Long-term Evaluation of Ultrasonic Flowmeters

Transcription

1 Long-term Evaluation of Ultrasonic Flowmeters A Report for National Measurement System Policy Unit DTI, London Project No: OR7 (OSDC57) Report No: 351/99 Date: 20 September 1999

2 Flow Centre Long-term Evaluation of Ultrasonic Flowmeters: A Report for National Measurement System Policy Unit DTI, London S U M M A R Y This report describes the results of the project Long-term Evaluation of Ultrasonic Meters. A series of simulated maintenance tests have been carried out on commercial transit time ultrasonic meters supplied by three manufacturers. The meters supplied were all 150 mm nominal diameter. The tests have investigated the potential effects of replacing component parts such as transducers, electronics and inter-connecting cables. The results show that for the majority of tests the effects of exchanging these parts are of the same order of magnitude as the calculated repeatability of the meters, a few tenths of a percent. In some specific cases effects of approximately 1% magnitude have been observed. In each case these have occurred when transducers have been exchanged. Periodic tests show that for all but one meter, a clamp-on design, the ideal long-term reproducibility is comparable with baseline repeatability and short-term reproducibility. Prepared by: Mr G J Brown Approved by: Dr A R W Hall Date: 20 September 1999 for Dr F C Kinghorn Director Report No: 351/99 Page 1 of 81

3 C O N T E N T S 1 INTRODUCTION BACKGROUND SCOPE EXPERIMENTAL PROGRAMME 4.1 The Project Meters Test Facilities and Method Calibration Schedule ANALYSIS AND PRESENTATION RESULTS 6.1 Danfoss Sonoflo 3000/ Panametrics XTM Krohne Altosonic UFM SUMMARY AND CONCLUSIONS Page APPENDIX 1: SET-UP DATA - DANFOSS SONOFLO APPENDIX 2: SET-UP DATA - PANAMETRICS XMT APPENDIX 3: SET-UP DATA - KROHNE ALTOSONIC UMF Report No: 351/99 Page 2 of 81

4 1 INTRODUCTION This report describes the results of research project OR7 Long-term Evaluation of Ultrasonic Flowmeters of the National Measurement System Three Year Programme for Flow Four complimentary reports are available which describe the results of two further projects conducted in the programme. Velocity Distribution Effects on Ultrasonic Flowmeters Part 1 Theoretical Analysis, NEL Report No 357/99 Velocity Distribution Effects on Ultrasonic Flowmeters Part 2 - Determination by Computational and Experimental Methods, NEL Report No 348/99 Multiphase Flow Effects on Ultrasonic Flowmeters, NEL Report No 350/99 Research into Clamp-on Ultrasonic Meters, NEL Report No 359/99 2 BACKGROUND The potential of ultrasonic flowmeters for non-intrusive, accurate, and repeatable measurement over a wide range of conditions was assessed in the previous National Measurement System Flow Programme. However, the long-term performance, reliability and maintainability of modern meters is largely undocumented. In the past, ultrasonic meters have been reputed to have poor reliability and long-term stability in comparison with mechanical meters, largely due to shortcomings in electronic design. Electronics related problems are thought to have been effectively eliminated in modern meters by the adoption of digital technology. There is no information that suggests a suitable interval for recalibration, the implication being that, intrinsically the meters are not subject to drift. Similarly little information is available on the effect that the exchange component parts will have on calibration. Data quantifying the capabilities and any limitations of ultrasonic meters in these respects would be of great value to industries where low maintenance and calibration requirements offer substantial savings, e.g. offshore oil and gas. 3 SCOPE The specification of the project was prepared in consultation with relevant parties during the formulation process for the Flow Programme. The approach adopted was based on a programme of laboratory calibration of proprietary meters following simulated maintenance actions and at intervals over an extended period of time. The calibration schedule was specified such that the results comprise sufficient data to establish linearity, repeatability and accuracy across a useful range. Commercial meters were selected for inclusion on the basis of meter performance in the supplementary project Ultrasonic meters for oil flow measurement of the Flow Programme. 4 EXPERIMENTAL PROGRAMME 4.1 The Project Meters Meters were obtained from three manufacturers for the programme. A brief description of the instrumentation supplied follows DANFOSS Sonoflo Converter type: Converter Serial No: Sensor type: Sensor Serial No: SONO F N355 SONO (DN150, CLASS150) 085B N187 The SONO3100 is a sensor tube with four transducers forming two parallel tilted chordal sound paths. Each path is formed at a horizontal mid-radius position in the cross-section and is inclined at an angle of 39 to the tube axis. The transducers are metal encapsulated piezoelectric crystals that are mounted in a welded boss in the pipe spool by means of a screw fitting. A memory chip that is pre-programmed with the primary physical parameters and factory calibration factor is provided with the meter tube. The calibration report issued by Danfoss lists the primary physical parameters and factory calibration factor as reproduced below. Report No: 351/99 Page 3 of 81

