Vibration is a wide subject area that has continued to attract research over the years because of its import in almost every facet of day-to-day life. From aircraft modelling to building design in earthquake prone regions of the world; from asset condition monitoring in various industrial plants to wheel balancing and alignment at the local car garage, an application of the knowledge of vibration principles can be observed.
The vibration of an object can simply be referred to as the oscillatory movement of that object about a mean equilibrium position. The motion is brought about by the application of some force or excitation. Common examples of this phenomenon include the motion of plucked guitar strings, the motion of tuning forks, and the shaking felt at the station floor when a train passes, the shaking observed on the road when a heavy truck passes or the rattling of the road workers' impact hammer. Some vibration however is not as pronounced as the examples given. For instance houses and bridges vibrate as well. Normally, these vibrations cannot be detected by merely looking. When the excitation is large enough, the vibrational motion can be seen and this would probably result in the collapse of the structure.
From the foregoing, it can be seen that some vibration is useful while some is destructive. The failure or destruction brought about by vibration is not an immediate one. Vibration ultimately leads to a fatigue failure and this should be of interest to the operators and maintainers of plant equipment.
For the maintenance engineer or asset management practitioner, these facts about vibration can be translated to intelligent and informed decisions for industrial plants. With numerous different devices, vibration levels on plant equipment can be detected, measured and recorded. It is possible to analyse the data collected to determine the condition of an asset and even predict an imminent failure. All rotating or stationary plant equipment have acceptable vibration levels stipulated by standards such as the International Organization for Standardization (ISO) Codes or developed in-house by Vibration Monitoring Engineers. A deviation from the acceptable vibration levels is an indication of the onset of deterioration which is undesirable. This knowledge gives insight into the actual condition of a piece of equipment, eliminates guesswork and enables the maintenance engineer to plan a fitting response to the asset's deterioration. Such control and planning can lead to a huge saving in maintenance costs, prevent unnecessary downtime, improve safety and performance for any asset.
Furthermore, vibration monitoring and analysis incorporates some advanced techniques for the discovery of the root cause of frequent machinery failures. Applying these techniques can mean the difference between constant breakdowns and good asset availability.
It is the objective of this report to provide a clear cut proposal as to how to solve the problem of the frequent failures of the Yoho High Pressure (H.P.) Flare Scrubber Pumps by the application of these advanced vibration analysis techniques and also provide a method for monitoring the condition of the pump to pre-empt any deterioration that can spring surprises. This will ensure a reduction in present maintenance costs and alleviate the work burden of maintenance personnel whose time is tied up with the care of these pumps.
Case Study and problem Description
The Yoho H.P. Flare Scrubber pumps are vertical, turbine, submerged pumps that transfer recovered liquids from the Flare Scrubber vessel to the main crude oil production header on the Yoho offshore oil and gas installation. The pumps are submerged in barrels that receive feed from the H.P. scrubber vessel by gravity. The recovered liquid is a mixture of water and light crude oil. There are three pumps on skid and these pumps are critical to plant operations. One pumps runs at a time and an additional pump or the other two pumps can be put in service, if the need arise due to an increasing level of liquid in the H.P. Flare Scrubber. In the event of an outage of all three pumps, the plant could lose production to the tune of one hundred thousand barrels of crude oil per day (4100 barrels/hour).
The pumps were commissioned in 2006 at plant start-up and have had numerous failures since. They have proven unreliable and presently a contractor's pumps (which are portable and of a different design), are relied upon to perform their function. There have been occasions when all three pumps are in a state of disrepair.
These pumps usually run smoothly for a while and then become noisy, vibrate and finally fail. After a pump fails, it is taken out of the hole, sent to the shore-base for repair, sent back to the platform for reinstallation and then reinstalled. Investigations of the numerous failure cases reveal that the pump bearings, riser liners and bushings have worn out given room for shaft play, impeller damage and mechanical seal failure. Installation, repair and rebuild procedures have been scrutinized and checked for quality. This has not yielded any dividends however, as the pumps keep failing after two or three months in service.
Fig. 1. overleaf shows a cut-away section of the pumps. The diagram is supplied by the manufacturers but is simplified as the actual H.P. Flare Scrubber pump has six impellers, a top column of 11inches length, two intermediate columns of 30 inches length, a bottom column of 30 inches length, and a pump bowl assembly of 36.25 inches length.
The length of the pumps makes it difficult for removal and installation and as such maintenance practitioners, plant operators and plant management have been constantly distressed by the frequent failure of these pumps.
Fig. 1. Gould Pump Model VIC-T (source ITT-Gould Product catalogue, [online] www.gouldspumps.com/pump_VIC.html [Accessed 6th May 2010]
Pump Specification
Manufacturer: ITT-Gould
Head shaft Length: 129 Inches
Head shaft Diameter: 1 Inch
Seal Method: Mechanical Seal
Drive: 40 HP Electric Motor
Differential Pressure: 200 Psi
Capacity: 180 Gallons per Minute
RPM: 3000
Impeller: 5 Vanes, Enclosed
Failure and Maintenance History
The table below shows the failures and some maintenance activities performed on the three H.P. Flare Scrubber Pumps over a two-year period. The rows 69, 71, 72, 90,91,97,98,102,103 and 110 show times when none of the three pumps were operational. The cost of this unavailability of the pumps is manifold. It ranges from the loss of production of about one hundred thousand barrels of crude oil per day, to penalties for non-compliance with environmental regulations and most importantly, safety.
