CALORIFIC VALUE MEASUREMENT WITH HONEYWELL I-£VT- 100

by

I Bates and T DaviesTotal Oil Marine pic

Paper 1.1

NORTH SEA FLOW METERING WORKSHOP 198818-20 October 1988

National Engi.l\~ring LaboratoryEast Kilbride, Glasgow

CALORIFIC VALUE MEASUREMENT WITH THE HONEYWELL HVT-lOO

IAN P. BATES

TOTAL OIL MARINE PLC, ABERDEENJULY 1988

1. INTRODUCTIONThis paper describes the requirement to replace theContinuous Recording Calorimeter with a more reliable andaccurate instrument for the purposes of Fiscal CalorificValue measurement of Natural Gas.The options considered are reviewed andattention is given to the instrument selected,its evaluation and final installation.

particulardetailing

The instrument selected was the 'Heating Value Transmitter'manufactured by Honeywell.1.1 The Installation

Natural gas processed by the St. Fergus Terminal istransferred to the British Gas site for nationwidedistribution. This transfer takes place through twoparallel pipes, each metered by separate meteringstations termed Phase I and Phase II.Revenue from British Gas is based on energy, thereforethe calorific value (CV) of the gas flowing throughboth pipes is required to be measured and transmittedto the flow computers for heat flow computation.The measurement of the sales gas gross calorific valuehas been performed by 4 continuous recordingcalorimeters, two units installed on each Phase in aclosely controlled and monitored air conditioned room.A sample line runs from the metering station pipeheader to both calorimeters, allowing a continuousanalysis of the same gas sample.

1.2 Existing CalorimetersSince gas first flowed at St. Fergus calorific valuehas been measured by the industry accepted standarddevice the continuous recording calorimeter,manufactured by Cutler Hammer.

PAGE 1

PAGE 2

As these units were designed in the 1930's theirconstruction is almost entirely of moving mechanicalparts, many operating in direct contact with water.The units display a sensitivity to ambient temperatureand must therefore be operated in a closely controlledenvironment at all· times. The moving parts aresubject to wear which results in instrument drift.As with any piece of analysis equipment, themanufacturers quote an expected accuracy. In the caseof the Cutler Hammer calorjmeter this is ± 0.5%,corresponding to + 0.20 MJ/smNormal operation calls for close monitoring ofinstrument performance on a daily basis, this includeschecks relating to the air conditioning systems. Theaccuracy of measurement is verified on a weekly basisby placing each unit on standby in turn and passingcerti fi ed test gases through the devi ce under test.A close examination of the unit is also required atthis stage.Planned maintenance is required on each calorimeterevery three months by taking the units out of servicein rotation. In addition to routine maintenance, themechanical and electrical performance has to berigorously tested. If unacceptable wear is found, theinstrument is stripped and the area of wear locatedand remedied. This operation may take several daysand large quantities of spares have to be stocked sothat there are no unnecessary delays due to shortages.During this maintenance period no standby unit isavailable. If a fault develops with the dutycalorimeter, a preset CV may be required which usuallycalls for corrections to production totals to be made.

2.0 THE REQUIREMENT FOR A REPLACEMENT2.1 The Continuous Recording Calorimeters gave rise to the

following areas of concern.a) Obsolescence

Cutl er Hammer no longer manufacture compl etecalorimeters or spares. Al though spares areavailable their quality is often poor. Thishas resulted in a programme of refurbishmentbeing carried out, replacing critical itemswith specially manufactured items followed byaccurate alignment.

b) ReliabilityIn addition to instrument drift detected byclose monitoring, operational problems havebeen experienced due to a number of failures ofthe calorimeters. Occasionally, a calorimeter

PAGE 3

would trip while on-line, completelyunexpectedly, rigorous examination revealing noapparent fault. These failures weresubsequently traced to porous castings.

c) AccuracyAs gas revenue is directly proportional tomeasured calorific value, we are obliged tooperate the calorimeters at the best accuracylevel achievable. However, this level islimited by relative inadequacies in the design,materials used and manufacturing tolerancesemployed in these rather antiquated devices.Although the calorimeters may be calibratedaccurately at a single point it is ratherdifficult to maintain the accuracy at thevarious CVs encountered when in operation.These units are sensitive to changes in ambienttemperature, if faults develop with the airconditioning system, inaccurate measurementwill result. To a lesser extent thecalorimeters are affected by changes inatmospheric pressure, this results inmisleading calibration results if attemptedduring periods of extreme barometic pressure.The units are slow to respond to transients andmay therefore not follow CV excursionscorrectly. The slow response time is aconsiderable problem when relighting followinga shutdown. During this operation thecalorimeters are extremely temperamental.

d) MaintenanceDue to the high level of manual monitoring andfrequent intervention required in addition tothe 1engthy pl anned rnaintenance peri ods, thecalorimeters are unreasonably dependent onskilled manpower. In view of the obligation tomaintain accurate measurement at all times,little can be done to reduce this manpowerrequirement.As the calorimeters become older and sparesbecome rarer, the maintenance problems willincrease with a corresponding deterioration inreliability and accuracy.

e) New TechnologyIt is now widely recognised that significantimprovements in metering accuracy andreliability have been achieved through thewidespread adoption of digital flow computers

PAGE 4

and largely sol id state instrumentation, suchas modern differential pressure transmittersand densitometers. Essentially moving partshave been either reduced to an absolute minimumor eliminated entirely in order to providethese advances.Rather than continuing with obsolete equipment,it has been our policy to appraise deviceswhich employ new technology.