5 Di = [m] L1 = L2 = [m] h1 = h2 = 352 [m] θ1 = θ2 = [ ] Cal. Factor = {Meter tube internal diameter} {Distance between the windows of each transducer pair} {Radial displacement of the chordal paths} {Angle between the pipe centreline and sound paths} The SONO3000 signal converter controls the excitation and detection of the ultrasonic signals and subsequent processing and flow computation. A sine wave of 8 cycles in duration and frequency of around 1 MHz is utilised to excite the transducers. The received upstream and downstream signals are digitised and stored for processing. Automatic gain control (AGC) is utilised to optimise the analogue to digital conversion. The flow computation is based on measurement of the overall transit-time and the upstream-downstream transit-time difference, which are both treated separately by correlation techniques. The transit-time difference is determined by cross-correlating the upstream and downstream signals whereas the overall transit-time is determined by cross-correlating one of the received signals with a stored reference signal. Set-up parameters for this meter are listed in Appendix Panametrics XMT868 Boss-mounted transducer configuration Model No.: XMT Serial No.: 224 Transducer No.: 42 (Standard) and 74 (Standard) Externally mounted configuration Model No: XMT Serial No: 223 Transducer No: 30 (Standard) The Panametrics XMT868 is a dual channel flow transmitter for use with clamp-on or wetted transducers. Panametrics provided two XMTs with both boss-mounted and clamp-on transducers. Two pairs of boss-mounted transducers were supplied with a meter spool. One path comprised a pair of 1 MHz 1-inch NPT extended-well type transducers which are threaded directly into a welded bosses which are drilled through into the pipe. The other path comprised a pair of 1 MHz 1-inch transducers which are threaded into Panadapta plugs which are in turn threaded into 1-inch NPT bosses allowing the transducers to be removed under operating conditions. Each path is formed at a horizontal mid-radius position in the cross-section and is inclined at an angle of 45 to the tube axis. The clamp-on configuration employed two pairs of 1 MHz shear-wave transducers which are mounted on the exterior of a length of 150 mm nominal bore Schedule 40 carbon steel pipe by means of a yoke-and-strap arrangement and coupled to the pipe wall using a coupling grease. The transducers were mounted such that single-reflection paths were in perpendicular planes at 45º and 135º to the horizontal. The Panametrics XMT868 uses a coded excitation and correlation detection scheme to determine the transit time of the ultrasonic signals. The excitation signal, rather than an impulse or sinusoid, is a phase-coded squarewave burst. AGC is utilised to optimise the signal level before digitisation and averaging of several successive signals. The digitised receive signals are then cross correlated with a stored version of the excitation signal to determine the upstream and downstream transit times. Neither of the configurations were flow calibrated prior to delivery. Set-up parameters for these meters are listed in Appendix 2. Report No: 351/99 Page 4 of 81

6 4.1.3 Krohne Altosonic UFM500 Primary head Type: UFS500F DN: 6 inch Flanges ASA 150 lbs. RF Test pressure: 30 Bar Measuring tube: Steel Sensor frequency: 1 MHz Protection class: IP65 Converter Type: UFC500F PC value: The UFS500S is a sensor tube with four transducers forming two parallel tilted chordal sound paths. Each path is formed at a horizontal mid-radius position in the cross-section and is inclined at an angle to the tube axis. The transducers are 1 MHz piezoelectric crystals that are mounted in a welded mount in the sensor tube, isolating them from the fluid. The UFC500F signal converter controls the excitation and detection of the ultrasonic signals and subsequent processing and flow computation. Upstream and downstream transit-time measurements are performed by a microprocessor-controlled threshold-armed zero-crossing detection technique. Set-up parameters for this meter are listed in Appendix Test Facilities and Method The test programme of this project was conducted in the oil flow measurement National Standard facility at NEL. The main oil flow facility comprises three separate flow circuits each with a 100 l/s flow rate capability using kerosene, gas oil and lubricant oils. Figure 4.1 shows a schematic diagram of one of these circuits. The oil for each circuit is stored in 30 m3 tanks and is maintained to within 1 C of a pre-determined temperature within the range of 5 to 45 C by a conditioning circuit linked to each tank. The test lines themselves can each accommodate 30 m of horizontal straight lengths or other configurations as required. At the outlet of each test section a manifold directs the fluid either back to the storage tank or to one of the weigh tanks. Line temperature and pressure are monitored both upstream and downstream of the test section. Key: Flow direction Valve 1500 kg Tank 30m Test Section 6000 kg Tank Bypass Pump Supply Tank Figure 4.1 A Schematic Diagram of the Test Facility Report No: 351/99 Page 5 of 81

7 The facility is fully traceable to Primary National Standards and is accredited by the National Accreditation of Measurement and Sampling (NAMAS). The results reported here were obtained by a master-meter calibration method using a sliding-vane positive displacement (PD) meter. Pulses are collected simultaneously from the test meter and reference PD meter for a given test interval. The uncertainty in the measurement of the quantity of fluid passed through the meter is ±8% for the master meter method, this uncertainty being estimated at a confidence level of 95%. The calibration of the master meter was checked during the programme of tests to ensure that measured effects were due to the meters under test and not the calibration system. The calibration factors used for the master meter were not altered at any time. Figure 4.2 shows the baseline and final calibrations of the master meter. Figure 4.3 shows the deviations from the polynomial curve fitted to the first calibration. It can be seen from this figure that all deviations are less than ±6%. 180 K-factor (P/l) Poly. ( ) Poly. ( ) 150 y = E-07x E-05x E-04x E Flow rate (l/s) Figure 4.2 Baseline and final calibrations of the PD master meter Deviations (%) Flow rate (l/s) Figure 4.3 Percentage deviations from the baseline calibration Report No: 351/99 Page 6 of 81