S/N
DATE
EVENT/ACTIVITY
EQUIPMENT TAG
NO. OF PUMPS AVAILABLE
COMMENTS/ FINDINGS
1
11-Jan-06
HP Flare Scrubber Pump B Auto Operation problem
YP-G-180
3
2
12-Jan-06
Troubleshooting of LSLL-6207 on HP Flare Scrubber
YP-G-180
3
3
1-Mar-06
PM on HP Flare Scrubber Pumps A and B
YP-G-180A/B
3
4
11-Mar-06
Pumps Operation Started
YP-G-180A/B /C
Earliest recorded date of crude and produced water movement with YP facilities
5
5-Jun-06
Test-run HP Flare Scrubber Pump C with vibration group, QIT to determine cause of excessive vibration
YP-G-180C
3
Awaiting result
6
7-Jun-06
Investigating cause of excessive vibration on HP Flare Scrubber Pump C
YP-G-180C
3
Vibration traced to damaged Mech. Seal
7
9-Jun-06
Removed HP Flare Scrubber Pump C for repairs
YP-G-180C
2
8
9-Jun-06
Unblocked strainers on HP Flare Scrubber Pumps A & B
YP-G-180A/B
2
9
12-Jun-06
Rectified poor discharge and noisy operation on HP Flare Scrubber Pumps A & B
YP-G-180A/B
2
10
19-Jun-06
PM on HP Flare Scrubber Pump A
YP-G-180A
11
3-Jul-06
Completed installation of HP Flare Scrubber Pump C
YP-G-180C
3
12
4-Jul-06
Removed HP Flare Scrubber Pump B for repairs in QIT
YP-G-180B
2
13
15-Jul-06
Remove and clean HP Flare Scrubber Pump C suction strainer
YP-G-180C
2
14
12-Aug-06
Remove and clean HP Flare Scrubber Pump C suction strainer
YP-G-180C
2
15
24-Aug-06
Remove and clean HP Flare Scrubber Pumps A & C suction strainer
YP-G-180A/C
2
16
27-Aug-06
Rigged out HP Flare Scrubber Pump A and transferred to loading bay
YP-G-180A
1
17
30-Aug-06
Rigged in HP Flare Scrubber Pump B after QIT repairs
YP-G-180B
2
18
2-Sep-06
Removed shaft coupling on HP Flare Scrubber pump B, removed motor and installed motor from pump A
YP-G-180A/B
2
Electricians investigating high voltage on motor
19
3-Sep-06
Transfer motor, pump and all associated parts from YP-G-180B to G-180A
YP-G-180A/B
1
20
4-Sep-06
Installed pump head, motor and shaft hub on HP Flare Scrubber Pump A
YP-G-180A
1
21
5-Sep-06
Carried out coupling runout checks, impeller lift ( Reinstallation) on pump A
YP-G-180A
2
22
8-Sep-06
PM on HP Flare Scrubber Pump C
YP-G-180C
2
23
11-Sep-06
Remove and clean HP Flare Scrubber Pump C suction strainer
YP-G-180C
2
24
29-Oct-06
Bleed off gas from HP Flare Scrubber Pump C suction and discharge line
YP-G-180C
2
25
12-Nov-06
Replacement of Mech. Seal on Pump C
YP-G-180C
1
26
13-Nov-06
Removed HP Flare Scrubber Pump C for repairs in QIT
YP-G-180C
1
27
28-Nov-06
Reinstallation of Pump C
YP-G-180C
1
28
2-Dec-06
Completed installation of HP Flare Scrubber Pump C and test-ran it.
YP-G-180C
2
29
21-Dec-06
PM on HP Flare Scrubber Pumps A and C
YP-G-180A/C
2
30
6-Feb-07
Commenced installation of pump B
YP-G-180B
2
31
7-Feb-07
Continued installation of pump B
YP-G-180B
2
Awaiting Mech. Seal
32
24-Apr-07
Installation of Mech. Seal on Pump B
YP-G-180B
2
33
25-Apr-07
Completed Mech. Seal Installation of Pump B
YP-G-180B
3
34
28-Apr-07
Troubleshoot overload trip fault on Pump B
YP-G-180B
3
35
30-Apr-07
Rechecked alignment of Bump B
YP-G-180B
3
36
1-May-07
Rectified overload trip on Pump B- cleaned inlet strainer
YP-G-180B
3
Discharge line was filled with sand and sludge
37
14-May-07
Mech. Seal replacement on Pump C
YP-G-180C
2
Shaft worn around Mech Seal
38
15-May-07
Rigged out HP Flare Scrubber Pump C for repair at QIT
YP-G-180C
2
39
27-May-07
Troubleshoot high vibration on Pump B
YP-G-180B
2
40
30-May-07
Removed HP Flare Scrubber Pump B for repairs in QIT
YP-G-180B
1
41
1-Jun-07
Assessed materials for stiff valves associated with pumps
YP-G-180A/B /C
1
42
3-Jun-07
Reinstallation of Pump C
YP-G-180C
1
43
6-Jun-07
Completed Pump C Installation
YP-G-180C
2
44
7-Jul-07
Reinstallation of Pump B
YP-G-180B
2
45
10-Jul-07
Completed Pump B Installation
YP-G-180B
3
46
13-Jul-07
Troubleshoot frequent tripping of Pump B
YP-G-180B
3
Electric motor problem. This was fixed
47
21-Jul-07
Rigged out HP Flare Scrubber Pump A and transferred to loading bay
YP-G-180A
2
48
22-Jul-07
Rigged out HP Flare Scrubber Pump B and installed in A barrel
YP-G-180A/B
1
Pump B becomes Pump A
49
23-Jul-07
Alignment and commissioning of Pump A
YP-G-180A
2
50
26-Jul-07
Remove and clean HP Flare Scrubber Pumps A & C suction strainers
YP-G-180A/C
2
51
10-Aug-07
Rigged out HP Flare Scrubber Pump A and transferred to loading bay
YP-G-180A
1
52
17-Aug-07
Commenced installation of pump A
YP-G-180A
1
53
18-Aug-07
Continued installation of pump A
YP-G-180A
1
54
22-Aug-07
Alignment and commissioning of Pump A
YP-G-180A
2
55
11-Sep-07
