We are evaluating the ideal locations for installing wireless triaxial accelerometer vibration sensors on a group of horizontal and vertically installed electrical motor driven pump assemblies (between 15 to 150 KW power ratings)for a cargo vessel.

We understand that ideally speaking, one would install one senor at each bearing location i.e. at the motor NDE, DE and one at the pump bearing end.

Due to cost and other constraints we shall have to restrict ourselves to one sensors max per machine.

What could be the ideal location for such a configuration, without losing the capability of reliable monitoring of the pump operation to prevent failure.   Kindly advise.

Original Post

Centrifugal pumps have multiple areas of interest for monitoring in a PdM setting.

-volute.
-stuffing box/seal area which is the junction that prevents fluid in the pump from coming out of the pump between the volute and pump shaft.
-last but not least, the bearing case, which can be monitored to detect bearing failure.
-For horizontally mounted Centrifugal pumps the sensor should be installed perpendicular to the shaft on the bearing case. On a vertically mounted pump two sensors should be installed 90 degrees from each other and perpendicular to the shaft on the bearing case. For axial measurements for a vertically mounted pump, a sensor can be installed on a location near the pump casing.

 

if you have any additional questions, please don’t hesitate to reach out. 

We are evaluating the ideal locations for installing wireless triaxial accelerometer vibration sensors on a group of horizontal and vertically installed electrical motor driven pump assemblies (between 15 to 150 KW power ratings)for a cargo vessel.

We understand that ideally speaking, one would install one senor at each bearing location i.e. at the motor NDE, DE and one at the pump bearing end.

Due to cost and other constraints we shall have to restrict ourselves to one sensors max per machine.

What could be the ideal location for such a configuration, without losing the capability of reliable monitoring of the pump operation to prevent failure.   Kindly advise.

Quite limiting, one per machine.  But with that restraint, I would put one at the pump drive end.

Dear Jake,

Thanks for your swift response. Much appreciated.

Good idea about moving sensors if attached magnetically - the system in this instance shall be on board a sailing merchant ship and the onboard engineers are not trained for this and will not know how to take reliable readings. The intention is to establish a fixed sensor system with no human intervention. 

As we shall be installing a triaxial sensor  - would 2 sensors perpendicular to the pump shaft be mandatory for reliable monitoring? 

Intention is to monitor both the electrical motor bearings and pump performance for deterioration from existing values and be able top predict an imminent failure with possible root cause to allow remedial repairs/maintenance in time. 

Accelerometers are mounted perpendicular to the shaft rotation on the pump bearing housing, as close to one bearing as possible. An axial measurement can be made near the pump casing. While mounting both a vertical and horizontal sensor is best, if only one direction is to be monitored, the horizontal direction is usually preferred. Vibration monitoring of horizontally mounted centrifugal pumps can detect problems related to soft foot, cavitation, balance, alignment, flow and bearing wear. 

Have you monitored the pumps previously?

Dear John,

Thank you for your response. Much appreciated.

Yes agree it is very limiting. Would one triaxial sensor at pump bearing end (for horizontal pumps) also manage to keep tabs (in a manner of speaking) on the electrical motor bearings too??

Most pump sets are with rigid or flexible flange couplings and some of the vertical ones do not have any bearings on the pump side. 

I know this appears to be asking for much but am simply trying to understand if basic monitoring to predict a major operational change can be captured to provide a reliable condition monitoring program to the vessel. 

Accelerometers are mounted perpendicular to the shaft rotation on the pump bearing housing, as close to one bearing as possible. An axial measurement can be made near the pump casing. While mounting both a vertical and horizontal sensor is best, if only one direction is to be monitored, the horizontal direction is usually preferred. Vibration monitoring of horizontally mounted centrifugal pumps can detect problems related to soft foot, cavitation, balance, alignment, flow and bearing wear. 

Have you monitored the pumps previously?

Well noted. No the pumps have not been monitored before - we are entering into a new system architecture for a ship, wherein client wishes to start a monitoring program and has shackled us with limitations due to cost constraints. 

So establishing existing base line values will be our first job before we start making comparative  evaluation. Plus this being on a vessel  - the externally induced vibrations will have to be sifted out during initial evaluation.

You minimum need one horiz on pump DE and one on the motor and that may as a compromise be on the mid lift hook position to monitor both bearings and rather have 2 single axis than 1 triax if that is lower cost. Don't trust the coupling to transfer any signal. I would not use a magnet mount on a ship. Only what I would do and there may be another better way. You need as minimum supply the crew with a "yellow handle analyzer" eg. a screwdriver and ears to confirm the data you receive and position of problem eg. bearing fail.

