GB2314412A - Method of monitoring pump performance - Google Patents

Abstract

Levels of acoustic noise (the mean level of an acoustic signal over a predetermined time) and acoustic distress (the peak level of an acoustic signal over a predetermined time) associated with a pumping installation are measured at a predetermined time. The levels of acoustic noise and acoustic distress are then monitored with time and compared with these predetermined levels to obtain a measure of pump performance. Additional parameters such as pump power consumption, pump speed and pump head are also measured and compared with predetermined levels, and can be combined with the measurements of acoustic noise and distress in a manner that is dependent upon the relative importance of each parameter for the particular pumping installation under investigation. The results obtained are also adjusted according to the medium being pumped.

Description

METHOD OF MONITORING PUMP PERFORMANCE The present invention is concerned with a method of monitoring pump performance and may be used, for example, to assist in scheduling maintenance for a pumping installation.

Pumping efficiency (or pump performance) is generally based on the relationship between pump head and capacity (flow rate) for a particular design of pump and the power required by the pump to produce the necessary throughput for a given location. The relationship between head, throughput and power is initially calculated at the design stage of the pump and is subsequently checked using empirical data from tests carried out by the manufacturer to produce a set of head, capacity and power curves for that design of pump. A pump is therefore selected on its ability to deliver the required capacity at a given head at the optimum power rating for the required duty. These form the key elements of pump selection for a given installation.

Pumping installations are designed to operate at an optimum efficiency for their prescribed duty to ensure that the required pumping capacities are realised and running costs (in respect of energy and maintenance) are kept within budget. To this end, new pumping installations are type tested using an AEMS (or Yates) meter which provides a BSI standard for absolute measurement of pumping efficiency for the new installation. The AEMS meter determines the work done by a pump by measuring the change in temperature across the inlet and outlet stages of the pump and combines this measurement with other measured parameters such as power, pressure and flow. For high capacity, high power pumping installations continuous monitoring of efficiency can be justified and an AEMS meter is permanently installed onto each pump. However, on many lower capacity installations, the high cost of the AEMS meter precludes it being an integral part of the installation and, unless the installation is seen to be showing progressively increasing running costs (energy, maintenance, etc.) when compared with the original costs, no further testing of performance is undertaken during the working life of the pump. As an alternative to the AEMS meter it is possible to obtain a less accurate measurement of pump performance by examining changes in power demand, outlet pressure and flow rates from the pump.

The drawbacks of these two known techniques are that they rely on measuring secondary effects of a degradation in pump performance which result from losses induced by wear and tear of the moving components within the pump and also ageing of pipework due to corrosion and deposition of, for example, minerals. The changes in temperature and other parameters due to pump wear and tear can be very small indeed and are therefore subject to experimental error whenever measurements are taken. In the case of the AEMS meter, this requires a lengthy testing regime to be able to ensure that decreases of 1 percent in pumping efficiency can be accurately resolved. The technique of examining changes in power demand provides a relatively quick but less sensitive and less accurate means of estimating pump performance, but is considerably less expensive to employ.

In general, errors in estimation of pump performance increase with smaller capacity pumps.

The condition of smaller capacity pumping installations is generally determined by periodic inspection of the equipment, the interval between inspections being set to prevent as far as possible any unplanned failures. If it were possible effectively to monitor the condition of such pumping installations by remote means it would, in most cases, be possible to increase the interval between inspections, thereby effecting a cost saving.

There is therefore a need to provide an accurate and inexpensive means for estimating pump performance, especially in respect of smaller capacity pumps.

According to the present invention there is provided a method of monitoring pump performance, which method comprises the steps of: determining levels of acoustic noise and acoustic distress (as hereinafter defined) associated with a pumping installation at a predetermined time; monitoring levels of acoustic noise and acoustic distress associated with the pumping installation with time; and comparing the monitored levels of acoustic noise and acoustic distress relative to the predetermined levels and comparing the results to obtain a measure of pump performance.

The term "acoustic noise" as used herein is a measurement of the mean level of an acoustic signal over a predetermined time.

The term "acoustic distress" as used herein is a measurement of the peak level of an acoustic signal over a predetermined time, which predetermined time may be the same as, or different from, the predetermined time over which acoustic noise is measured.

The method may include the additional step of determining power consumption associated with the pumping installation at the predetermined time, monitoring the power consumption with time, and comparing the monitored power consumption relative to the predetermined power consumption and combining the result with the first-mentioned results to obtain the measure of pump performance.