PM on HP Flare Scrubber Pump A
YP-G-180A
2
56
17-Oct-07
Removal and relocation of Pump A to Pump B barrel
YP-G-180A/B
2
Pump A becomes B
57
21-Oct-07
Investigate low discharge pressure on Pump B
YP-G-180B
2
58
22-Oct-07
Replacement of Mech. Seal on Pump C
YP-G-180C
2
59
23-Oct-07
Rigged out HP Flare Scrubber Pump B and transferred to loading bay
YP-G-180B
1
60
24-Oct-07
Rigged out Pump A barrel for leak repairs by FMS
YP-G-180A
1
61
25-Oct-07
Rigged in Pump A barrel after leak repairs by FMS
YP-G-180A
1
62
28-Oct-07
Commenced installation of pump A
YP-G-180A
1
63
29-Oct-07
Completed installation of pump A
YP-G-180A
2
64
31-Oct-07
Remove stiff hub from Pump B Motor
YP-G-180B
2
65
3-Nov-07
Remove Motor to W/Shop to drill and tap broken bolts of Motor Fan Cover
2
66
29-Dec-07
Remove damaged pump C and commenced installation of refurbished pump
YP-G-180C
1
67
30-Dec-07
Rig out Pump C due to lack of keyway on shaft
YP-G-180C
1
68
1-Jan-08
Transferred bad pump to loading bay for QIT W/Shop
1
69
8-Jan-08
Removed HP Flare Scrubber Pump A for repairs in QIT
YP-G-180A
0
70
9-Jan-08
Installed Flare Scrubber Pump A
YP-G-180A
1
71
10-Jan-08
Removal of Pump A
YP-G-180A
0
Pump stiff.
72
12-Jan-08
Commenced installation of pump A
YP-G-180A
0
73
13-Jan-08
Completed installation of pump A
YP-G-180A
1
74
27-Jan-08
Rigged in Pump B and commissioned same
YP-G-180B
2
75
4-Mar-08
PM on HP Flare Scrubber Pumps A & B
YP-G-180A/B
2
76
13-Apr-08
Removal and relocation of Pump A to Pump C barrel
YP-G-180A/C
1
Pump A barrel leaking crude from cut bleed line. Pump A becomes C
77
14-Apr-08
Commenced installation of pump C
YP-G-180C
1
78
15-Apr-08
Completed installation of pump C
YP-G-180C
2
79
17-Apr-08
Rigged out HP Flare Scrubber Pump B and transferred to loading bay
YP-G-180B
1
80
81
18-Apr-08
Removal of debris from Pump barrel
YP-G-180B
1
82
21-Apr-08
Commenced installation of pump B
YP-G-180B
1
83
22-Apr-08
Completed installation of pump B
YP-G-180B
2
84
9-May-08
Commenced rigging out Pump C
YP-G-180C
1
85
10-May-08
Completed rigging out Pump C
YP-G-180C
1
86
17-May-08
Commenced installation of pump C
YP-G-180C
1
87
18-May-08
Completed installation of pump C
YP-G-180C
2
88
20-May-08
Rectified failure to lift on Pump B
YP-G-180B
2
89
23-May-08
Rigged out HP Flare Scrubber Pump B for QIT repairs
YP-G-180B
1
90
26-May-08
Rigged out HP Flare Scrubber Pump C for QIT repairs
YP-G-180C
0
91
27-May-08
Commenced installation of pump B
YP-G-180B
0
92
28-May-08
Completed installation of pump B
YP-G-180B
1
93
13-Jun-08
Commenced installation of pump C
YP-G-180C
1
94
14-Jun-08
Completed installation of pump C
YP-G-180C
2
95
17-Jun-08
Rigged out HP Flare Scrubber Pump B for QIT repairs
YP-G-180B
1
96
17-Jun-08
Troubleshoot failure to lift on Pump C
YP-G-180C
1
97
18-Jun-08
Rigged out HP Flare Scrubber Pump C for QIT repairs
YP-G-180C
0
98
20-Jun-08
Commenced installation of pump B
YP-G-180B
0
99
21-Jun-08
Completed installation of pump B
YP-G-180B
1
100
22-Jun-08
Reconfirmed Pump B alignment
YP-G-180B
1
101
24-Jun-08
PM on HP Flare Scrubber Pump B
YP-G-180B
1
102
26-Jun-08
Rigged out HP Flare Scrubber Pump B for QIT repairs
YP-G-180B
0
103
30-Jun-08
Commenced installation of pump B on newly designed base
YP-G-180C
0
FMS welding / design inaccurate. Pump pulled out to C-Barrel
104
1-Jul-08
Completed installation of pump C
YP-G-180C
1
105
5-Jul-08
Swapped Barrel B to 'A' Position
YP-G-180A
1
106
11-Jul-08
Investigated unusual noise on Pump C
YP-G-180C
1
107
22-Jul-08
Worked with FMS to assess modification of Pump bases
YP-G-180A/B /C
108
26-Jul-08
Installed pump A
YP-G-180A
2
109
29-Jul-08
Removed pump C
YP-G-180C
1
110
31-Jul-08
Commence removal of Pump A
YP-G-180A
0
Fig.2. Failure and Maintenance Summary for H.P. Flare Scrubber Pumps, YP
Literature Review
Graham and Nurcombe (2003), observed that many vertical submersible pump failures happen rather unexpectedly, without due warning and harsh economic climates and competition has become an incentive for equipment operators to desire to get the best service out of their equipment. This translates to higher life expectancy with plant equipment and as such condition monitoring technologies such as Vibration, Lubricant and Exhaust gas analyses have become very popular to prevent unwanted and unplanned machinery outages. Of the afore-mentioned techniques, Vibration analysis they say, is probably the most important because of its proven efficacy and world-wide acceptance in many industries.