8 The ultrasonic meters tested under this programme were configured to provide frequency output scaled to an appropriate range (eg khz corresponding to l/s). When recording a frequency output the volumetric flowrate is determined from the frequency obtained by dividing the totalised pulse count by the counting interval. 4.3 Calibration Schedule A nominal bore size of 150 mm was chosen for the project to obtain a reasonable range of velocities within the volumetric flowrate capacity of the facility. The fluid used in these tests is a processed hydrocarbon oil of viscosity ~24 cst and density ~0.85 kg/l at 20ºC, the nominal test temperature. The points chosen for the calibration schedule are given in Table 1 below. TABLE 1 The Calibration Schedule Adopted for the Test Programme % of Max. Flowrate Velocity Reynolds No Typical No of (l/s) (m/s) repeats , , , , , , ,800 5 As shown in the table, at the lowest test points (9 and 5 l/s) the Reynolds number indicates that the flow is in either in the laminar or laminar-turbulent transition region. For clarity vertical broken lines are shown on graphs which include data in these regions to indicate the boundaries of the transition region (ie Re = ). 5 ANALYSIS AND PRESENTATION Results are presented in graphical and tabular form and described in the following terms: Error: The difference between the volumetric flowrate indicated by the meter under test and the recorded reference volumetric flowrate, expressed as a percentage and plotted against velocity. V E = ind V V std std 100% (1) Meter factor: The factor by which the indicated volume should be multiplied in order to obtain the reference or true volume F = V std Vind (2) Repeatability: A measure of the random uncertainty in measurement defined as: R = 2. 83s E, where s E = ( E E ) n 1 2, (3) Report No: 351/99 Page 7 of 81

9 n is the number of repeat measurements at a single test condition. Non-linearity: The maximum difference between errors within a given velocity range (based on the average of a number of repeats at each point). Residual error: The error calculated after the indicated volumetric flowrate has been corrected by meter factor determined from calibration results as a function of indicated flowrate. 6 RESULTS 6.1 Danfoss Sonoflo 3000/ Baseline calibration The results of the baseline calibration on the Danfoss Sonoflo, presented in Figures and 6.1.2, show a nonlinearity of approximately 8.5% over the velocity range tested. For velocities above m/s the average error is 1.2% and the non-linearity is approximately 1%. The repeatability is 1.4% at 5 m/s and is better than 5% for velocities above m/s Error (%) Figure Test file 6722/4, Baseline calibration Repeatability (%) Figure Test file 6722/4, Repeatability Report No: 351/99 Page 8 of 81

10 Figure shows a third order polynomial curve fitted to meter factor versus indicated flowrate for velocities greater than m/s. Using the parameters of this curve to correct the indicated flowrate, the residual errors are less than ±0.12% as illustrated in Figure Meter factor y = E-08x E-05x E-04x E-01 R 2 = E Indicated flowrate (l/s) Figure Test file 6722/4, Polynomial fit, v > m/s Figure Test file 6722/4, Residual error after correction, v > m/s Reproducibility test A reproducibility test was conducted six days after the baseline calibration. The results are presented in Figure which shows the baseline calibration as a broken line. Using the parameters of the baseline calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the residual errors are less than ±0.1% for all but the two points at 1 m/s as illustrated in Figure Report No: 351/99 Page 9 of 81

11 /5 6722/4 Error (%) Figure Test file 6722/5, Reproducibility test /5 (ref. 6722/4) 6722/4 (ref. 6722/4) - - Figure Test file 6722/5, Residual error after baseline correction, v > m/s Electronics replacement test The signal converter (type SONO3000 compact housing, Serial No N147) was removed and replaced with a nominally identical unit (Serial No N147). No changes were made to software parameters prior to recalibration. The recalibration results, presented in Figures and 6.1.8, show little deviation from the baseline in terms of error, linearity or repeatability. Using the parameters of the baseline calibration curve fit to correct the indicated flowrate for velocities greater than m/s, residual errors are less than ±% as illustrated in Figure The residual errors indicate a slight negative bias, having a mean value of -72%. A new polynomial curve fit to meter factor versus indicated flowrate for velocities greater than m/s was derived to serve as the new baseline. Report No: 351/99 Page 10 of 81

12 /6 6722/4 Error (%) Figure Test file 6722/6, Electronics replacement test 1.6 Repeatability (%) /6 6722/4 Figure Test file 6722/6, Repeatability /6 (ref. 6722/4) 6722/4 (ref. 6722/4) - - Figure Test file 6722/6, Residual error after baseline correction, v > m/s Report No: 351/99 Page 11 of 81