The method may include the additional step of determining the pump head associated with the pumping installation at the predetermined time, monitoring the pump head with time, and comparing the monitored pump head relative to the predetermined pump head and combining the result with the first-mentioned results to obtain the measure of pump performance.

The method may include the additional step of determining the pump speed associated with the pumping installation at the predetermined time, monitoring the pump speed with time, and comparing the monitored pump speed relative to the predetermined pump speed and combining the result with the first-mentioned results to obtain the measure of pump performance.

The predetermined and monitored values may be subject to an offset before they are compared. This is particularly the case where the values can be expected to have operating minima such as power consumption and head.

The compared levels in each category may be combined in a manner that is dependent upon the relative importance of that category for the particular pumping installation being monitored.

The method may include the further step of adjusting the combined results according to the medium being pumped in order to obtain the measure of pump performance.

For a better understanding of the present invention and to show more clearly how it may be carried into effect reference will now be made, by way of example, to the accompanying drawings in which: Figure 1 is a diagrammatic illustration of one embodiment of a system for use in monitoring pump performance; Figure 2 is a diagrammatic illustration showing in more detail one way in which pump performance data can be utilised; and Figure 3 is a diagrammatic illustration showing another way in which pump performance data can be utilised.

Figure 1 shows two pumps 1 and 3 linked to a monitoring system 5 by way of acoustic emission transducers 7, 9 for providing information on pump dB levels and pump distress levels and analogue transducers 11, 13 for example for providing information on power consumption, head and speed.

The actual number and type of transducers used will depend on the type of pump being monitored as will be explained in more detail hereinafter.

The acoustic emission transducer may be, for example, of the type SWT/2 which is a 90kHz acoustic transducer manufactured and sold by Holroyd Instruments Limited, Matlock, Derbyshire, United Kingdom. These transducers produce a noise signal corresponding to a range from 0 up to about 100 decibels (dB) and a distress signal in a range from 0 to 50 (the distress signal is in non-dimensional units).

The monitoring system 5 includes a display 15, 17, such as a liquid crystal display, for each pump together with an alarm indicator 19, 21 for each pump. The monitoring system includes means (not shown) for storing data which can be evaluated remotely as indicated by arrow 23.

Figure 2 shows how the data provided by a number of monitoring systems can be transmitted by modem 25 over the public telephone network 27 to a computer 29 or the like.

The computer 29 can be used to archive and analyse the data and to identify trends in pump performance. Reports can be produced and printed or otherwise distributed as indicated by arrow 31. In the event of an alarm signal being generated, the computer 29 can issue a signal to a paging service 33 which in turn communicates with alpha-numeric pagers 35 carried by service personnel and which can be used to summon assistance to a pumping installation which is experiencing a problem.

Where the monitoring systems 5 are close to each other, they can be networked together in a manner not shown so as to require only a single modem.

Where it is not practical to transmit data from the monitoring system 5 to the computer 29, the data can be accumulated in a removable data storage device 37 provided in the monitoring system as shown in Figure 3. The removable data storage device may be, for example, a PCMCIA SRAM card for use with a portable computer. The storage device 37 can be removed from the monitoring system and inserted into a portable computer 39 as indicated in Figure 3 and can then be transmitted to the computer 29 for example by cellular telephone 41 or by floppy disk 43 or a direct link 45 between the two computers 39 and 29. As with Figure 2, the computer 29 can be used to archive and analyse the data and to identify trends in pump performance. Reports can also be produced and printed or otherwise distributed as indicated by arrow 31.

I have found that pump performance can be reliably estimated by means of one or more acoustic emission transducers disposed at one or more suitable locations relative to the pump. Indeed, when combined with readily measurable parameters such as power consumption and head, acoustic emission can be used to monitor aspects of a pumping installation beyond simple pump performance. For example, it is possible to monitor the general mechanical condition of a pump, such as loss of lubrication and/or bearing damage, increased internal clearances and failure of seals. It is also possible to monitor pump behaviour under load in dependence on the fluid dynamics, for example changes in fluid density, particle sizes, solids content, cavitation, speed and the like. It is also possible to monitor pump duty with time, such as transient load conditions at start up and shut down. Beyond the pump itself, it is possible to monitor external influences and pipework condition, such as pipe bursts, blockages and the like.

I have found that it is possible to derive an index of pump performance on the basis of energy losses incurred during operation of the pump. Essentially the energy losses are: power consumption by the pump, differential pressure across the pump (e.g. pump head), an acoustic determination of pump distress, and an acoustic determination of pump noise.