In a case history of submersible pumps at Saudi Aramco, they highlighted the need to know the condition of the pump internals which were submerged in liquid and are usually without condition monitoring. ISO codes specify that bearings be monitored but this is not directly done for submerged pumps because the bearings are normally inaccessible. Instead, ISO allows measurements to be taken from the accessible parts of the machine i.e. from the Electric motor and the downside of this is that plenty of vibration information gets missing or attenuated. A lot of the faults associated with vertical submersible pumps however arise from those inaccessible positions e.g. cavitation, flow induced vibration etc., and as such monitoring the condition of the submerged parts directly provides a valuable source of information for predictive and diagnostic measures that can cause big cost savings for equipment operators.
They have developed and tested transducers and vibration monitoring equipment which can be used to directly get information from the submerged pump parts.
It is not enough just to monitor the vibration condition of these pumps however. The problem at hand is that of the frequent failures of the H.P. Flare Scrubber pumps from the very time they were commissioned. Installing the submersible vibration monitoring devices would definitely help to predict the failures but would do nothing to identify the underlying cause of the frequent failures. So, the failures might be predicted but would happen frequently nonetheless. The root-cause of these failures needs to be identified so that a lasting solution can be developed. Vibration analysis also makes this possible.
Sinha (2008) shows that site installation of machines has effects on their vibration and dynamic characteristics, even when they are well designed. He points to the fact that many newly installed machines vibrate badly and fail frequently just as has been described in the introduction and in the maintenance history of the H.P. Flare Scrubber pumps. Hence it is important to resolve any machine installation problems if equipment availability is desired. A vibration analysis and dynamic characterization technique known as Modal Testing can be used to reveal the natural frequencies of the machine installation assembly and the operating speed checked to see if close to any of the natural frequencies. Operating machines at speeds at or close to structural natural frequencies brings about resonance which is seen as excessive vibration. This test makes it possible to identify the right positions on the structure to apply stiffening in order to reduce vibration by modifying the structural natural frequencies. He gave some examples were these had been done successfully to eliminate frequent machinery failures.
DeMatteo (2001) presents a case study of how the vibration analysis methods of Modal Testing and Operating Deflection Shape have been used to solve the problem of excessive vibration on vertical pumps which are similar to the H.P. Flare Scrubber pumps.
A common yet debilitating fault for these pumps is cavitation. A mentioned earlier, it occurs at the submerged portions of this pump where there is no condition monitoring as of the present. Cavitation is a phenomenon that takes place in these sorts of pumps when the impact of a collapsing vapour of the fluid been pumped causes damage to the impellers and other pump internals. Vapour-bubbles can be formed within the pumped fluid at low pressure pump internals as a result or restricted suction, fluctuating liquid levels of the H.P. Flare Scrubber vessel etc. When these bubbles move on to higher pressure areas within the pump, they collapse and cause damage to the pump. Wilcoxon Research says that, "The collapse of the bubbles is a violent process that creates an impacting action inside the pump. This impact will excite high frequency resonances in the pump structure." For this reason, they advocate the use of vibration sensors in pumps. Cernetic (2009) says in the same vein, that vibration signals should be used to detect cavitation at the early stages of development since this phenomenon causes pump damage and a reduction in efficiency.
From the foregoing, the case for preventing failures is being made and the need for Condition Based Maintenance (CBM) emphasised. Prickett and Eavery (1991) compared preventive maintenance and Breakdown maintenance with CBM. CBM is shown to be cheaper and as such is required for the H.P. Flare Scrubbers if business profitability and equipment availability are desired.
