Centrifugal Pump Condition Monitoring Standard 2

[VaughnQ]

I am developing a standard of condition monitoring for centrifugal pumps, inclusive of visual inspections and vibration analysis. Does anyone have any such standard document etc??

 

[DiegoMartinez]

There is no specific standard you can develope. The company you contract with to do the equipment condition monitoring will advise if the equipment vibration (Pump, fin fan cooler, turbine, etc.) is satisfactory or not; I mean within acceptable range or not. They are the specialist. Normally equipment vendor should indicate the vibration range in the operation manual.

 

[JJPellin]

We have standards of the type that you are describing. Unfortunatly, I am not allowed to send those out as they are controlled documents. However, I can describe the basic elements. We include pump and driver condition monitoring as part of our Mechanical Integrity program as required by OSHA 1910.119. We specify that we will take portable vibration data using accelerometers and hand hand data collectors on each pump once per month for running machines. For machines which do not run, we take data a minimum of once every three months. This requires our operators to run their spares at least this often so we can get data. For each machine, we take two radial readings at each bearing location (vertical and horizontal) and axial readings at each bearing housing. We take special data sets on motors (to distinguish run speed and line frequency). We have a standard that defines the starting point for the alert and danger alarms. But these levels are adjusted based on baseline data following an overhaul. We take post repair data following all overhauls. We take data on all spares when the main pump is out for maintenance. We take pre-repair data on all machines that are going to be taken out for repair to make sure we are not missing something in the scope. For example, if a pump is out for a seal leak, the vibration techs will include scope notes if we need to replace the bearings or balance the impeller at the same opportunity. We do not have such a clear program when it comes to visual inspection. We used to have an inspector check the case and head on every pump we worked on and generate a report. Now, we ask the mechanics to inspect case and head and if they see anything that concerns them, they request an inspector to look at it.

 

[svanels]

www.hq.nasa.gov/office/codej/codejx/Vibration_Spec.doc is in the public domain.

Basically it has guidelines on acceptance criteria of new and overhauled equipment as used by Lockeed Martin Space Systems.
There are other similar public domain standards like the General Motors Vibration Standard which is almost the same.
It gives ideas about numbering and placement of mesuring points.

Alarm limits, you have to develop yourself

 

[lvq78]

Svanels, the link looks very hopeful, but its dead. Is there any other?

 

[svanels]

Looks like the link is broken

Maybe this thread from maintenanceforums helps

http://maintenanceforums.com/eve/forums/a/tpc/f/3751089011/m/6991081052?r=6821082052#6821082052

 

[DiegoMartinez]

JJPellin,

You said:

For machines which do not run, we take data a minimum of once every three months. This requires our operators to run their spares at least this often so we can get data.

Could you expand it abit more? I'd like to know what your approach is to do vibration analysis on offline machines.

 

[JJPellin]

Each unit in the plant has a program for running their spare machines. In one unit, for example, they run their spares the firth Thurdsay of every month. When that day comes, the go out in the morning and start up the spare pumps and switch off the main pumps. They let the spares run for a minium of 4 hours to give them a chance to get up to temperature and stabilize at normal operating conditions. Later that day, the vibration technician goes around and takes vibration data on those spare. The pump switching is initiated by a program we call "routine duties calendar" which is simplet caledar program that puts out a list of duties for each shift for things that need to be done daily, weekly or monthly. Our vibration technicians have a schedule that tells them when each unit is doing their switching so they know when to go to get data on the spares. If a pump is not run as part of the normal spares switching, and we reach 3 months without getting data, the technician will contact the control room and specifically request that they run that pump. There are very few pumps that they are not willing to run for us. Certain ones pose too great a risk to the process and these are exempted from the program. Out of 1400 centrifugal pumps, we usually only have 8 or 10 that are missed each month using this program. And those are usually offsites pumps at storage tanks that do not run on any set schedule for only as needed.

 

[svanels]

JJPelin what is your experience with these standby pumps? Do they have a higher failure rate, compared to the duty pump. I am especially interested in pumps with hot viscous products, mechanical seals, barrier fluids etc.
I am principally intersted in mechanical seal failures.

 

[JJPellin]

My experience would suggest that spare pumps are often less reliable that the main pump that runs all time. Since most of the failures that I see are not wear-out failures, other mechanisms are more important. Specifically in hot, heavy products, there are a number of potential issues that result in higher failure rates among the spares. First, if the product can thicken up or set up between the seal faces, then failure can result on start-up. It can be tricky to keep the product hot enough to stay fluid but not so hot that it cracks and cokes. Services like coker charge have this problem especially. Our standby coker charge pumps are all turbine driven and we have them keep those turbines on slow roll. This keeps the product in the pump moving so it doesn't set up. It keeps the fluid between the seal faces moving so they don't set up or coke up as badly. Slow rolling is not good for the turbine (in general) but we do it for the sake of the pump and mechanical seal. Standby pumps in asphalt service are tricky in this regard, too. Offsites pumps at storage tanks set idle for much of the time but have to run on occasion to transfer product. In order to keep the pumps and seals hot, we use steam tracing, steam jacketing and steam quenches. These seals have jackets built into the seal glands. Steam is passed through the jackets and then is injected on the atmosphere side of the seal as a quench. This keeps the asphalt from setting up between the faces when the pump is idle and helps prevent coking on the ID of the seal faces when the pump is running. The rest of the pump case is also jacketed, then wrapped in steam tracing and insulated to keep it hot.

