Standby pumps reliability 3

[pumphead]

Are there any studies showing how swapping pumps with their standby spares impacts reliability? I realize different pump designs, services, fluids, etc. may require different strategies. In general, its a question of what is the desired benefit, what are the costs and potential risks, and then is it worth it. This will vary for different applications, and organizations - I get that. There are arguments for both sides, so what I'm curious is if there are any documented studies on the topic that can be shared.

 

[MFJewell]

I've never seen a study on it and like you said it will vary for each unique scenario.

I did an analysis for a project (power gen) for various pump configurations (2x100%, 3x50%, 2x50%) on two large water systems. Since it was a power plant, I used lost generation revenue versus capital cost and additional O&M expense to determine a payback period. Payback period was quite long, so we went 2x50% on one system and 3x50% on the other.

 

[LittleInch]

Yes, they are called RAM studies ( Reliability, Availability and Maintenance studies)

But not quite sure what you mean by " swapping pumps with their standby spares impacts reliability" though. Can you expand?

ultimately it becomes risk versus cost.

I know one company that commonly uses 2 x 50%. they work that on the basis that one unit will probably do 70% of the requirement so the losses are reduced. occasionally it hurts them, but overall they save money.

However things like pumps might be operationally more efficient as a single unit so overall costs including OPEX might be better with 2 x 100%

All in the detail....

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[pennpiper]

pumphead and MFJewell
Both of you left out any information about the Operation protocol for the different pump configurations.
Example:
2-100% Pumps - Alternate with each 100% pump on a 48 hour on and 48 hour off.
Pro: This tends to maximize your operation life for both pumps. It also gives you a 48hour for maintenance window for routine or simple repairs.
Con: this tends to be a higher initial installation cost

3-50% Pumps - Alternate pumps with two pumps on and one off on a 48 hour cycle
(A & B on w/C off, then A & C on w/B off, then B & C on w/ A off)
Pro: This tends to extend the working life of all three pumps
Con: This tends to require more space

2-50% Pumps - Both pumps run 100% of the time This might work good for a batch operation.
Pro: This tends to be less initial cost
Con: This tends to place your operation at risk

Sometimes its possible to do all the right things and still get bad results

 

[MFJewell]

Both of you left out any information about the Operation protocol for the different pump configurations.

I didn't leave it out of my analysis. I just didn't discuss it here. Instead I gave a brief example of what I have done without diving into details since he is asking about studies. To me, one would need real operational data with service intervals and maintenance costs (corrective and preventative), which would be after the equipment was in service for several years. My analysis was on the front end, so I made assumptions (since that data does not exist) based on maintenance schedules recommended by the manufacturers, anticipated operational profile of the unit, and maintenance outage durations and costs for other similar units.

 

[pumphead]

My interest would be in configurations where you have 2 X 100% sized pumps.

swapping pumps with their standby spares: Say you have 2 X 100% ANSI pumps that feed a process. To ensure the spare pump is available when needed every month operations shuts down the running pump and starts the idle pump. This also prevents bearings from brinelling, seals from sticking, tests the electrical power supply. The reason I hear most often is to ensure the spare pump will start.

On the other hand, motors send a high inrush current during starts which over time eat away at the life of bearings and couplings. If the pump swap is done manually, there is also risk of operator error. In some cases, the cost of swapping is too high and not worth it. So the counter argument I hear is you are shortening the life of both pumps, decreasing the reliability, and there may come a time when whichever is on standby mode does not start due to a failure induced by increased run life or too frequent starts.

And there are variations of this: A primary/secondary for instance. Where the primary pump runs for 3 weeks, then the secondary runs for 1 week. The intent here is to separate life so that they are not both due for major overhauls at the same time. Another version is to jog the idle pump or turn by hand. But with LOTO and work permits, this is not always practical. A third version is to swap equal run time once a week.

Various arguments for why one should and why one shouldn't alternate pump run time exists. I accept that each site has to consider their own priorities. Still, I was curious if anyone had actually studied the practice and shared their findings with the general public.

 

[EdStainless]

I know of people running 2x100% and they run the "A" pump twice as long as the "B" pump. They are running the "A" pump about 2 weeks between changes, but they always change over on Mon or Tues. That way if there is an issue with the other pump they can fix it during the week and make sure that it is functional before the next weekend.
You can start a centrifugal pump against a closed valve, so the switch over is fairly painless. In a plant it would be programed into the controls.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[JJPellin]

Our current practice is to swap pumps every 30 days when it is practical to do so. If both pumps are identical and have motor drivers, we run the A pump for 30 days then run the B pump for 30 days and so on. Our most common pumps tend to be API designs in the range of 50 to 500 horsepower. Most of our pump services are in pairs (2 at 100%).

We have other plants owned by our company that have extended the interval between pump switching to 60 or 90 days. We hired a consultant with 40 years experience with one of the largest oil companies in the US. They are emphatic that they improved their pump reliability by extending pump switching intervals to as much as 1 year in some services.

We are going to a 60 day switching interval currently and plan to extend further to 90 days in some services.