Overview of Proposed Vibration Monitoring and Analysis Based Solution to Frequent Failures
First and foremost, it must be ascertained that the site installation of the three pumps is not causing any vibration problems. The Vibration analysis techniques of Modal Testing and Operating Deflection Shape are proposed for use to determine the root-cause of the frequent failures. Modal testing on the one hand would show structural natural frequencies, node points and mode shapes for the three pumps. The mode shape is the deflection of the structure at any natural frequency. This information would help to determine: if operating speed is dangerously close to the structural natural frequencies, the dynamic characteristic of the structure and the points of least or no deflection on the structure (nodes) - where supports or stiffening may be added in order to alleviate vibration levels. The Impact hammer method shall be used to carry this test out, in situ.
The Operating Deflection Shape (ODS) on the other hand, would as the name implies show the deflection of each pump structure at the operating frequency of 50Hz (3000RPM). In other words, the ODS would show the effect of the operating speed on various parts of the structure and it can be seen if points on the structure vibrate in phase or not. Should parts of a structure not move in phase with the other parts, destructive loading can occur which can give rise to frequent failures.
Carrying out these tests as mentioned above would identify problems with the installation. Solving the installation problems would eliminate the frequent failures. The solution as mentioned earlier, usually involves the application of supports in identified positions or the stiffening of existing supports.
After the installation problems are taken care of, it is proposed that permanently mounted vibration sensors are put in place. For the exposed parts of the pump assembly i.e. the electric motor, it is proposed that two accelerometers be mounted at each of the antifriction bearing housing areas. The accelerometers would be stud mounted at each bearing housing at right angles apart. Having these accelerometers installed in addition to the analysis of the generated signals would make it possible to detect bearing faults at their incipient stages, such that something could be done to prevent a more costly damage to the whole pump assembly.
As for the submerged parts of the pump, i.e. the journal bearings, the shaft, the impeller/bowl assembly, the Bently Nevada designed submersible proximity probes are proposed for use to monitor the vibration and give diagnostic and predictive capabilities for such faults as cavitation, impeller damage, fatigue shaft crack, imbalance etc., which are the common faults of machinery such as these and can only be detected by submerged sensors. Understanding what goes on in the pump hole is vital to keeping the pump healthy. For instance cavitation can be detected on time with these probes, process conditions changed and the reliability of the pumps maintained. The vibration information collected by these sensors would be analysed and used to make quality decisions regarding the required maintenance responses. Analysis techniques would include frequency spectrum analysis, envelope analysis, polar plots, orbit plots etc.
This three prong approach is strongly believed to eliminate the frequent failures, reduce the overall maintenance costs and help in assuring the availability of the three H.P. Flare Scrubber Pumps.
The techniques shall be expounded in more detail within this report and all the necessary tools and vibration signal processing methods shall be specified.
In-Situ Modal Testing for the Pumps
This vibration test is to reveal any problems with the installation of the pumps which might be responsible for the frequent failures experienced within the past few years. In this test, the natural frequencies, mode shape and nodes will be determined for each pump structure.
An instrumented hammer would be used to supply impact or energy to each structure at a known frequency and responses measured. When there is resonance, amplification would be seen in the response spectrum. A Frequency Response Function (FRF) is obtained using the force and the response spectra. The response can be represented as:
FRF= = A + j B
Where A= Real Part, B= Imaginary Part and Phase =
At Resonance, the exciting frequency from impact hammer = Natural Frequency of Pump Structure.
A 0, B gives the Mode Shape, and Phase 90°
The calculations are performed and graphs displayed by the FFT (Fast Fourier Transformation) Analyzer as shown below.
Fig. 3. Time domain and frequency domain signals. (Source: M14 Lecture Notes, 2010, MSc Maintenance Engineering and Asset Management, University of Manchester)
The FRF shows frequency peaks which may or may not be structural natural frequencies. However, for the structural natural frequencies, the relationships shown above would all apply. The real part of the curve (A) would pass through zero and the phase would change by 90°. The imaginary part of the FRF gives the mode shape.
So, the required equipment for the in-situ modal testing of the pump structures are as follows:
Some accelerometers positioned along points on a pump structure (accelerometers can be secured by magnetic means)
An instrumented hammer
An FFT Analyzer
Cable connections for hammer and accelerometers to analyzer
Post Processing Software.
The diagram below shows the layout for the test. The instrumented hammer is used to excite the pump structure and the responses taken from the measurement points and analysed to give all the information required i.e. mode shape, nodes and natural frequencies.
Fig.4. Schematic of Vertical Pump Impact Test
Fig. 5. Impact Test and Modal parameters (source: Richardson M.H. (1997), "Is It a Mode Shape, or an Operating Deflection Shape", Sound and Vibration Magazine, 30th Anniversary Issue.)
Fig.6. Mode shapes from Imaginary Part of FRF (source: Richardson M.H. (1997), "Is It a Mode Shape, or an Operating Deflection Shape", Sound and Vibration Magazine, 30th Anniversary Issue.)
Fig.7. Example of mode shape obtained from pump modal testing. (Source: Sinha J.K., (2008), "Vibration-based Diagnosis Techniques Used in Nuclear Power Plants: An Overview of Experiences", Nuclear Engineering and Design, Elsevier B.V., Volume 238, Issue 9, pp. 2439-2452.
From the resultant mode shapes and observed natural frequencies, insight can be obtained as to the exact positions for stiffening application or mass removal in order to change the natural frequencies. Experience has shown that resonance in these sort of cases is due to the closeness of the operating speed (operating frequency or 1x) or multiples thereof, to one or more structural natural frequencies. For the pump described in fig.7. above, the problem was solved welding a thick plate to the stool in order to stiffen it and by installing additional u-bolts on the discharge line.