Pump head (pressure) is a primary indicator of pumping medium and flow characteristics. With regard to pumping medium characteristics, pump head for a given flow rate is proportional to the specific gravity of the medium being pumped. With regard to flow characteristics, variations in pump head are representative of changes in flow due, for example, to leakage and/or blockage in the supply and/or outlet main.

Acoustic distress of a pump is a primary measure of worn or damaged components within the pump, but can also be used to identify processes linked to the operation of external mechanisms such as valves. Acoustic noise of a pump is a secondary measure of damage and is used to quantify the level of permanent wear and damage to the pump; it is also a secondary indicator of pumping medium and flow characteristics by detecting variations in the turbulence of the medium being pumped and/or variations in loading of the pump bearings.

The level of these energy losses is dependent on the pumping medium and the type of pump. Specific values can be derived by straightforward experimentation which in itself requires no inventive activity.

The pumping medium can be classified according to specific gravity and homogeneity. In water pumping installations (i.e. water and sewage pumping installations) I have identified three different categories of pumping medium: water (both treated and untreated), sludge and humus (i.e.

homogeneous material having a specific gravity greater than 1) and raw sewage (i.e. non-homogeneous material having a rapidly varying specific gravity greater than or equal to 1). Other pumping installations could well display other classes of pumping medium.

I have found that pumps can be divided into a number of main types, reciprocating (or displacement) pumps, rotary pumps and centrifugal pumps and these main types can be sub-divided. That is reciprocating (or displacement) pumps can be either piston pumps or diaphragm pumps; rotary pumps can be vane pumps, helical (progressive cavity (PCP)) pumps, Archimedean screw pumps, peristaltic pumps or gear/lobe/screw pumps; and centrifugal pumps can be single or multistage axial pumps, submersible pumps or torque flow pumps. Inevitably, other types of pumps will be identified in due course.

I have found that the energy losses have values associated therewith, which values are characteristic of the type of pump and pumping medium. The characteristic values are derived by developing a model of the pump type from statistical analysis of empirical measurements taken from different pump types, which model is subsequently tested and tuned as necessary. The characteristic values can be considered as "loss contribution constants" (LCC).

The loss contribution constants are used to adjust the measured energy losses according to pump type and pumping medium. For example, head losses would not be included in the calculations for a helical rotor pump and the corresponding LCC for a helical rotor pump is therefore assigned a value of zero.

I have also found that the expected ranges of power and pressure may need to be taken into consideration for a given installation and pump type. This is accomplished by employing offsets for power and pressure figures, which offsets can be considered as "sensitivity calibration values" (SCV). An SCV is generally subtracted from a corresponding measured value to adjust the zero point or "floor value" for pump head or power, thus effectively increasing the significance of changes in the measured value. An SCV is determined by a number of factors relating to the operation of a pumping installation and the medium being pumped. For example, a significant deterioration in the operating performance of a pump may be accompanied by a net change in head of only 0.15 bar. If the head at commissioning of the installation is 1.5 bar then the overall change to 1.35 bar is only 10 percent of the initial head. However, where this is required to represent a change in pump performance of 100 percent then the SCV would be set at 1.35 bar.

Where the speed of a pump is variable, the effects of pump speed may also need to be taken into account.

The energy losses, as adjusted for pump type through the loss contribution constants and the sensitivity calibration values, in combination with pump speed and pumping medium can be used to derive a pump performance index (PPI), for example having a value in the range from zero to ten. PPI is dependent upon a knowledge of the baseline characteristics for a particular pumping site at the time of commissioning and these baseline values are used to normalise the current measured values: thus a PPI of 10 represents the performance of a pump when newly commissioned. After commissioning, high PPI values indicate efficient performance and low PPI values indicate poor performance.