@OLI posted:

You minimum need one horiz on pump DE and one on the motor and that may as a compromise be on the mid lift hook position to monitor both bearings and rather have 2 single axis than 1 triax if that is lower cost. Don't trust the coupling to transfer any signal. I would not use a magnet mount on a ship. Only what I would do and there may be another better way. You need as minimum supply the crew with a "yellow handle analyzer" eg. a screwdriver and ears to confirm the data you receive and position of problem eg. bearing fail.

Thanks OLI,

Noted - atleast one sensor on pump DE and another possibly mid point between 2 bearings.

Didnt know that couplings could be such dead ends for vibration transmission. noted on that.

Magnet is defintely out - shall used screwed connections to the extent possible or glued with Loctite where not feasible.

Good question about using single axis - am of the impression that a triax would give a better analysis potential for fault detection as opposed to single axis? 

Like your idea of the yellow handle analyser :-) - though many in the field has lost that skill set lately it might appear.

@fburgos posted:

I would do one accelerometer per bearings, why triax, wireless IOT?

 

If you're budget tight wireless can wait

fburgos,

Good point -

Ideally I agree best to do one accelerometer per bearing.

Maybe I didnt clarify properly - this is a sailing cargo ship and we are looking at retrofit installation on multiple projects on various vessels if we can convince the client. They are all sailing in different parts of the world globally and the wired solution would require serious wiring over long distances like 200 - 300 mtrs for some equipment on the ship with multiple metal bulkheads in between and high voltage power cables running across which could interfere with the vibration sensor signal quality.

The wired option would work out even more expensive as the installation team would have to fly to the port where the vessel is and sail with the vessel to some other ports to complete the project and fly back again.

Hence believe that the wireless option would give an overall better solution.

for the triax - isnt this better than a single axis?

I would have one transducer per bearing so I can evaluate bearing failures that is the most common due to lack of greasing. Next worn bearings due to just worn out. So I rather monitor all bearings than over evaluate a few bearings for no good. In my view single axis for on-line is good enough unless it's free with 3-axis. If you guide the chief to listen to a good bearing and then one that your data indicate as bad he will learn in a few months...... In a ship you have doors that should be shut and that split up radio transmission and if machines are spread in several rooms as they should it may be a problem. Just FYI. I have a few instruments on tankers and cruise ships but they are handheld and require a person able to read a manual. I have also collected data on a few ro-ro ships and Swedish Navy vessels, majority found were bearing faults........

@OLI posted:

I would have one transducer per bearing so I can evaluate bearing failures that is the most common due to lack of greasing. Next worn bearings due to just worn out. So I rather monitor all bearings than over evaluate a few bearings for no good. In my view single axis for on-line is good enough unless it's free with 3-axis. If you guide the chief to listen to a good bearing and then one that your data indicate as bad he will learn in a few months...... In a ship you have doors that should be shut and that split up radio transmission and if machines are spread in several rooms as they should it may be a problem. Just FYI. I have a few instruments on tankers and cruise ships but they are handheld and require a person able to read a manual. I have also collected data on a few ro-ro ships and Swedish Navy vessels, majority found were bearing faults........

Ideally I would agree, and that is something I intend to push - one sensor per bearing.

If there is a good Chief engineer on board - many of the monitoring system porblems are obviated - its just on some vessels there is very little or no attention paid to this system due to various reasons.

With regards to water tight doors and bulkheads - I am banking on info from some makers that with the new bluetooth technology - the new wireless sensors using BLE 5.0 wireless system can actually transmit over larger distances and through metal partitions too. Shall have to add gateways to strengthen the signal transmission/reception capabilities of the system.  

I didn't know about LoRa, so I did a quick search:

https://www.link-labs.com/blog/lora-faqs

There are a lot of issues for getting good data from a measurement point on a machine on a ship 1/2  world away to a central office/analysis location. I have done a lot of machine surveys and diagnostic tests on various types of vessels in various situations (underway, dock, dry dock, and remote underway), and I can say it is a challenge!

The chain of data usefulness and quality starts with the sensors and the measurement location. A single triaxial accelerometer per machine or one per component (say motor and pump) may not meet your monitoring program goals. It would be wise to start with very specific program goals that consider individual machine component common failure modes and fault detection time to allow corrective actions. The issues you don't know about may far exceed the issues you are aware of!