Operating Deflection Shape
The Operating Deflection Shape (ODS) simply shows how much the pump structure is moving at a particular frequency (most importantly, the normal operating speed) and how much difference there is in phase between different points of the pump structure as it operates.
The set-up is just as was used for the modal testing. The difference however, is that the instrumented hammer is not used to excite the structure. Instead the machine would be run at its normal operating speed and vibration data collected from the accelerometers is fed to the multi-channel analyzer. The output from the analyzer is then fed into the computer which would have installed specialist software for ODS display. The display would show the actual vibration pattern of the structure. It would be clearly seen, if the pump structure is bending, if parts are moving out of phase with one another etc. It would be seen if any condition exists which contributes to frequent failures.
Fig.8. Example of Software Animation of ODS FRF data. (Source: Richardson M.H. (1997), "Is It a Mode Shape, or an Operating Deflection Shape", Sound and Vibration Magazine, 30th Anniversary Issue.)
It is proposed that both Modal Testing and ODS analysis be carried out by contractors who are specialists in the area of study and have a proven track record of success.
Proposed Permanently Mounted Vibration-based Condition Monitoring System
Fig.9. Set-up of Proposed Permanently mounted vibration monitoring system for the H.P. Flare Scrubber Pumps (Influenced by Fig.5., Graham K.M. and Nurcombe B., (2003), "Vertical Water Pumps- What's Happening Down The Hole", Orbit Magazine, 1Q 2003, pp. 4-9)
Fig. 10. Orthogonally mounted proximity probes [online] Available from: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/obt_tbs_shctln/ [Accessed 6th May 2010]
The proposed permanently mounted vibration monitoring and analysis system for each pump would have the following:
Four Bently Nevada 330400 Accelerometers mounted as shown on the diagram
Four 3300XL 8mm underwater Proximity Probes mounted in Custom Housings (two each for two different journal positions along shaft length)
A Bently Nevada 1900/65 General Condition Equipment Monitor
A tacho-sensor for shaft position reference (from any reputable manufacturer- Bently Nevada, SKF, Endevco)
and,
A laptop with the Bently Nevada System 1 version 6.5 Diagnostic software installed for the analysis of information collected by any of the monitors.
The accelerometers would be used to measure the vibration levels of the anti-friction bearings on the pumps' electric motors while the submerged proximity probes would measure shaft vibration, within the impeller casing and intermediate columns. All the transducer information would be collected by the 1900/65 monitor for each equipment. The monitors are designed for continuous monitoring and read-outs from them can be checked from time to time by dedicated personnel or plant operators.
Also, these monitors have the capability of being tied into existing plant control systems such that vibration warning levels and danger limits for each pump can be announced in control rooms via audible alarms or lights when these levels are breached. Additionally, there is the capability to configure trips and shutdowns in the case of high vibration levels occasioned by faults such as cavitation. The monitors can be installed close to the equipment in sheltered enclosures, using a short cable run. The System 1 software is capable of advanced vibration analysis through the use of displays such as Bode Plots, Spectral displays, Polar plots, Envelope analysis, etc. It is also capable of data acquisition and storage which makes trending possible. Detailed specifications for the various equipment are supplied in the appendix.
Signal Conditioning and Processing
As is well known, the output from the accelerometers and proximity probes are analogue and time domain signals. These have to be converted to digital outputs and frequency domain signatures for fault diagnosis to be carried out. This is achieved by the Fast Fourier Transformation (FFT) algorithm. Studying the frequencies shown in a spectrum is essential for understanding underlying machinery faults, as certain faults have distinct frequency characteristics. For instance, pump rotating speed would be shown in the frequency spectrum and faults on the shaft could be presented as multiples of rotating frequency. The Bently Nevada 1900/65 monitor and System 1 software facilitate this.
Fig. 11. Time domain and frequency domain signals. (Source: M04 Lecture Notes, 2010, MSc Maintenance Engineering and Asset Management, University of Manchester)
The vibration measurement devices have been chosen carefully to minimise the noise and unwanted interference to measurement signals. The Bently Nevada 1900/65 has the capability for low pass filtering and high pass filtering and these can be configured to suit user needs. This helps to eliminate the aliasing effect and other instrument related noise. (See product specification sheet in appendix for details).
There is also the capability for envelope analysis by the use of standard or enhanced demodulation. This is supported by monitor and analysis software and is particularly useful for the early detection of faults on the electric motor anti-friction bearings.
Diagnosis Software Display Plots and Uses
The following show some of the display plots that can be generated by the diagnostic software:
Bode
Performance map
Rotor stator profile
Rotor shape
Hydro air gap
Phasor
Histogram
Octave
Casacade/Full Casacade
Current values
Bar graph
Machine train diagram
Alarm/System event list
Trend / Multivariable trend
Tabular list
Time base (with option for superposition
of baseline data)
Orbit / Time base (with option for
superposition of baseline data)
Orbit (with option for superposition of
baseline data)
Shaft average centerline
Spectrum / Full spectrum (with option
for superposition of baseline data)
X vs. Y
Waterfall / Full waterfall
Polar/Acceptance region
Of the list above, emphases would be placed on the Bode, Polar, Orbit, Shaft average centreline and Waterfall plots. These plots can be used during normal and transient machine conditions to expose the commonly experienced faults.