If a pump sits idle and never runs for a long period of time is likely to be among the least reliable pumps in the plant. A pump that runs continuously at a good operation point, with good lubrication and seal support piping is likely to be among the most reliable pumps in the plant. We run our spares to check them for problems (performance or vibration) and to keep them reliable (circulate the oil, rotate ball bearings to discourage false brinnelling, etc.).

 

[svanels]

My experience would suggest that spare pumps are often less reliable than the main pump that runs all time

JJPelin you just made my day, several years ago we abandoned a practise of switching pumps every week for 1 day. The advise of "old and wise operators", which could have been valid in the days of packing glands, but is complete disaster for mechanical seals. Guess what happened, our failure rate fell significantly. In your case the switch every month, still is not giving satisfactory results, there is room for improvement.
Currently we (maintenance) are trying to implement an operating protocol, where we want to run the duty pump for 3 months (12 weeks), switch to the backup pump for 3 weeks, so we have enough time to monitor the beheaviour of main and backup pump without unnecesarry start-stop or too short operation cycles.
This comes down to a 88-12% operating rule.
I expect too have hard evidence a year after implementation, since now we are coming out of the experimenting phase with our vibration monitoring equipment and set the layout for a durable PdM strategy.

 

[DiegoMartinez]

JJpelin & Svanels,

Thanks guys for the interesting discussion. JJpelin, do you mean you run the spare pump for one day per calender month (every first thurseday of each months)? Svanels, I am interested to know about your operation protocol, obviously if possible.

The thing is we are getting a company to do vibration analysis survey on our equipment every three months. What do you think on how we can get the most benefit from their servise/ or what is the best effective way we can get them to do the survey? I am thinking of scheduling the pumps operation so that one runs for 10 weeks, being monitored and surveyed, being switched off and the other pump get started the day after, being surveyed and being kept running for 2 weeks and gets monitored during these period and we follow up this cycle of 10 weeks, 2 weeks again!

By the way, we have 2 turbine driven compressors and 2(A&B)x2 fin fan coolers. When one compressor is running, exchanger A is operating. How can we efficiently run those exchangers in order to:
1- be able to monitor them and ensure they are in good condition and can be swapped over if needs be.
2- How often we'd better swapping them over.

Appreciate your recommendations.

 

[DiegoMartinez]

By the way, in our plant the guys tend to run the main and spare pumps in a way that they have close total working hours at any specific time! Is it a good way of maintaining them?

 

[svanels]

Mebrahim, you are exactly on the same path, I am building the framework to do this in a structured manner, it starts with getting the approval from the equipment owners, which can be the hardest part, they don't like to be told how to run their equipment, I would like to invite you and jjpelin to join

http://maintenanceforums.com

were this item is a hot discussion point right now.

 

[JJPellin]

Yes, we normally have them run the spare pumps for one day (or at least 4 hours, anyway) once per month. This tends to be the normal method. But I should explain why. There is a certain risk in starting and stopping process pumps in an oil refinery. We don't want to have excessive starts because of this risk. Most of our pumps are set up in pairs with one motor driver and one turbine driver. They usually have a strong preference to run one versus the other. This is sometimes because the turbine has an auto-start feature that the motor does not. Running the turbine places them at risk. If the turbine were to trip, they would shut the unit down before they could respond. In other cases, there are steam balance issues the make one driver preferable over the other. If we did not have these differences, my preference (and we do this on some pumps) would be to switch pumps once per month and not switch back until the following month. Then each pump gets equal run time and we have half as many pump switches to minimize risk. I think that is ideal.

Regarding testing frequency, I have some comments. We test our pumps once per month if they are running continuously. We do this based on good data that suggests that the average time between when a bearing failure becomes detectable and when it goes catastrophic is about 2 months. By testing once per month, in theory you should be able to catch the vast majority of the bearing failures in the early stages (before sparks are shooting out). If you test every three months, there is a good chance that a bearing could begin to fail the day after you test and go catastrophic before you test again. Full blown vibration analysis using a contractor could be done every three months. But I would also suggest getting your operators a simple hand held meter to take overall levels on the pumps at least monthly so you are somewhat protected in between.