3.0 REPLACEMENT OPTIONS3.1 Several devices are available to measure the calorific

value of natural gas. These devices may becategorised into three types.a) Calorimetry methods which measure the heat

released by the combustion of a metered volumeof gas. The Cutler Hammer calorimeter was anexample of this type.

b) Gas analysis methods which determine thecalorific value by calculation from an analysisof the concentrations of individual componentsin the natural gas mixture. Instruments whichmay be included in this classification are:gas chromatographs, and spectrometers.

c) Inferred measurement methods which calculatethe calorific value by measurement of aparameter which is related to the calorificvalue. Examples of this type are the HoneywellHVT-I00 and the Therm-Titrator.

3.2 Each of these options was considered, but we requiredthat items to be short listed should have already beenproven in service as we could not evaluate allpossibil ities.In the UK sector of the North Sea, Operators haveconsidered the gas chromatograph as a suitablereplacement to the continuous recording calorimeterand this device has gained acceptance as a point ofsale instrument.The Honeywell HVT-IOO has received good reports and iswidely used in the USA.The gas chromatograph and HVT-IOO are now discussed indetail.

3.3 Gas Chromatograph3.3.1 Advantages

a} Already acceptedmeasurement in UK.

for Point of sale

b} A fundamental advantage of using gaschrrimatography is that additionalparameters such as relative density andcompressibility may be calculated and usedin metering calculations, reducing theoverall number of instruments required.Special air conditioned rooms are notrequired, further simplifying theinstallation.However, as Operator of a large gastransportation system, we have considerableexperience in the use of gaschromatographs, both in the 1 aboratoryenvironment and offshore on MCPOI whereon-line GCs are used.

3.3.2 For on line use at St. Fergus the followingdisadvantages arise.a} The nature of Frigg gas does not allow a

simple analysis of light carbon componentsdue to the long tail of heavy components.In order to allow an accurate analysis andsubsequent determination of calorific valuea protracted cycle time would be required.In addition, the type of chromatograph toperform this analysis on-line has not yetbeen sufficiently developed.

b} Even if accuracy were to be compromi sed, atypical cycle time for the evaluation of anupdated calorific value would take aminimum of 10 minutes and this wouldpresent problems in monitoring gas quality,as a continuous output will be required forcontrol purposes if blending/nitrogeninjection is to be performed efficiently.

c} The detection of unexpected components mayresult in large errors.

d} The chromatography columns are subject tocontamination and associated inaccuracy.

PAGE 5

3.4 Honeywell HVT-I003.4.1 Advantages

The HVT-I00 hadorganisations andfollowing areas:a) The manufacturers specification offers good

accuracy under our process conditions.

good reports from othershowed promi se in the

b) The unit isdesigned tointervention.

self calibratingoperate with

and isminimal

c) The minimum amount of moving parts are usedwhich promises good reliability. Thedesign is microprocessor based with modernself diagnostic routines to aid rapid faultfinding. A modular construction isemployed which should allow quick exchangeof faulty components.

d) The HVT provides a continuous output ofmeasured calorific value and a fastresponse time is offered.

e) Unlike a conventional calorimeter the HVTis designed to tolerate variations inambient temperature and would thereforemaintain accurate measurement if the airconditioning system for the calorimetryroom were to fail.

3.4.2 Disadvantagesa) Not accepted for point of sale measurement

in UK at time of investigation.b) A Specific Gravity analyser is required in

addition to the HVT-I00 for heat flowcomputation to Fiscal Standards.

3.5 Following a two week familiarisation period with aunit on site, the HVT-I00 was selected as the devicemost likely to be able to demonstrate a worthwhileimprovement in accuracy and large savings inmaintenance costs over the existing calorimeters.