The pump performance index for any particular pumping installation after commissioning can be calculated as follows: PPI = {[A1..nαAl..An]SNX1 + [B1..nαB1..Bn]SNX2 + HαHSNX3 + PαPSNX4} X MP X 10 where n = number of acoustic emission sensors A1..n = (baseline distress level) . (measured distress level) B1..n = (baseline noise level for pump) . (measured noise level) H = (measured outlet pressure - B) ; (baseline pressure - H) P = (measured power - p) . (baseline power 5N = (baseline speed value) # (measured speed value) Mp = pumping medium adjustment constant A1..n,B1..n,H,P # 1 αA1..An,αB1..Bn,αH,αP = 1 and H = SCV for pressure P = SCV for power αA1..An = LCC for distress measurement αB1..Bn = LCC for noise measurement aH = LCC for head measurement op = LCC for power measurement The values of X1, X2, X3 and X4 may vary depending for example on the level of sophistication required for the analysis. For a straightforward analysis, for example for an Archimedean screw pump, X1, X2, X3 and X4 may all be unity giving rise to the formula: PPI = {A1..nαA1..An + B1 .naB1.. + HOH + pap} x SN X Mp x 10 For a more sophisticated analysis, particularly for a centrifugal pump, X1 and X2 may be unity, but X3 may be 2 and X4 may be 3 giving rise to the formula: PPI = {[A1..nαA1..An + B1..nαB1..Bn]SN + HαHSN2 + PαPSN3} X Mp x 10 Other types of pump may require different values for X1, X2, X3 and X4.

EXAMPLE 1 I have applied the above-described evaluation of pump performance index to a single axial stage variable speed centrifugal pump.

In a preliminary step it was found that a single acoustic emission transducer is necessary for determining distress and dB levels at the impeller bearing of such pumps and the baseline loss contribution constants (LCC) for distress, noise level, head and power were determined. Because the pump can operate at different speeds, the power and head ranges were determined and the sensitivity calibration values (SCV) for power and head set at the lower end of each range. Additionally, the optimum pumping speed was determined.

I have found that the primary indications of deterioration for pumps of this type are increasing bearing noise level and falling head. Thus the loss contribution constants relating to these two sources of energy losses are more heavily weighted than the other two sources.

The pump was used to pump water which I have determined as having a pumping medium adjustment constant of 1.

The predetermined factors and measured values for the pump are shown in Table 1 below:

Energy loss Sensitivity Loss Optimum/ Range Measured parameter calibration contribution Baseline value value constant value Power p = 160 kW αP = 0.15 200 kW 230 - 160 180 kW kW Head H = 1.0 = = 0.35 1.4 bar 1.4 - 1.0 1.215 bar bar bar Distress at n/a B1 = 0.20 i10.0 n/a 14.3 impeller bearing Noise level n/a = = 0.30 i30.0 n/a 42.1 dB at impeller bearing Table 1 The pump was operating at 900 rpm, its optimum value, although it could operate in the range from 730 - 950 rpm.

The pump performance index can be calculated using the values in Table 1 and those given above and using the equation given above (with X1 = X2 = 1, X3 = 2 and X4 = 3) to give a figure of 5.76.

The condition of the pump is such that the impeller bearing is deteriorating as indicated by the rising distress and noise levels. The high distress level indicates that damage is taking place within the bearing and the high noise level indicates the permanence of that damage. In addition the head (outlet) pressure has fallen as a result of increased clearances between the impeller and its casing: there has also been a small reduction in power required. In this case, it is likely that the increased clearances between impeller and casing and the bearing damage are closely related to each other.

An alarm threshold may be set when the pump performance index falls to a predetermined level, such as 5.0. This level does not mean that the pump is about to fail, but implies that there has been significant loss in performance and that maintenance should be scheduled based on recommendations produced by further off-line analysis of the data.

However, should a leak or a blockage develop the substantial drop in pump head that results will more than likely cause the pump performance index to go negative thus making it easy not only to detect this condition, but also to correct it promptly.

EXAMPLE 2 I have also applied the above-described evaluation of pump performance index to an Archimedean screw pump rotating at a steady 45 rpm.

As a preliminary step it was found that two acoustic emission transducers are necessary for determining distress and noise levels in this type of pump, one being positioned at the top bearing and the other being positioned at the lower bearing. In the pump tested the upper bearing was a combined thrust and journal roller bearing, while the lower bearing was a plain journal bearing. Because of the nature of the pump, there is effectively no head and this term is not used in the calculation. Thus baseline loss contribution constants (LCC) were determined for distress and dB level both for the upper and lower bearings and also for power. Because power will vary with clearances between the screw and its channel a sensitivity calibration value (SCV) was determined.

I have found that the primary indications of deterioration for pumps of this type are increasing bearing distress levels for the upper and lower bearings and falling power level. In addition it is necessary to take into consideration the different nature of the upper and lower bearings. Thus the loss contribution constants relating to these two sources of energy losses (distress and power) are more heavily weighted than the other two sources.

The speed of the pump was constant and therefore has no effect on the pump performance index and the pumping medium was raw sewage, but again with this type of pump the pumping medium has no effect on the pump performance index.