13 6.1.4 Transducer cable replacement test The four single-core coaxial cables used to connect the SONO3200 transducers to the terminals of the SONO3000 remote mounting fixture were removed and each shortened from 15 m to 4 m in length. The correct cable length (to the nearest half meter) was then entered in the meter software in the application parameter submenu. The recalibration results, presented in Figure , show little deviation from the previous test. Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, residual errors are less than ±% as illustrated in Figure The residual errors indicate a slight positive bias, having a mean value of 79% /7 6722/6 Error (%) Figure Test file 6722/7, Cable replacement test /7 (ref. 6722/6) 6722/6 (ref. 6722/6) - - Figure Test file 6722/7, Residual error after baseline correction, v > m/s Application zero test To evaluate the effect of automatic adjustment of the meter zero setting on the linearity of the meter at low velocity, an application zero was performed. The application zero setting was conducted by closing the valves at either end of the test section and selecting the new auto zero option in the set zero menu. Report No: 351/99 Page 12 of 81

14 Successful zero setting was indicated by the result 0000 l/s being displayed on the LCD screen. The results of a calibration performed subsequent to this action, presented in Figure , show little deviation from the results of the previous test. Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, residual errors are less than ±% as illustrated in Figure The residual errors indicate a slight positive bias, having a mean value of 58% /8 6722/7 Error (%) Figure Test file 6722/8, Application zero test /8 (ref. 6722/7) 6722/7 (ref. 6722/7) - - Figure Test file 6722/8, Residual error after baseline correction, v > m/s Transducer assembly exchange test The SONO3200 O-ring type transducer comprises a piezoceramic crystal seated in stainless steel housing which isolates the active element from the fluid. All four transducer assemblies were removed from the flow tube when the line was drained down. They were then reinstalled in different locations, the transducers taken from locations A (lower track, upstream) and C (upper track, upstream) being interchanged with those in locations D (upper track, downstream) and B (lower track, downstream) respectively. The recalibration results, presented in Figure , show a significant deviation from the previous tests. Non-linearity is approximately 6.5% over the velocity range tested. For velocities above m/s the average error has increased to approximately 2% and the non-linearity is approximately 1.3%. Report No: 351/99 Page 13 of 81

15 /9 6722/8 7.0 Error (%) Figure Test file 6722/9, Transducer assembly exchange test Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the mean residual error is 0.81% and the variation about this mean is less than ±0.15%, as shown in Figure /9 (ref. 6722/8) 6722/8 (ref. 6722/8) -0.1 Figure Test file 6722/9, Residual error after baseline correction, v > m/s First periodic test The first periodic test was performed on the Danfoss Sonoflo three months after the transducer exchange test. The recalibration results, presented in Figure , show little deviation from the results of the previous test. Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the mean residual error is -52% and the variation about this mean is less than ±0.13% as shown in Figure Report No: 351/99 Page 14 of 81

16 /7 6722/9 7.0 Error (%) Figure Test file 6737/7, First periodic test /7 (ref. 6722/9) 6722/9 (ref. 6722/9) - Figure Test file 6737/7, Residual error after baseline correction, v > m/s As illustrated in Figure the repeatability is essentially unchanged in comparison with the baseline even though the value calculated at 3 m/s is approximately twice the baseline result. The scatter of results around the best-fit in previous calibrations qualifies this remark. Report No: 351/99 Page 15 of 81

17 1.6 Repeatability (%) /7 6722/4 Figure Test file 6737/7, Repeatability Second periodic test The second periodic test was performed on the Danfoss Sonoflo ten weeks after the first periodic test. Prior to the test a correction factor of was entered into the meter software to adjust the output by -1.68% and remove the offset apparent in the results. The recalibration results, presented in Figure , show the expected shift in calibration. To compare for characteristic deviations the flowrates were divided by to back-calculate the results. Using the parameters of the previous calibration curve fit to correct the backcalculated flowrate for velocities greater than m/s the mean residual error is -64%. However, there is a trend evident in the residual error versus velocity, as illustrated in Figure which shows overall variation from -% at 1 m/s to 3% at 4.5 m/s. It is considered that this may be evidence of an installation effect as the test line configuration was changed in the interval between the first and second periodic tests. The line change involved insertion of a spool of shorter length but nominally identical bore upstream of the Danfoss meter, shortening the distance from the upstream manifold by approximately five diameters. As illustrated in Figure the repeatability is essentially unchanged in comparison with the baseline /7 6722/9 Back-calculated Error (%) Figure Test file 6752/7, Second periodic test (correction factor entered) Report No: 351/99 Page 16 of 81

18 /7 (ref. 6722/9) 6722/9 (ref. 6722/9) - Figure Test file 6752/7, Residual error after baseline correction, v > m/s 1.6 Repeatability (%) /7 6722/4 Figure Test file 6752/7, Repeatability Third periodic test A third periodic test was performed on the Danfoss Sonoflo three months after the second periodic test. The recalibration results, presented in Figure , show little deviation from the previous calibration. Using the parameters of the previous calibration curve fit to correct the flowrate for velocities greater than m/s the mean residual error is 23%, as illustrated in Figure As illustrated in Figure the repeatability is essentially unchanged in comparison with the previous test. Report No: 351/99 Page 17 of 81