PAGE 6

PAGE 7

4.0 HVT-IOO4.1 Principle of Operation for the HVT-IOO

The HVT-IOO employs the well recognised principle,that for the gases in the paraffin (alkane) series,there is a direct relationship between the air to fuelratio required to achieve stoichiometric combustionand the calorific value of the gas. This relationshipalso holds for a mixture of gases such as present innatural gas (Fig 1).The air to fuel ratio is determined by a rotary mixingvalve, where the rotational speed influences theratio.Mixed gas and air is then burnt in a combustionchamber where stoichiometric combustion is determinedby monitoring the exhaust gas with a zirconium oxidesensor. This sensor detects the presence of oxygen inthe exhaust gas and provides an output to themicroprocessor controller which controls therotational speed of t h.e mixing valve. Thesecomponents form a closed loop control system operatingin the following manner to maintain stoichiometriccombustion (Fig 2).The zirconium probe develops its greatest outputvoltage when the oxygen content in the exhaust gas ~sat its lowest level. As the air to fuel ratio 1Sadjusted towards a leaner mixture, the sensors outputdecreases until a point is reached where it dropssharply, in a stepped fashion, to a level at which anyfurther increase in the air would cause the output totaper off into a nerstian region (Fig 3). The stepresponse in the sensors output marks the period duringwhich stoichiometric combustion occurs at the burner.The speed of rotation of the mixing valve is adjustedin accordance with the oxygen sensor output and themeasured rate of rotation is related to the calorificvalue. The calorific value is in turn shown on adigital display on the front panel and also scaled foroutput as a 4-20 mA signal.Self calibration is achieved by automaticallyswitching from sample gas to a calibration gas such asmethane. The HVT measures the CV of the cal ibrationgas and then adjusts itself to measure the known valueof the calibration gas input by the user. Theauto-calibration interval is also preset by the user.

4.2 InstallationAn HVT-IOO was purchased for evaluation and installedin the Phase II calorimeter room and connected inparallel with the No.2 calorimeter so that both theHVT and the calorimeter receive the same sample gaswithout affecting calorimeter operation.

The test unit, when compared to the existingcalorimeters under the same operating conditions,should prove to be a worthwhile improvement in thefollowing areas:a) Accuracy and repeatabilityb) Sensitivity to environmental conditionsc) Response timed) Stability of outpute) Reliabilityf) Maintenance requirementg) Availability of spare parts.

5.2 Accuracy and Repeatability5.2.1 Accuracy

In order to assess the accuracy of the HVT-IOOunder operating conditions, the unit was testedusing a number of specially formulated gaseswhich had been analysed and certified by theGas & Oil Measurement Branch of Department ofEnergy, Wigston.The test gas mixtures specified were chosen torepresent the range of natural gas mixtureslikely to be processed as St Fergus.Gross calorific values for each test gas werecalculated in accordance with ISO 6976 based onthe certified gas analysis provided byDepartment of Energy.

5.2.2 Method

5.0 EVALUATION TEST5.1 Evalyation Test Programme

An evaluation test programme was drawn up to includetests which allowed an assessment of the HVT'ssuitabil ity for accurate measurement of natural gascalorific value.If favourable results were obtained then the testfindings would be used as a basis for gaining approvalfor the HVT-IOO.Test Criteria

The tests were performed with the HVT-IOO andthe stand-by Cutler Hammer calorimeter piped inparallel, so that both instruments received thesame gas test. This allowed a directcomparison to be made under normal operatingconditions.The Cutler Hammer CV was read from the on-lineflow computer displays and the HVT CV was takenfrom the instruments digital display.

PAGE 8

This section may be split intosensitivity to changes in ambientsecondly, atmospheric pressure.

two main areas:temperature and

5.2.3 ResyltsHVT-IOO Results within +0.05MJ/sm3 for

gases likely to be measured atSt Fergus.The synthetic ass~ciated gas waswithin 0.10 MJ/sm

CH Calorimeter Results within +0.05MJ/sm3except the synt~etic associatedgas: +0.15MJ/sm

5.3 Sensitivity to Environmental Conditions

5.3.1 Ambient TemperatyreThe existing Cutler Hammer calorimeters arerecognised as being sensitive to ambienttemperature and for this reason are required tooperate in an air conditioned room having aclosely controlled air temperature.As the HVT was installed in the same room itwould only have been possible to alter the airtemperature if the cal ori meters had not beenoperating for some period during the evaluationtest. Unfortunately Phase II was required toflow gas continually during this period and soambient temperature effects could not bechecked.However, the HVT-I00 was designed from theoutset to operate in the open air and thiswould indicate a degree of insensitivity toambient temperature.This has been confirmed by other organisationscarrying out similar evaluation tests.

PAGE 9

5.3.2 Atmospheric PressureThe Cutler Hamm~r calorimeters at St Fergus areknown to display a certain degree ofsensitivity to atmospheric pressure.In order to assess any atmospheric pressureeffect on the HVT-IOO,an analysis of the autocalibration results was made.A sensitivity to atmospheric pressure changesshould result in a deviation from the expectedcalibration results.Barometric pressure readings were taken fromcontinuous barograph charts (kindly loaned byBritish Gas St Fergus).The effect of changing atmospheric pressure onthe HVT is shown graphically in graphs A-E.These graphs analyse data for 5 days chosen togive a variety of barometric patterns. Theseincluded i) Steady barometric pressure.ii) Steadily falling barometric pressure.iii) Steadily rising barometric pressure.iv) A pressure peak and v) A pressure trough.The measured CV data is for certified highpuri ty methane (37.77 MJ/sm3) using a 2 hourautocalibration period.Each graph consists of three parts:Top: Tabulated calibration gas

measured tV and correspondingbarometric pressure at 2 hourintervals. Calculated valuesare CV error (assuming 37.77MJ/sm3 is correct) and %change in barometric pressuresince the last calibration.

Middle: Plot of daily barometricpressure variation.

Bottom: Plot of calculated values fromtable.