The predetermined factors and measured values for the pump are shown in Table 2 below:

Energy Sensitivity Loss Optimum/ Range Measured loss calibration contribution Baseline value parameter value constant value Power p = 200 kW ap = 0.2 250 kW n/a 220 kW Head n/a n/a n/a n/a n/a Distress n/a aM = 0.25 #10.0 n/a 15.3 at upper bearing dB level n/a apl = 0.15 i10.0 n/a 9.0 dB at upper bearing Distress n/a aS = 0.2 i10.0 n/a 9.3 at lower bearing dB level n/a aBX = 0.2 i15.0 n/a 14.1 dB at lower bear in Table 2 The pump performance index can be calculated using the values in Table 2 and the equation given above ( with X1 = X2 = X3 = X4 = 1) to give a figure of 7.93. This is an acceptable PPI value for this pump although the upper bearing is running in a damaging condition and some deterioration has taken place. A reduced power demand is indicating losses in pumping capacity due to increased clearance between the screw and its channel. For slowly rotating shafts I have found that the distress levels are more important indications of bearing damage than noise levels.

In this case, the increased clearance between impeller and casing and top bearing damage have no relation to each other.

Thus the value of the pump performance index enables a maintenance engineer to be able to assess and prioritise maintenance for each pumping installation. Over and above this, it is possible for the apparatus to issue an alarm should a predetermined threshold be exceeded for example for pump performance index, distress or noise level.

Claims (12)

1. A method of monitoring pump performance, which method comprises the steps of: determining levels of acoustic noise and acoustic distress (as hereinbefore defined) associated with a pumping installation at a predetermined time; monitoring levels of acoustic noise and acoustic distress associated with the pumping installation with time; and comparing the monitored levels of acoustic noise and acoustic distress relative to the predetermined levels and comparing the results to obtain a measure of pump performance.

2. A method according to claim 1 and including the additional step of determining power consumption associated with the pumping installation at the predetermined time, monitoring the power consumption with time, and comparing the monitored power consumption relative to the predetermined power consumption and combining the result with the first-mentioned results to obtain the measure of pump performance.

3. A method according to claim 1 or 2 and including the additional step of determining the pump head associated with the pumping installation at the predetermined time, monitoring the pump head with time, and comparing the monitored pump head relative to the predetermined pump head and combining the result with the first-mentioned results to obtain the measure of pump performance.

4. A method according to claim 1, 2 or 3 and including the additional step of determining the pump speed associated with the pumping installation at the predetermined time, monitoring the pump speed with time, and comparing the monitored pump speed relative to the predetermined pump speed and combining the result with the first-mentioned results to obtain the measure of pump performance.

5. A method according to any preceding claim, wherein one or more of the predetermined and monitored values are subject to an offset before they are compared.

6. A method according to claim 5, wherein values expected to have operating minima, such as power consumption and head, are subject to an offset before they are compared.

7. A method according to any preceding claim, wherein the compared levels in each category are combined in a manner that is dependent upon the relative importance of that category for the particular pumping installation being monitored.

8. A method according to any preceding claim and including the further step of adjusting the combined results according to the medium being pumped in order to obtain the measure of pump performance.

9. A method according to any preceding claim and including the step of determining a pump performance index for any particular pumping installation after commissioning as follows: PPI = {[A1..nα1..An]SNX1 + [B1..nαB1..Bn]SNX2 + PαPSNX4} x Mp x 10 where n = number of acoustic emission sensors A1. = (baseline distress level) . (measured distress level) B1..n = (baseline noise level for pump) + (measured noise level) H = (measured outlet pressure - H) t (baseline pressure - ssE) P = (measured power - ssp) + (baseline power 5N = (baseline speed value) # (measured speed value) Mp = pumping medium adjustment constant # A1..n,B1..n,H,P # 1 # αA1..An,αss1..Bn,αH,αP = 1 and - = SCV for pressure ssP = SCV for power An = LCC for distress measurement αB1..Bn = LCC for noise measurement OH = LCC for head measurement αP = LCC for power measurement.

10. A method according to claim 9, wherein the step of determining a pump performance index for any particular pumping installation after commissioning as follows: PPI = {A1..nαA1..An + B1..nαB1..Bn + HαH + PAP} X SN X MP X 10.

11. A method according to claim 9, wherein the step of determining a pump performance index for any particular pumping installation after commissioning as follows: PPI = {[A1..nαA1..An + B1..nαB1..Bn]SN + HαHSN2 + PαPSN3} X Mp X 10.

12. A method of monitoring pump performance substantially as hereinbefore described with reference to the accompanying drawings.