19 /1 6752/7 Error (%) Figure Test file 6769/1, Third periodic test /1 (ref. 6752/7) 6752/7 (ref. 6752/7) - Figure Test file 6769/1, Residual error after baseline correction, v > m/s Report No: 351/99 Page 18 of 81

20 1.6 Repeatability (%) /1 6752/7 Figure Test file 6769/1, Repeatability Repeat transducer assembly exchange test Due to concerns over the magnitude of the shift in the transducer replacement test (section 6.1.6) the transducer assemblies were removed from the meter and replaced in their original location. Danfoss advised that in the event of transducer failure the piezoceramic element could be removed from its housing without the necessity of removing the whole transducer assembly. The calibration results obtained after the transducer assemblies were returned to their original locations are presented in Figure The results show a significant deviation from the previous test. Non-linearity is approximately 7 % over the velocity range tested. For velocities above m/s the average error is approximately -0.7 % and the non-linearity approximately 0.7 % /1 6769/1 Error (%) Figure Test file 6780/1, Transducer assembly exchange test Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the mean residual error is 1 % and the variation about this mean is less than ±0.15%, as shown in Figure Note that this deviation is of similar magnitude but opposite sign to the first transducer replacement test. As illustrated in Figure the repeatability is essentially unchanged in comparison with the previous test. Report No: 351/99 Page 19 of 81

21 /1 (ref. 6769/1) 6769/1 (ref. 6769/1) Figure Test file 6780/1, Residual error after baseline correction, v > m/s /1 6769/1 Repeatability (%) 0.8 Figure Test file 6780/1, Repeatability By back-calculating to remove the effect of the change of the correction factor, we can compare the calibration with the result obtained before the transducers were first exchanged as shown in Figures and These results show that the calibration result is close to that obtained when the transducers were first in their original locations, the residual error of % at 1 m/s diminishing to a negligible level as the velocity increases. Report No: 351/99 Page 20 of 81

22 /1 6722/8 Back-calculated Error (%) Figure Test file 6780/1, Comparison with transducer assemblies in original locations /1 (back calculated, ref. 6722/8) 6722/8 (ref. 6722/8) - Figure Test file 6780/1, Comparison with transducer assemblies in original locations Transducer element replacement test Following the return of the transducer assemblies to their original positions, one of the piezoceramic elements was removed from its housing and replaced by a nominally identical element. The correction factor entered in the Danfoss SONO3000 electronics was set back to 1 to facilitate comparison of the results with the calibration conducted prior to any interference with the transducers (test 6822/8). Report No: 351/99 Page 21 of 81

23 The calibration results obtained are presented in Figure By back-calculating to remove the effect of the change of the correction factor, we can compare the calibration with the previous result. These results show that the calibration result is very close to that obtained before the transducer element was replaced, the residual error having an average value of 9% as illustrated in Figure /1 6780/1 Back-calculated Error (%) Figure Test file 6797/1, Transducer element test /2 (ref. 6780/1) 6780/1 (ref. 6780/1) Figure Test file 6797/1, Transducer element test As illustrated in Figure the repeatability is essentially unchanged in comparison with the previous test. Report No: 351/99 Page 22 of 81

24 /1 6780/1 Repeatability (%) Figure Test file 6797/1, Transducer element test The reason for removing the correction factor was to allow comparison with the test conducted before any change to the transducers. The comparison results are shown in Figures and These figures show that the result is close to that obtained when the transducers were first in their original locations and before exchange of the piezoceramic element. The residual error is approximately 0.3 % at 1 m/s and is practically zero at velocities above 2 m/s. It should be realised that this result is the combined effect of removing the transducer assemblies, replacing one of the piezoceramic elements and returning the transducer assemblies to their proper locations /1 6722/8 6.0 Error (%) Figure Test file 6797/1, Transducer element test comparison Report No: 351/99 Page 23 of 81

25 /2 (ref. 6722/8) 6722/8 (ref. 6722/8) - Figure Test file 6797/1, Transducer element test comparison Forth periodic test The fourth periodic test was performed on the Danfoss Sonoflo three months after the transducer element replacement test. The recalibration results, presented in Figure , show little deviation from the previous calibration. Using the parameters of the previous calibration curve fit to correct the flowrate for velocities greater than m/s the mean residual error is 13%, as illustrated in Figure As illustrated in Figure the repeatability is essentially unchanged in comparison with the previous test /1 6797/1 6.0 Error (%) Figure Test file 6810/1, Forth periodic test Report No: 351/99 Page 24 of 81

26 /1 (ref. 6797/1) 6797/1 (ref. 6797/1) Figure Test file 6810/1, Residual error after baseline correction, v > m/s /1 6797/1 Repeatability (%) 0.8 Figure Test file 6810/1, Repeatability Final periodic test The final periodic test was performed on the Danfoss Sonoflo seven months after the previous test. The recalibration results, presented in Figure , show little deviation from the previous calibration. Using the parameters of the previous calibration curve fit to correct the flowrate for velocities greater than m/s the mean residual error is 61%, as illustrated in Figure As illustrated in Figure the repeatability is essentially unchanged in comparison with the previous test. Report No: 351/99 Page 25 of 81