5.3.3 FindingsGraphs A and B show that there is a definiteproportional linear relationship between CVerror and change in barometri c pressure. Therelationship can be roughly defined as:

tV = 1.0 MJ/sm3bPi n Hg

PAGE 10

t ::::T~

t =Tod(CV)dt -+ 0 as (T - To) ~ co

If all drift is due to the barometric pressurechange and fits the above relationship then:

t ::::T5 d(CV)t ::::To

::::

t ::::T

t ~ ToKdP dt

If pressure variation is assumed to be linearbetween any two cal ibrat ions then due to thecyclical nature of barometric pressure it wouldbe expected that:

i.e. long term bias error due to atmosphericpressure should tend to zero over a long periodif the above assumptions are correct.Taking the complete test period for the 2 hourcalibration data from 15 October 1987 to7 January 1988 (spanning more than 600calibrations); calculation gives averagemethane measured CV ::::37.763.Based on the assumption that pressure varies1 inearly with time and that measured CV errorwill be zero immediately following acalibration.Mean CV error ::::37.763 - 37.77 ::::-0.0035 MJ/sm3

2

Since3calibration gas CV is certified as ~7.77MJ/sm an uncertainty of + 0.005 MJ/sm isimpl ied.Therefore long term bias error in CV is withinthe tolerance of the calibration gas. It canbe confidentially stated that there was nodetectable long term bias error introduced intoCV measurement due to atmospheric pressurechanges when using the HVT with a 2 hourcalibration period.

SYMBOLST, t = TimedP :::: Change in barometric pressure

over time dtK =

d(CV) ::::Proportionality constantChange in measured calibration

gas CV over time dt.PAGE 11

PAGE 12

AUTOMATIC CALIBRATION5 .3 .4 The H V Tis de st g ned too per ate u sin g reg u 1a r

automatic calibration to compensate for anymeasurement drift. The calibration intervalmay be set by the user between 10 minutes and 4hours.In order to assess the effect of differentinterval s on the accuracy of CV measurement,the HVT was operated on normal sample gas withcalibration intervals of 4 hours, 2 hours and 1hour.During each cal ibration cycle the HVT samplesthe calibration gas with normal sample gasisolated. The calibration gas is measured fora 30 second peri od and if stabi 1 ity cri teri aare passed, the result given on the printout isthe mean value over the period.Ideally, this result should be identical to theknown calibration gas value, but if this is notthe case the HVT will automatically realignitself to remove the measured error beforereturning to normal sample gas.An analysis of these reported errors will allowan impression to be gained of the generalaccuracy of the HVT at different calibrationintervals, allowing the optimum calibrationinterval to be set.

5.3.5 FindingsIf the HVT-IOO does not drift or display anyerror then the reported val ues at eachauto-cal ibration test should always match thecalibration gas value held in the HVT as afixed parameter.The cal ibration gas used throughout was highpurity methane certified at 100% by Departme~tof Energy, ISO 6976 gives a CV of 37.77 MJ/SM .Thus ideal results should be a straight line atthis value with a standard deviation of O.

PAGE 13

4 Hoyr IntervalThe overall 3average resul t over 45 days is37.760 MJ/SM wi th a mean std. devi at ion of0.0300 MJ/SM3. This would infer a slightunder-regi~tration of calorific value by0.01 MJ/SM over this period if constant CV gashad been sampled. Although the errorencountered in measuring sample gas may bedi fferent to thi s val ue as one represents apure gas and the other a mixture, the smallmagnitude of this error is very impressive.2 Hour IntervalThe overall average resu1 t for 94 days was37.764 MJ/SM~ with a mean standard deviation of0.0308 MJ/SM. Thus this calibration intervalgives sl ight1y better results than a 4 hourinterval.This result is most impressive and shows that a2 hour calibration interval maintains accuratemeasurement when operating on sample gas and ishighly effective in reducing instrument drift.1 Hour IntervalThis test duration was limited to a period ofone week due to the high calibration gasconsumption rate.The overall average obtained was 37.754 MJ/SM3with 3a mean standard deviation of 0.0226MJ/SM .Although this interval resulted in less scatterthan the larger intervals, the overall averagewas sl ightly poorer. Thus the resu1 ts fromthis brief test indicate that there is nobenefit to be gained by using a 1 hourcalibration interval.

5.4 Response Time5.4.1 Method

The response time of the HVT-IOO to changes insample gas calorific value has been compared toan existing Cutler Hammer calorimeter usingtest gases.The gas inlet pipework to the calorimetersallows gas to be instantaneously switchedbetween normal sample gas and a test gas, orbetween two test gases. This operation isperformed using manually operated valves.

