for long periods of time without having to go through the start up process. In runs of shorter time frames, only one pump at a time is exposed to the regular stresses of start-up, while the other remains dormant and tested infrequently (but regularly) to ensure reliability.
So, higher reliability through frequent starts is not a valid argument in many cases. What about the second argument, that of increased time between failures of the process due to interchanging the pumps?

Pumps, to continue with the example, contain components that wear out at a rate dependant on what is being pumped, the rate required, and the design of the pump impellers and housing in particular. So, if they are pumping the same media, at the same head pressure and rate, then they will wear out at roughly the same rate.

If we discount the effects of frequent start up then this will allow us to swap pumps and extend the overall time between failures. But then what? The fundamental reason for having a stand by unit is because we do not want to run the risk of losing the function of the process, whatever it is. Stand-by units are effectively an insurance policy against the failure of the prime or duty unit.

Instead of operating the assets in a way that will decrease the risk of failure this strategy will in fact increase the likelihood that when one pump fails due to age related wear; the other one will fail at around the same time! Almost guaranteeing a loss of function instead of protecting it! When combined, these two effects actually call into question the whole strategy of having a standby pump in the first place! It would be cheaper to have just one and to lose the function when it fails.

In the vast majority of cases it is hard to justify anything less than a 90/10 split of operational time. In fact, even this could be reduced without any adverse effects at all.
All we want to do is to prove it will work when needed. Where safety is not an issue, the frequency of testing intervals is determined by cost of the failure and the likelihood of it occurring, and balancing that against the cost of doing the task itself, in this case the functional test. Because the consequences of failure are purely economic, there will come a point where the risk of failure is less than the cost of doing the task at the required frequency.
The failure we are trying to avoid is the one that will lead to total system failure. This would be a failure of the stand by asset, and in this case could be flat spots in bearings, hardening of whatever is being pumped, or some other reason.

A recent case that I reviewed included the lime dosing pumps in a water treatment plant. The configuration was three duty units and one standby pumping unit. One of the factors we needed to take into account the fact that lime hardens over time, so if the standby pump was left for too long it would never start as it would have set solid! As the cost of failure was too high, the frequency of the functional test was based on not allowing the lime to harden.

Other similar cases include a factory where we pumped milk, and this included a system clean out on changeover, and a gold processing plant. In the gold processing plant what we had to avoid was the settling out of the gold in the slurry, representing a loss of profit even though small.