27 /2 6810/1 7.0 Error (%) Figure Test file 6841/2, Fifth periodic test 6841/2 (ref. 6810/1) 6810/1 (ref. 6810/1) Figure Test file 6841/2, Residual error after baseline correction, v > m/s /2 6810/1 Repeatability (%) 0.8 Figure Test file 6841/2, Repeatability Report No: 351/99 Page 26 of 81

28 Summary of Danfoss Sonoflo results The results obtained show that there was no significant effect of exchanging electronics, replacing cables or replacing a transducer element on the calibration of the Sonoflo. The only exceptions were the tests where the transducer assemblies were interchanged (See sections and ). During all of the tests the repeatability was essentially unchanged. The repeat calibrations showed no sign of drift outside of the accuracy envelope defined by the meter repeatability and calibration system uncertainty. All of the results obtained with the Danfoss Sonoflo, with the exception of tests 6722/9 and 6780/1 are summarised in Figures and /5 (ref. 6722/4) 6722/6 (ref. 6722/4) 6722/7 (ref. 6722/6) 6722/8 (ref. 6722/7) 6737/7 (ref. 6722/9) 6752/7 (ref. 6722/9) 6769/1 (ref. 6752/7) 6797/1 (ref. 6780/1) 6810/1 (ref. 6797/1) 6841/2 (ref. 6810/1) Flowrate (l/s) Figure A summary of all residual errors, excluding 6722/9 and 6780/1 1.6 Repeatability (%) /4 6722/6 6737/7 6752/7 6769/1 6780/1 6797/1 6810/1 6841/2 Figure A summary of all repeatability results Report No: 351/99 Page 27 of 81

29 6.2 Panametrics XTM Baseline calibration Single-channel mode using extended well transducers The XMT868 Serial No 224 was configured to operate in a single-channel transit time mode using the extended well transducer pair. The results of this baseline calibration, presented in Figures and 6.2.2, show a nonlinearity of approximately 11% over the velocity range tested. For velocities above m/s the uncalibrated average error is 4.7% and the non-linearity is approximately 0.9%. Application of a correction factor of shifts the curve to the position indicated by the bold line in Figure The repeatability is 5.7% at 5 m/s and is better than % for velocities above m/s Error (%) Figure Test file 6737/1, Single channel baseline calibration % 1.4 Repeatability (%) Figure Test file 6737/1, Repeatability Figure shows a third order polynomial curve fit to meter factor versus indicated flowrate for velocities greater than m/s from test file 6737/1. Using the parameters of this curve to correct the indicated flowrate, the residual errors are less than ±2% as illustrated in Figure Report No: 351/99 Page 28 of 81

30 Meter factor y = E-08x E-05x E-04x E-01 R 2 = E Indicated flowrate (l/s) Figure Test file 6737/1, Polynomial fit, v > m/s Figure Test file 6737/1, Residual error after correction, v > m/s Transducer replacement test - Single-channel mode using extended well transducers The type 42 transducers were removed from the welded 1 inch NPT bosses and reinstalled in opposing locations while the test line was drained down. Prior to reinstallation of the transducers, data for velocity of sound (VOS) versus temperature was determined for the lubricant oil. Following reinstallation and reconnection of the transducers, line temperature and indicated velocity of sound were monitored and the face-to-face separation distance parameter was subsequently reduced by 2.8 mm to bring the indicated result into agreement with VOS calibration data. The recalibration results, presented in Figure 6.2.5, show fairly minor deviations from the previous calibration results. Non-linearity is approximately 3.6% over the velocity range tested. For velocities above m/s the average error is approximately 4.6% and the non-linearity is approximately 0.95%. Report No: 351/99 Page 29 of 81

31 /4 6737/1 Error (%) Figure Test file 6737/4, Extended-well transducer exchange test Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the mean residual error is -5% and the variation about this mean is less than ±0.3%. Figure shows a slight negative trend evident in residual error versus velocity. As illustrated in Figure 6.2.7, repeatability is essentially unchanged in comparison with the baseline /4 (ref. 6737/1) 6737/1 (ref. 6737/21) Figure Test file 6737/4, Residual error after baseline correction, v > m/s Report No: 351/99 Page 30 of 81

32 % 5.7 % Repeatability (%) /1 6737/4 Figure Test file 6737/4, Repeatability Baseline calibration Single-channel mode using Panadapta transducers The XMT868 Serial No 224 was configured to operate in a single-channel transit time mode using the Panadapta transducer pair. The results of this baseline calibration, presented in Figures and 6.2.9, show a nonlinearity of approximately 13% over the velocity range tested. For velocities above m/s the average error is 2.8% and the non-linearity is approximately 0.7%. Application of a calibration factor of shifts the curve to the position indicated by the bold line in Figure The repeatability is 4.5% at 5 m/s and varies between 1% at 1 m/s and 0.12% at 5 m/s Error (%) Figure Test file 6737/2, Single channel baseline calibration Report No: 351/99 Page 31 of 81