5.4.2 FindingsThe HVT was found to respond immedi ately to astep change 1n gas calorific value. The trendin all cases was near critical damping whetherin response to an increase or decrease in CV.No sign of overshoot was noted, the initialresponse stopping just short of the finaloscillating trend.The results may be summarised as follows:Time taken to reach 98.5%of step change in CV :20 sec per MJ/SM3

Additional time taken toattain final 1.5% of stepchange :60 secondsIn contrast to the HVT-I00 the C.H. calorimeterhas a far poorer response time.Switch!ng between test gases which differ by4MJ/SM , first response does not occur for over4 minutes.In this case, HVT-I00 would take 2 minutes 20seconds.Start-upFollowing a shutdown the HVT-I00 takesapproximately 4 minutes before the out put isrestored. The start-up procedure 1s verysimple: the sample gas supply is turned on andthe HVT front panel ON/OFF button is pressed.No output is provided during the 4 minutestabilisation period, the output is restored asa step change.Following a start-up the CH calorimeters areusually allowed to stabilise for a m r n t mun oftwo hours before being considered for duty.

PAGE 14

PAGE 15

5.5 STABILITY OF OUTPUT

5.5.1 The Cutler-Hammer calorimeters tend to displaya slightly oscillating output under steadyinput conditions. This trend is usuallyexaggerated by the 1ength of servi ce between"rebuilds and is thought to be caused by wearresulting in small degrees of misalignment ofthe rotating parts. A limit has been set as tothe max:'Jnum acceptable oscillation of0.10 MJ/SM peak to peak. Although anoscillation should not lead to poorer overallaccuracy, as the mean result over the course ofa day is the same for a constant output, thecalorimeter instantaneous maximum and minimumval ues may not be acceptab1 e. Thi sis due toboth duty standby calorimeters being constantlymonitored for a difference in measured value,if one reads a maximum and the other reads aminimum the differential alarm may be raisedalthough the measured mean values are correct.

The HVT-IOO also disp1ay~ an oscillating trend.Amplitude is +0.05 MJ/SM with a time period ofapproximately 20 seconds.

The cause is the necessary variations isair/fuel rate in order to maintainstoichiometric combustion.

5.5.2 In order to check compatabi 1 i ty wi th the flowcomputers, a spare flow micro-computer (869R)was linked to the HVT-IOO via the analogueoutput.

A constant std vol ume f l ovr a t e was preset inthe flow micro with CV as a measured variable.As energy and std vol ume fl ows are he1 d for a24 hour period, a daily average CV could becalculated and then compared with the HVT-I00internally generated daily average CV.Very close agreement was found which indicatesthat the oscillation and the analogue

converters do not degrade the daily energytotals.Although not really necessary for ourinstallation, we would prefer that theoscillations are reduced so that it is easierto interpret the measured CV at any instant.It would seem that the most feasible way toreduce the amplitude of oscillation is tointroduce a form of damping.

5.6 RELIABILITY5.6.1 As the sample gas supplied to the HVT is

essentially identical to both calorimeters onPhase II, a comparison may be made betweenreported daily average CVs.

5.6.2 FindingsAn example of the results is shown in Graph F.The HVT displayed daily averages are shownplotted against day number together with theCutler-Hammer derived daily average taken fromthe flow computers.The 1 arge swi ngs in average CV are due to theeffect of Piper/Tartan gas being injected intothe pipeline in different quantities each day.This oil associated gas is much richer than thenatural gas from the Frigg area.The results from both instruments, essentiallysampl ing the sa~e gas, agree well, generallywithin O.05MJ/SM and appear to follow similartrends.Following initial commissioning the HVT-IOO ranreliably for the duration of the six monthevaluation period.

5.7 MAINTENANCE REQUIREMENTNo regular maintenance of the HVT-IOO is specified bythe manufacturers. However, it is apparent thatcertain parameters require regular monitoring toprevent unexpected loss of information. Theseparameters are listed below:a) Calibration Gas Bottle Pressure

The gas pressure wi thi n the cal ibrat ion gascylinder must be checked regularly and thecylinder changed before calibration gas isexhausted as this would result in the shutdownof the HVT.

PAGE 16

PAG E 17

b) Digital Printer PaperAs a consid~rable amount of importantinformation is reported by the digital printerconnected to the HVT, the amount of reservepaper in the printer should be monitoredregularly in order to prevent the loss ofreported daily averages, alarm states andauto-calibration results.

c) Analogue OutputThe accuracy of the analogue output to the flowcomputers should be verified regularly in orderto correct any drift which may occur. This maybe performed quickly by comparing the HVTdisplayed value to the flow computer displayedvalue or by comparing the HVT daily average tothe inferred flow computer average CV.

d) On Line DiagnosticsThe on-line diagnostics routine should be runon a regular basis in order to monitor thestate of internal parameters which greatlyaffect the operation of the HVT. If any of thereported parameters exceed predetermined levelsthen further action should be taken toinvestigate the fault.

e) It is recommended that a certified test gasother than methane is sampled by the HVT-IOO ona regular basis in order to verify linearity.A binary gas such as the currently usedmethane/8% ethane mix would be suitable.

f) The most critical moving part in the HVT-IOO isthe rotary mixing valve which determines theair/fuel ratio. As wear may affect the unit'sperformance, this component should be inspectedor replaced at annual intervals.