Walt

A ship propulsion is basically a small powerplant in a confined space one way or another and we have been monitoring these type of plants for 50+ years using some kind of overall values and a 4-20mA signal to the PLC and here it is still like that if it is 1 signal per bearing do vary more and more. I have seen some very strange optimizations lately, like 2 on a pump or a fan and no on the motor.
So for monitoring a basic parameter or 2 and temperature would be a minimum a FFT of some sort would be of great added value in my view. 
Regarding the transmission of signals we have used WiFi with good success in industrial environments and that is these days about the same frequency range as BT. If it goes thru closed doors on a ship I doubt but it goes thru 3 stories shaft in a hydroplant and a close fire door anyway but not 50m and 2 fire doors in a gasturbine site, just what I tested. LoRa do technically rely on a separate backbone that would have to be supplied locally on the ship as it still can't be depended on that it would exist and reach inside a ship. So basically you need a interface to the ship network and use that internet connection that the ship have and that may be tricky but there are examples where it has been done like off shore in support vessels in the North Sea. So everything is possible, good luck. You can also wait for Elons world satellite wide network and just hook up to that. 

 

 

My 2 cents:

1. You were very specific of using a wireless triaxial accels on pumps and limiting to one per machine. But ask yourself why one location when you can easily mount on all the bearing housings?   As a few have suggested, why not move the accel to each bearing point and get data from the whole machine for that day?   Does it have to be continuous monitoring or periodic data evaluation?

2. Why limit to just wireless accel? Have you considered epoxying  mounting discs and using a portable data collector with a single triaxial accel and get data from each bearing housing?  IMO, It is simple to train personnel to take meaningful data at all the important points on a machine; perhaps just once every few weeks or so

3. You probably already know this, but some of the most important parameters on a pump are its flow rate, discharge pressure and suction pressure. Whenever I analyze pump problems those are the FIRST parameters I ask for (along with the pump curve).

Regards

Jim P

 

Ideally I would agree, and that is something I intend to push - one sensor per bearing.

If there is a good Chief engineer on board - many of the monitoring system porblems are obviated - its just on some vessels there is very little or no attention paid to this system due to various reasons.

With regards to water tight doors and bulkheads - I am banking on info from some makers that with the new bluetooth technology - the new wireless sensors using BLE 5.0 wireless system can actually transmit over larger distances and through metal partitions too. Shall have to add gateways to strengthen the signal transmission/reception capabilities of the system.  

Have you looked into ZigBee networking technology?  It has been found to be quite useful for shipboard applications.  See https://www.mdpi.com/1424-8220/20/6/1668/pdf for some insight into the technology.

Last edited by John from PA

We currently use LoRa transceivers with our sensors that we sell, and I think for this application, BLE may not be the best bet. I’ve attached a graph for a good reference.

 

Thanks Jake - whilst we can increase gateways if needed - penetration through metal bulkheads would be of more rlevance as opposed to range - especially in a complete steel cage scenario existing on board vessels.

@Walt Strong posted:

I didn't know about LoRa, so I did a quick search:

https://www.link-labs.com/blog/lora-faqs

There are a lot of issues for getting good data from a measurement point on a machine on a ship 1/2  world away to a central office/analysis location. I have done a lot of machine surveys and diagnostic tests on various types of vessels in various situations (underway, dock, dry dock, and remote underway), and I can say it is a challenge!

The chain of data usefulness and quality starts with the sensors and the measurement location. A single triaxial accelerometer per machine or one per component (say motor and pump) may not meet your monitoring program goals. It would be wise to start with very specific program goals that consider individual machine component common failure modes and fault detection time to allow corrective actions. The issues you don't know about may far exceed the issues you are aware of!

Walt

Thanks Walt - fully understood - this is very good feedback - intend to discuss critical equipment set up with high failure rate for a 2 - 3 sensor installation per machine. 

@OLI posted:

A ship propulsion is basically a small powerplant in a confined space one way or another and we have been monitoring these type of plants for 50+ years using some kind of overall values and a 4-20mA signal to the PLC and here it is still like that if it is 1 signal per bearing do vary more and more. I have seen some very strange optimizations lately, like 2 on a pump or a fan and no on the motor.
So for monitoring a basic parameter or 2 and temperature would be a minimum a FFT of some sort would be of great added value in my view. 
Regarding the transmission of signals we have used WiFi with good success in industrial environments and that is these days about the same frequency range as BT. If it goes thru closed doors on a ship I doubt but it goes thru 3 stories shaft in a hydroplant and a close fire door anyway but not 50m and 2 fire doors in a gasturbine site, just what I tested. LoRa do technically rely on a separate backbone that would have to be supplied locally on the ship as it still can't be depended on that it would exist and reach inside a ship. So basically you need a interface to the ship network and use that internet connection that the ship have and that may be tricky but there are examples where it has been done like off shore in support vessels in the North Sea. So everything is possible, good luck. You can also wait for Elons world satellite wide network and just hook up to that. 

 

 

OLI - Intend to try it out the wireless triaxial set up on a vessel to check if it works well before rolling it out for more vessels. Thanks for your feedback - quite interesting.