The Bode plot is very useful in identifying the critical speed (natural frequency) of a machine, as it shows the vibration behaviour of the said machine during start-up or shut-down (transient conditions).
Fig. 12. Bode Plot Example. [Online] Available from: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/bode_polar_plots/ [Accessed 6th May 2010]
The polar plot gives the amplitude of 1X (machine RPM or operating frequency) and its phase difference from the reference position. The amplitude and phase behaviours can be interpreted to actual equipment health or defect.
Fig. 13. Polar Plot example [online] Available from: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/bode_polar_plots/ [Accessed 6th May 2010]
The Orbit plot traces out how the shaft is rotating within the bearing. This tells how much clearance exists between shaft and bearing wall. This information is invaluable as it can be used to determine bearing load changes and the onset of bearing wear.
Fig. 14. Orbit Plot example [online] Available from: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/obt_tbs_shctln/ [Accessed 6th May 2010]
The Shaft centreline plot is used in much the same way as the orbit plot in that it can be used to tell how much wear has happened within a bearing. The plot checks the concentricity or eccentricity of shaft running within a journal bearing, as the name implies.
Fig. 15. Shaft Centerline plot example [online] Available from: http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/obt_tbs_shctln/ [Accessed 6th May 2010]
The Waterfall plot is useful during the transient machine operations. It shows how frequency components such as 1X, 2X, 3X etc change with time or any other variable. The information obtained can be used to make good judgments as to actual machine conditions.
Fig. 16. Waterfall plot example [online] Available from: http://integratedpro.com/content/?p=1114 [Accessed 6th May 2010]
Diagnosis Chart
Fault
Steady state Characteristic
Transient State Characteristic
Shaft rub
0.3 X shown in frequency spectrum,
Funny Orbit plot shapes and discontinuities thereof.
Imbalance
Only 1X is seen in frequency spectrum, 1X increases with time and the Phase angle changes
Bode Plot remains the same, There is no change in critical speed or phase angle when compared with the healthy condition.
Misalignment (or Preload in the case of fluid bearings)
1X,2X,3X,4X etc are shown in frequency spectrum, Phase angle remains constant
The orbit plot will not vary with speed and polar plot remains the same.
Crack
1X, 2X, 3X, 4X etc are shown in frequency spectrum and these continually change in amplitude. The phase angle changes as well.
There is amplitude and phase change of 1X component in the polar plot,
The orbit plot changes from a figure eight shape to a loop containing a small loop.
Bend
Only 1X is seen in frequency spectrum, 1X increases with time and the Phase angle changes
A signal change of phase takes place at critical speed.
Mechanical Looseness
Presence of 0.3X, 0.5X, 1X,1.5X,2x, 2.5X in frequency spectrum
Motor Bearing Damage
Bearing Characteristic Frequencies would be seen in spectrum
Fluid induced instability
The presence of 0.45-0.48 X in spectrum when fluid natural frequency is approached by circumferential speed of fluid, representing Oil Whirl.
The presence of 0.45-0.48 X in spectrum when fluid natural frequency is approached by circumferential speed of fluid, representing Oil Whirl
Oil Whip results when Pump System rotor natural frequency equals fluid's.
Fig. 17. Diagnosis Chart for common faults (Source: Sinha J.K., M14 Lecture notes 2010, MSc Maintenance Engineering and Asset Management, University of Manchester)
Fault Diagnosis Process
The overall vibration levels measurement would be the first stage of protection for the pumps. The ISO recommends the use of RMS values of velocity for overall vibration measurement. Limits for acceptable vibration would be set and configured into the Bently Nevada 1900/65 monitors in terms of Velocity (RMS). These monitors can announced when the limits have been breached and this would prompt further investigation and tests. These limits can be obtained from ISO tables or decided upon in-house by the maintenance engineer. The monitor would show which particular sensor or sensors has detected a fault.
Furthermore, the monitor and diagnosis software proposed are capable of data acquisition and storage which make it possible for trending. The trends would be observed weekly and when a set vibration limit is approached, the frequency of inspection is increased and tests such as the ones mentioned before can be carried out to ascertain the fault type, so a fitting maintenance response can be planned.
Fig. 18. Trending example (source: M04 CBM Lecture notes (2010), MSc Maintenance Engineering and Asset Management, University of Manchester)
The fault diagnosis chart would be used in conjunction with the FMEA diagram, 1900/65 monitor event logs and various applicable display plots (frequency spectrum displays, Bode plot, Polar plot etc) to confirm the exact fault of the pumps. Pump related frequencies would be noted such that when they appear in the frequency spectrum, they can easily be identified.
FMEA, Symptoms of identified deterioration mechanisms
Potential Failure Mode
Potential Effects of Failure
Potential Failure Causes
Symptoms of identified deterioration mechanisms
1
Antifriction Bearing fault (Electric motor)
Bearing Seizure, Misalignment,
Damage to motor shaft, Mechanical seal failure.
Poor lubrication, Resonance
High Frequency Hump seen in Frequency spectrum related to bearing housing natural frequency
2
Shaft Cracks
Shaft Fracture, Loss of pump action
Resonance, Manufacturing defects
1X, 2X, 3X, 4X etc are shown in frequency spectrum. These increase in amplitude over time.
3
Cavitation
Impeller Damage, Reduced output, Pump loss.