5.8 EVALUATION TEST - SUMMARYThe results may be summarised as follows:a) Accuracy and Repeatability

The HVT-IOO has proven to be highly accurateover a wide range of calorific values and willbe able to operate accurately when sampling thegases likely to be processed at the terminalfor the foreseeable future.Both accuracy and repeatabi 1 ity were found tobe well wi th in the manufacturers spec ifi cat ionwhich offers a significant gain over theexisting calorimeters.

Accuracy is claimed fO be ±0.10MJ/SM3 over therange 33 to 41 MJ/SM .The stated accuracy oS thecalorimeters is +O.20MJ/SM .

b) Sensitivity to Environmental Conditions

present CH

The HVT-IOO displays a sensitivity toatmospheric pressure variations, but this isovercome by the automatic calibration feature.Tests have shown that a setting of 2 hourlyauto-calibration interval is effective inmaintaining accurate measurement.

c) ResponseThe response time to step changes in gas CV wasfound3to be very quick at around 20 seconds perMJ/SM. This is a magnitude faster than theexisting calorimeters and will allow transientsto be followed much closer and also allowtesting to be carried out quicker. Start-up iscompleted within 5 minutes.

d) Stability of OutputThe Cutler-Hammer calorimeters tend to displayan 0 sci 11 at ing 0 u tput un de r3 stea dy inputconditions of up to O.10MJ/SM peak to peak.The HVT-IOO also displays an oscillating trendof the same level but with a much shorter timeperiod of approximately 20 seconds.Testing revealed that this oscillation does notdegrade the output signal in operation.

e) ReliabilityReliability of the HVT-IOO has proven to beexcellent during the long term test. Followingcommissioning the unit has operated withcomplete reliability with only minorinterruptions due to external circumstances.

f) Maintenance RequirementThe Cutler-Hammer calorimeters require regularintervention to maintain accurate operation andplanned maintenance involving laboriousadjustment. Often a complete rebuild is foundto be required.In contrast,maintenanceThis workroutines.

the HVT-I00other than

is simplifiedrequiresregularby self

no plannedmonitoring.

diagnostic

PAGE 18

g) Availabilitv of Spare PartsThis is an area of major concern for theexisting CH calorimeters both in terms of longterm availability and quality.

6.0 INSTALLATION AND COMMISSIONING6.1 A report on the evaluation test was submitted to the

Fiscal authorities, a non-objection to the use of theHVT was subsequently granted. An order for threeadditional HVT units was placed with Honeywellallowing installation to commence in January 1988.Each HVT was installed alongside its operationalcalorimeter and prepared for service.With a calorimeter providing the used calorific valuesignal the HVT was connected to the flow computers andassumed the role of comparison unit. Prior toacceptance, for Fiscal metering duty, each HVT wastested for a seven day peri od to establ ish correctoperation and presence of any systematic errors.In the final installation the HVT digital data isinternally converted to an analogue output using aOAC. The 4-20 mA signal then passes via a 2 channelplotter to each of the six flow computers where thesignal is converted to 0.2 1.0 v using 50 ohmresistors. The voltage is then converted to a digitalsignal within the computer using an AOC.During the acceptance period the accuracy of analogueconversion was monitored and calorimeter trendscompared with those generated by the HVT. In additionto the 2 hourly methane autocal ibration, daily spanchecks were performed using a certified methane-ethanemixture.

6.2 ResultsAn offset of 0.02 - 0.04 MJ/sm3 was found between theHVT internally generated daily average and the flowcomputer calculated average. This offset originatesat the HVT OAC and as it is not possible to adjust theOAC to el iminate this it was necessary to adjust theHVT output scaling to compensate.Other calibration and test results were will withintolerance.One unit was found not to restart after a shutdown.The HVT wi 11 not restart unt i1 it detects a fl arne inthe main burner, the unit apparently detects a flameby monitoring capacitance changes between the igniterplug and main burner earth due to flame ionisation.The problem was rectified by repositioning the igniterplug in the main burner using spacers.

PAGE 19

The dehydrator unit was found to be defective and wasreplaced.The main fault could be reproduced by open circuiti~gthe zirconium probe output. This caused the ma inburner assembly including the sensor and igniter plugto be replaced. .The power supply board showed signs of overheating andwas returned to Honeywell for investigation.The rotary valve showed signs of wear and wasrep1 aced. A1 though Honeywell adv ised that the rotaryvalve assembly is suppl ied 'factory set', it has beenfound that the analogue output oscillations may bereduced significantly by adjusting the valve discposition on the motor drive shaft.

On recommissioning the original HVT unit, problemswere encountered in making the unit run reliably,positive and negative spikes were observed on theanalogue output at unpredictable time intervals.Faults found:

6.3 All" four units have now been acceptedoperating well. The four ContinuousCalorimeters have been decommissioned.

and areReading

7. AREAS FOR IMPROVEMENTa) The automatic calibration interval could be increased

and accuracy further improved if automatic atmosphericpressure compensation were to be performed. Thelinear effect means that the addition of a simplepressure transducer is enough.

b) The digital to analogue output converter calibrationis factory set but small offsets could be removed byincluding an adjustment pot.

c) The user manual is rather superficial and could beimproved. Two areas where more informat ion isdefinitely required is in setting up the rotary valveand igniter sensor.