Johnny Pellin

 

[EdStainless]

JJ, in concept I understand what you are saying, but if a pump failure will shut down a unit would you really feel comfortable with your backup not having been run for a year? I could see 3 or 4 months maybe, but if it was critical service I would stick with more frequent swapping. The reason that you install 2x100% is for absolute certainty of backup, not to save money.
I find it hard to believe that if your pump will last for 5 years, that there is a significant difference between 6 starts and 60 starts. Maybe if we were talking about 600 starts it would start to have an effect. The cases that I know of where they switch every week or two are severe service and the pumps only last 2 years. I am sure that he size and style of pump is a factor, as well as the service. Heck in may cases the seals may well be the limiting factor.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[JJPellin]

Many of our pumps are 2 x 100% that are not in services that would shut a unit down. At this point I would not go beyond 90 days for switching critical pumps. But, for non-critical pumps in clean service, I will eventually consider going beyond 90 days.

Johnny Pellin

 

[Valvecrazy]

I don't know about studies, but I can tell you from 40+ years experience of fixing pumps like this, alternating pumps equally wears them out equally. When I am called to fix a pump that alternates equally with the other pump (2X100%), I know I have very little time to repair the pump, as the other pump is in EXACTLY the same condition. I prefer 2X100% with one pump running 24/7, and the other set up to work with what I call an exercise clock. Starting and running against an almost closed valve, I force pump #2 to come on every Monday at 10AM and run until 10;05AM. I like this time as there is usually somebody there when it happens to witness it. Since the pumps are controlled by pressure reducing type valves, pump #1 will just have its flow reduced by the same amount that pump #2 is allowed to flow for those 5 minutes. When the 5 minutes is up, pump #2 is shut off and pump #1 goes back to supplying 100% flow. Pump #1 never shuts off, and pump #2 starts and stops against and almost closed valve, which is easy on the pump and doesn't spike the system pressure.

I have found that pump #1 actually last longer because it is running 24/7/365. And pump #2 is still like brand new 20 years from now when I finally need a backup pump. I no longer have to be in a hurry to repair pump #1, as pump #2 is in good shape and makes a much more reliable backup.

On municipal type systems I like 2 X 100% plus a jockey of about 10-20% capacity. As these type systems are at low flow conditions a lot of the time, the jockey pump greatly increases the life of pump #1. Along with using the exercise clock for pump #2, this makes a very dependable system.

 

[EdStainless]

JJ and VC, I hear you.
Where I have seen pumps alternated they usually run one of them 2-5 times as much as the other. That way they don't wear out together, and you can prove that they both work. I like to see the second pump actually doing the job so that you can get accurate motor readings, flow readings, and vibration numbers. Better to find something wrong when you know that you can switch back to the other pump.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[JJPellin]

In our oil refinery, we rarely see pump failures that I would describe as “wear out” failures. By far, the most common single failure mode is mechanical seals that leak. Some of these failures have a time element in the cause. The formation of coke in hot services would tend to be affected by time in service. Most of the other failures would be classified as random or not related clearly to operating hours. Bearings tend to fail from excessive lubrication, inadequate lubrication or contaminated lubrication. We use non-lubricated couplings which should not wear out in a conventional sense.

I would feel more trust in the reliability of a pump which has run 24/7 for five years than a pump which set idle for five years other than 15 minutes of operation against a pinched valve once per week.

I have spoken to engineers at refineries that extended pump switching beyond 30 days and experienced increased pump reliability. Some of the pacesetters in the Solomon study are switching their pumps less often. I do not have study results to back up this position. But, the evidence I have seen is sufficient to drive me to extend switching intervals in our refinery.


Johnny Pellin

 

[TugboatEng]

I love this question. I feel that standby pumps need to be of the same duty as duty pumps or better. For example, in salt water, if a duty pump is bronze and the standby cast iron how can you be sure the iron pump is ready to operate when just sitting stationary is enough to cause it to fail. That is true for all pumps. They need to run regularly to keep things fresh and to verify things are fresh. Alternating duty of your pumps is the best way to keep your plant reliable.

 

[EmmanuelTop]

If you have an offline/standby equipment, failure-finding tasks (including functional tests as described by Valvecrazy) should be set based on the required availability (= risk tolerance for not be able to start on demand). Every RCM manual, including those published on the internet (i.e. ABS, NASA etc.) have guidelines on determining failure-finding maintenance task interval, either through reliability modeling or by using general guidelines.

Interesting point on having the standby pump in offline mode ALL the time. It is a novel concept to me, but it does sound intriguing. I was ready to bet on my 15 years of experience that keeping the standby pump out of service longer than 2-3 months will definitely cause problems like shaft bending, seal deformation, corrosion in some cases etc.

Dejan IVANOVIC
Process Engineer, MSChE

 

[lilliput1]

If you are concerned about reliability of pumps with spare consider adding vibration sensor/alarm on each pump and temperature sensor/alarm on each motor.

 

[pumphead]

Thank you to everyone who has participated in this thread. It helps me put together guidelines in developing a policy for my sites.