33 % 1.4 Repeatability (%) Figure Test file 6737/2, Repeatability Figure shows a polynomial curve fitted to meter factor versus indicated flowrate for velocities greater than m/s from test file 6737/2. Using the parameters of this curve to correct the indicated flowrate, the residual errors are approximately ±% at 1 m/s and less than ±6% at 5 m/s as illustrated in Figure Meter factor y = E-08x E-06x E-05x E-01 R 2 = E Indicated flowrate (l/s) Figure Test file 6737/2, Polynomial fit, v > m/s Report No: 351/99 Page 32 of 81

34 Figure Test file 6737/2, Residual error after correction, v > m/s Transducer replacement test - Single-channel mode using Panadapta transducers The type 74 transducers were removed from the Panadapta mounting plugs and reinstalled in opposing locations while the test line was flowing. No changes were made to software configuration parameters. The recalibration results, presented in Figure , show fairly noticeable deviations from the previous calibration results. Non-linearity is approximately 6.4% over the velocity range tested. For velocities above m/s the average error is approximately 3% and the non-linearity is approximately 0.8% /3 6737/2 Error (%) Figure Test file 6737/3, Transducer assembly exchange test Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the mean residual error is % and the variation about this mean is less than ±% at 1 m/s and less than 0.13% at higher velocities as shown in Figure As illustrated in Figure the repeatability is broadly comparable with the baseline. Report No: 351/99 Page 33 of 81

35 /3 (ref. 6737/2) /2 (ref. 6737/2) Figure Test file 6737/3, Residual error after baseline correction, v > m/s % 2.9 % Repeatability (%) /2 6763/3 Figure Test file 6737/3, Repeatability Electronics replacement test Maintaining the single-channel configuration (extended well transducers) the XMT868 signal converter (serial no 224) was removed and replaced with a nominally identical unit (serial no 223). Using the manufacturers configuration software the set-up parameters from the first unit were downloaded to the new unit from a PC file. No adjustments were made to these configuration parameters. The recalibration results presented in Figures and show relatively minor deviation from the baseline in terms of linearity or repeatability with the exception of the lowest velocity set-points close to and in the transition region. Using the parameters of the baseline calibration curve fit to correct the indicated flowrate for velocities greater than m/s, residual errors are less than ±0.35% as illustrated in Figure The mean value of the residual errors is -38%. Report No: 351/99 Page 34 of 81

36 Error (%) /6 6737/4-1 Figure Test file 6737/6, Electronics replacement test % 14.6 % Repeatability (%) /4 6737/6 Figure Test file 6737/6, Repeatability Report No: 351/99 Page 35 of 81

37 /6 (ref. 6737/4) 6737/4 (ref. 6737/4) Figure Test file 6737/6, Residual error after baseline correction, v > m/s Baseline calibration Dual-channel mode with invasive transducers The XMT868 (Serial No 224) was configured to operate in a dual-channel transit time mode using the both the extended-well and Panadapta transducer pairs. Calibration factors of and were input to correct the measurements on the extended-well and Panadapta transducers respectively. The results of the calibration, presented in Figures and , show a non-linearity of approximately 4.2% over the velocity range tested. For velocities above m/s the average error is 0.16% and the non-linearity is approximately 0.9%. The repeatability is 3.4% at 5 m/s and varies between 0.9% at 1 m/s and 0.18% at 5 m/s Error (%) Figure Test file 6737/7, Dual-channel baseline calibration Report No: 351/99 Page 36 of 81

38 % 1.4 Repeatibility (%) Figure Test file 6737/7, Repeatability Figure shows a polynomial curve fitted to meter factor versus indicated flowrate for velocities greater than m/s from test file 6737/7. Using the parameters of this curve to correct the indicated flowrate, the residual errors are within ±% at 1 m/s and within ±0.11% at velocities above this limit as illustrated in Figure Meter factor y = E-08x E-06x E-04x E-01 R 2 = E Indicated flowrate (l/s) Figure Test file 6737/7, Polynomial fit, v > m/s Report No: 351/99 Page 37 of 81

39 First periodic test Figure Test file 6737/7, Residual error after correction, v > m/s The first periodic test was performed on the XMT868 (with boss-mounted transducers) three months after the transducer exchange test. The recalibration results, presented in Figure , show little deviation from the results of the previous test. Using the parameters of the previous calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the residual error is within ±0.35% at 1 m/s and ±3% above this velocity as illustrated in Figure The mean value of the residual errors is -68% /1 6737/7 Error (%) Figure Test file 6769/1, First periodic test Report No: 351/99 Page 38 of 81

40 6769/1 (ref. 6737/7) 6737/7 (ref. 6737/7) Figure Test file 6769/1, Residual error after baseline correction, v > m/s As illustrated in Figure the repeatability is essentially unchanged in comparison with the baseline % /7 6769/1 0.8 Repeatability (%) Figure Test file 6769/1, Repeatability Second periodic test The second periodic test was performed on the XMT868 (with boss-mounted transducers) seven months after the transducer exchange test. The recalibration results, presented in Figure , show little deviation from the results of the baseline test although there does appear to be a slight positive tendency at the higher velocities. Using the parameters of the baseline calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the residual error is less than % across the range as illustrated in Figure The mean value of the residual errors is 0.125%. As illustrated in Figure the repeatability is essentially unchanged in comparison with the baseline. Report No: 351/99 Page 39 of 81