PAGE 20

d) During the commissioning period Honeywell support forthe HVT-IOO in respect of fully trained personnel wasonly available in the USA where most units are inservice.

In order to maximise the performance of the HVT-IOO Honeywellshould consider action on the above points.

8. CONCLUSIONSA survey of alternative devices lead to the HoneywellHVT-IOO being selected as the most suitable replacementdevice for the aging Continuous Recording Calorimeter.An evaluation over six months under actual operatingconditions showed that the HVT-IOO met the manufacturersspecification and indeed performed better than the existingcalorimeters.The performance and reliability of the HVT-IOO could befurther improved by Honeywell. Regular maintenance detailsare required in order to avoid unexpected failures.The HVT-IOO units installed have performed with greaterreliability, better accuracy and less maintenance than thereplaced Cutler Hammer calorimeters.

9. FIGURES AND GRAPHS£..ill1) Relationship between air/fuel ratio and calorific

value.2) HVT-IOO schematic diagram3) Zirconium probe output characteristic.

Graphsa) Effect of steady barometric pressureb) Effect of falling barometric pressurec) Effect of rising barometric pressured) Effect of a barometric pressure peake) Effect of a barometric pressure troughf) Comparison between HVT and calorimeter outputs

10. AcknowledgmentI would like to express my gratitude to Trevor Davis forhis assistance in the preparation of this paper.

PAGE 21

3000

-u..Uc.n-:l......co'--'III::s

2000ru>01C

".:;to:1>I

1000

FIG 1.

4000

50.% pentaneplus50% nitrogen

50%melhaneplus50%, nitrogen

o 20 3010

" Air-lo-fuel ratio(at stoichiometric combustion)

Heanng value versus air-to-tuet ralios 'or paraffin sertes gases

Rotaryvalve

Sample gas

----- --c Controller _J-Microprocessor

Combustion air

Zirconium oxidesensor

~-,"-----~- IIIIIIIII

...J

II I 111011131IBI TI uJ/lslclFI

1 r----I

LEOdisptay

FIG 2.

_____ Analog outputto recorder

Basic operation 01 the HVT 100 Heating Value TransmiUer

1000

>E

Excesshydrocarbonregion

Stepresponse

FIG 3.

Sloichiomelric

Air-to-fuel ratio

Performance of typical zirconium oxide sensor

Fuel lean

1.020 -

-:-J......

1.010 -

IIl.

1.000

0.9510 -

DATE: SUN 19 DEC 1986 GRAPH A

HOUR BAROMETRIC CHANGE MEASURED MEASUREDPRESSURE IN Patm CALIBRATION CV ERRORPatm GAS CV(bar) % ) (MJ/sm3) (MJ/sm3)

-1 0.9851 0.985 0.000 37.75 -0.023 0.985 0.000 37.77 0.005 0.985 0.000 37.75 -0.027 0.985 0.000 37.79 0.029 0.984 -0.034 37.74 -0.03

11 0.984 -0.034 37.74 -0.0313 0.984 0.000 37.79 0.0215 0.984 0.000 37.76 -0.0117 0.984 0.034 37.79 0.0219 0.984 0.000 37.77 0.0021 0.985 0.034 37.77 0.0023 0.985 0.000 37.77 0.0025 0.985 0.034 37.78 0.01

1.040

1.030 -

o o o o o oo o oo o D

0.980~---~------~---~----~I----r-,---~-----~r---~----~I---~----~,---~8

0.20

0.15

0.10

..... 0.05~......

i 0.00

f5~ -0.05

-0.10

-0.15

-0.20

o 4 12HOURS

11 20 24

---

8c

-

-

-

I

-0.5 -0.3 -0.1 0.1 0.3 0.5PERCENT CHANGE IN Palm

HOUR BAROMETRIC CHANGE MEASURED MEASUREDPRESSURE IN Patm CALIBRATION CV ERRORPatm GAS CV(bar) % ) (MJ/sm3 ) (MJ/sm3 )

-2 1.0280 1.027 -0.099 37.76 -0.012 1.026 -0.033 37.76 -0.014 1.025 -0.099 37.70 -0.076 1.024 -0.099 37.76 -0.018 1.023 -0.099 37.73 -0.04

10 1.022 -0.099 37.74 -0.0312 1.021 -0.166 37.71 -0.0614 1.019 -0.199 37.71 -0.0616 1.017 -0.199 37.73 -0.0418 1.013 -0.333 37.66 -0.1120 1.010 -0.301 37.68 -0.0922 1.007 -0.335 37.69 -0.0824 1.005 -0.168 37.75 -0.02

1.040

1.030 -

1.020 -

1.010 -

1.000 -

O.iSlO -

GRAPH BDATE: WED 24 DEC 1986

[][] n []

[]n

[][]

[]

c[]

0.i80 +-~-......"r---",--...,.--"'---r---........,--...--- ........,---,•...--....----14 8 12 H5 20

HOURS0.20

0.1S

0.10

..... O.OSM.....