Process upsets, Gas lock in Pump Barrel
Noisy operation, High frequency peaks in spectrum
4
Imbalance
Excessive Vibration, B
Damage to bearings and Impellers, Pump loss.
Wear, Impeller damage
1x component in frequency spectrum which increases in amplitude over time.
5
Journal Bearing Wear
Lateral shaft play, Shaft damage
Coupling misalignment
Noisy operation, Lateral shaft play,
6
Impeller Damage
Reduced output, Pump loss, Imbalance,
Improper assembly,
Cavitation, Flow-induced vibration
Blade Passing frequency present in frequency spectrum (5X,10X etc)
7
Bent Shaft
Bearing damage, High vibration
Coupling misalignment, Resonance
Axial vibration, 1X presence in frequency spectrum
8
Coupling Misalignment
Resonance, Damage to pump internals, Mechanical seal failure, Loss of pump.
Improper assembly, Resonance
1X,2X,3X,4X etc are shown in frequency spectrum
9
Shaft rub
Damage to pump internals, Mechanical seal failure, Loss of pump.
Coupling misalignment
0.3X, 0.5X presence in vibration spectrum plot
10
Looseness
Damage to pump internals, Mechanical seal failure, Loss of pump.
Resonance, Improper Assembly
The presence of 0.5X, 1X, 1.5X, 2X, 2.5X etc in frequency spectrum
Fig. 19. FMEA table for pump and motor assembly.
Cost and Man-Power Implications of Vibration Monitoring and Analysis Set-up
The prices given are estimates based on prices obtained from various internet shopping websites. They are not definitive as Bently Nevada gives prices based on different functionality requirements and applications worldwide.
Item
Quantity required for all three pumps
Unit Price ($)
Price ($)
Accelerometer, Bently Nevada 330400
12
500
6000
Monitor, Bently Nevada 1900/65
3
2500
7500
Submersible Proximity probes, Bently Nevada
12
1000
12000
Tacho Sensor, Bently Nevada
3
300
900
Laptop
1
1000
1000
System 1 Diagnostic Software Licence, Bently Nevada
1
20000
20000
Training for vibration engineer (from existing maintenance organisation)
2500
Total Price $ 49,900
Fig.20. Cost breakdown of required equipment for permanently mounted vibration monitoring and analysis system.
At lease one engineer with skills for vibration monitoring and analysis would be required to oversee the whole set-up. He must be trained and competent in the use of various display plots, signal processing and conditioning, for fault identification and detection.
This knowledge can also be used on other critical plant equipment such as the gas turbines and the centrifugal gas compressor.
The dynamic characterization tests are to be contracted out to experienced service providers with the adequate hardware and software for real-time animation of vibrational motion. It is estimated that the cost of this service would be circa $100,000. This brings the grand total of the proposed vibration programme from the frequent pump failures solution to permanently installed condition monitoring to about $150,000.
Benefits and Limits of the proposed Vibration-based Condition Monitoring System
The proposed set-up for monitoring and analysing the vibration from the pumps has quite a number of benefits.
From the business point of view, it is an investment because it can prevent costly failures. The dollar value of the pumps' failure within the two-year period considered in this report easily exceeds $1million when repair costs, spare parts, logistics and man-hours expended are considered.
For the maintenance organisation, the presence of these vibration-based condition monitoring equipment, makes it possible for maintenance to be pro-active rather than reactive. Furthermore, frequent failures are eliminated which give room for better planning and more time for effective and efficient maintenance.
The cost of the vibration monitoring and analysis equipment can be seen to be a small price to pay for asset availability, enhanced productivity and even safety.
The limitation to the proposed system is the skill, knowledge and competence of the engineer or engineers who are in charge of the set-up. The signals for any fault condition would always be picked up by the monitoring equipment. The proper and accurate diagnosis of faults and subsequent maintenance decisions made are the remit of the engineer(s) responsible for the vibration-based condition monitoring programme. In addition, vibration monitoring equipment could fail and require replacement.
Conclusion
A glance through the summarised failure and maintenance history of the H.P. flare scrubber pumps for a two year period reveals the amount of resources expended on them and their poor availability. Clearly then, something new and different from the previous approaches should be attempted.
This proposed system covers all the grounds- from installation problems check, process vagaries that cause cavitation, to common faults experienced by rotating machines such as bearing defects and coupling misalignments to mention a few.
Besides, the proposed methods are tested and trusted and can contribute to savings in maintenance cost, plant availability and safety which are key performance indicators for most industrial plants.
Appendix i- Bently Nevada 1900/65 Monitor
[Online] Available from: www.ge-energy.com/prod_serv/products/oc/en/bently_nevada.htm [Accessed 6th May 2010]
Appendix ii- Bently Nevada 330400 Accelerometers
[Online] Available from: www.ge-energy.com/prod_serv/products/oc/en/bently_nevada.htm [Accessed 6th May 2010]
Appendix iii- Bently Nevada System 1 Diagnostic Software
[Online] Available from: www.ge-energy.com/prod_serv/products/oc/en/bently_nevada.htm [Accessed 6th May 2010]
Appendix iv- Vibration Severity Limits for Machines
Fig. 21. ISO 10816 Vibration Severity Limit Chart
Appendix v- Submersible Proximity Probes
[Online] Available from: www.ge-energy.com/prod_serv/products/oc/en/bently_nevada.htm [Accessed 6th May 2010]