41 /2 6737/7 Error (%) Figure Test file 6797/2, Second periodic test /2 (ref. 6737/7) /7 (ref. 6737/7) - Figure Test file 6797/2, Residual error after baseline correction, v > m/s % Repeatability (%) /7 6797/2 Figure Test file 6769/2, Repeatability Report No: 351/99 Page 40 of 81

42 6.2.9 Third periodic test The third periodic test was performed on the XMT868 (with boss-mounted transducers) ten months after the transducer exchange test. The recalibration results, presented in Figure , show a general positive bias with respect to the baseline. Using the parameters of the baseline calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the residual error is less than 5 % across the range as illustrated in Figure The mean value of the residual errors is 6 %. As illustrated in Figure the repeatability is essentially unchanged in comparison with the baseline /1 6737/7 Error (%) Figure Test file 6810/1, Third periodic test /1 (ref. 6737/7) 6737/7 (ref. 6737/7) - Figure Test file 6810/1, Residual error after baseline correction, v > m/s Report No: 351/99 Page 41 of 81

43 % Repeatability (%) /7 6810/1 Figure Test file 6810/1, Repeatability Final periodic test The final periodic test was performed on the XMT868 (with boss-mounted transducers) 18 months after the transducer exchange test. The recalibration results, presented in Figure , show a very slight negative bias with respect to the baseline. Using the parameters of the baseline calibration curve fit to correct the indicated flowrate for velocities greater than m/s, the residual error is less than % across the range as illustrated in Figure The mean value of the residual errors is 0.1 %. As illustrated in Figure the repeatability is essentially unchanged in comparison with the baseline /2 6737/7 Error (%) Figure Test file 6841/2, Final periodic test Report No: 351/99 Page 42 of 81

44 /2 (ref. 6737/7) 6737/7 (ref. 6737/7) Figure Test file 6810/1, Residual error after baseline correction, v > m/s % Repeatability (%) /7 6841/2 Figure Test file 6810/1, Repeatability Report No: 351/99 Page 43 of 81

45 Summary of Panametrics XMT868 results (boss-mounted transducers) The results obtained show that there was no significant effect of exchanging electronics, replacing cables or interchanging transducers on the calibration of the XMT868. During all of the tests the repeatability was essentially unchanged. One repeat calibration result indicated a positive shift of a few tenths of a percent relative to the baseline. However, the subsequent calibration was in better agreement. All of the results obtained are summarised in Figures and Note that the repeatability in percent is shown on a logarithmic scale /3 (ref. 6737/2) 6737/4 (ref. 6737/1) 6737/6 (ref. 6737/4) 6769/1 (ref. 6737/7) 6797/2 (ref. 6737/7) 6810/1 (ref. 6737/7) 6841/2 (ref. 6737/7) Flowrate (l/s) Figure A summary of all residual errors 1 Repeatability (%) 6737/1 6737/2 6737/3 6737/4 6737/6 6737/7 6769/1 6797/1 6810/1 6841/2 0.1 Figure A summary of all repeatability results Baseline calibration (dual-channel mode with clamp-on transducers) The XMT868 (serial no. 223) was set-up in dual channel mode using two pairs of clamp-on transducers mounted to produce two single-reflection sound paths in orthogonal planes. The results of the baseline calibration, presented in Figures and , show a non-linearity of approximately 8.6% over the velocity range Report No: 351/99 Page 44 of 81

46 tested. For velocities above m/s the average error is 2.9% and the non-linearity is approximately 1%. The repeatability is 0.7% at 5 m/s and is better than 5% for velocities greater than 1 m/s Error (%) Figure Test file 6752/4, Baseline calibration 0.8 Repeatability (%) Figure Test file 6752/4, Repeatability Figure shows a third order polynomial curve fit to meter factor versus indicated flowrate for velocities greater than m/s. Using the parameters of this curve to correct the indicated flowrate, the residual errors are less than ±0.3% at 1 m/s and less than ±0.13% at higher velocities as illustrated in Figure Report No: 351/99 Page 45 of 81

47 Meter factor y = E-08x E-05x E-04x E-01 R 2 = E Indicated flowrate (l/s) Figure Test file 6752/4, Polynomial fit, v > m/s Figure Test file 6752/4, Residual error after correction, v > m/s Clamp-on transducer replacement test (1) The type 30 transducers were mounted using yokes and straps. To evaluate the sensitivity of the meter to removal and replacement, the transducers were removed and the mounting fixtures were left in-situ. The locations of the transducer pairs were interchanged as were upstream or downstream transducers (whilst the pair matching was maintained). The recalibration results, presented in Figure , show a significant deviation from the baseline calibration. Non-linearity is approximately 10% over the velocity range tested. For velocities above m/s the average error has increased to approximately 4% and the non-linearity is approximately 1.3%. Report No: 351/99 Page 46 of 81