~ 0.00

ffiis -O.OS

-0.10

-O.\~

-0.20

o 24

--

-

[] [][]

[][] []- [] 0

[][]

- []

[]

-

,-o.s •-0.3 O.~-0.1 0.1 0.3

PERCENT CHANGE IN Patm

HOUR BAROMETRIC CHANGE MEASURED MEASUREDPRESSURE IN patm CALIBRATION CV ERRORPatm GAS CV(bar) ( \ ) (MJ/sm3 ) (MJjsm3)

-2 0.9870 0.988 0.069 37.80 0.032 0.990 0.171 37.78 0.014 0.992 0.205 37.79 0.026 0.994 0.205 37.82 0.058 0.996 0.204 37.80 0.03

10 0.998 0.238 37.84 0.0712 1. 002 0.339 37.81 0.0414 1. 003 0.169 37.81 0.0416 1.005 0.202 37.80 0.0318 1.007 0.135 37.81 0.0420 1.009 0.202 37.73 -0.0422 1. 009 0.067 37.80 0.0324 1. 010 0.067 37.78 0.01

1.0..0

1.020 -

'L:'1.....

1.010 - I

i 0 0

c cC

1.000 -C

C0

CC

0.900 t c

0.5180 -T I I I T J

0 .. IS 12 HS 20 24HOURS

0.20

0.15

0.10

..... 0.05tie~II::

0.00i15(; -0.05

-0.10

-0.15

-0.20-0.5

DATE: WED 31 DEC 1986 GRAPH C

1.030 -

--

0

- DC C D·0 DD

D D

- C

--

I I

-0.3 0.1 0.50.3-0.1

PERCENT CHNI.GE IN Patm

1.030 -

"

II 1.020 -

'0:'J 0 0 c 0c....,

1.010 - c D 0E c c c"6a.

1.000 -

0.20

0.115

0.10

..... 0.05~....,II:

0.00iISe -0.05

-0.10

-0.15

-0.20

DATE: SAT 6 DEC 1986 GRAPH 0

HOUR BAROMETRIC CHANGE MEASURED MEASUREDPRESSURE IN Patm CALIBRATION CV ERRORPatm GAS CV(bar) ( % ) (MJ/sm3) (MJ/sm3)

-2 1.0040 1.006 0.202 37.82 0.052 1.008 0.202 37.77 04 1.009 0.101 37.79 0.026 1.010 0.101 37.78 0.018 1.012 0.134 37.81 0.04

10 1.013 0.100 37.79 0.0212 1.013 0.067 37.75 -0.0214 1.013 -0.067 37.76 -0.0116 1.012 -0.033 37.72 -0.0518 1.011 -0.134 37.75 -0.0220 1.010 -0.067 37.73 -0.0422 1.008 -0.168 37.71 -0.0624 1.006 -0.201 37.73 -0.04

1.040

0.t90 -

0.980 +--...-.--,...---,,;---"T"'"---r---,r---.....---"""'T--r---.....----.,-- .--1o e 12

HOURS

20 2-4-4

--- D

0

D0

00 0

- 0 c0

0

-I •

-0.3 -0.1 0.1 0.3 0.15

PERCENT CHANGE IN Patm

HOUR BAROMETRIC CHANGE MEASURED MEASUREDPRESSURE IN Patm CALIBRATION CV ERRORPatm GAS CV(bar) ( % ) (MJ/sm3) (MJ/sm3)

-2 1.0060 1.005 -0.168 37.70 -0.072 1. 001 -0.404 37.67 -0.104 0.997 -0.338 37.66 -0.116 0.994 -0.306 37.68 -0.098 0.992 -0.204 37.75 -0.02

10 0.995 0.239 37.81 0.0412 0.995 0.068 37.87 0.1014 0.998 0.306 37.76 -0.0116 1.000 0.136 37.81 0.0418 1.001 0.136 37.80 0.0320 1.002 0.135 37.80 0.0322 1.004 0.169 37.81 0.0424 1.006 0.169 37.82 0.05

1.040

1.030-

1.020-

1.010-

1.000 -

DATE: FRI 5 DEC 1986 GRAPH E

f,I

cc c c c

o D

c c cD

0.~80~----'------~----'----'------~---~Ir----~----'----~J----r-,---,----~12 16 20

HOURS

0.20

0.15

0.10

....... 0.05tt....iii:

~ 0.00

f5(; -0.05

-0.10

-0.15

-0.20-0.5

o 8 244

-

- c

- Bn cc

0c

-c

0- 0 c

-

I I I ,-0.3 -0.1 0.3 0.50.1

PERCENT CHANGE IN Patm

1 5 10 15 20 25 30 35 40 45

HVT EVALUATIONDAILYAVERAGES(4 hr col.int.)

39.00

"r') 38.95E0)<,""')

::E 38.90~

>Uw 38.85o~w 38.80~

38.75

39.05

38.70

[J HVT-=100DAY+ CUTLER HAMMER

References

[1] Paper presented at the North Sea Flow Measurement Workshop, a workshoparranged by NFOGM & TUV-NEL

Note that this reference was not part of the original paper, but has been addedsubsequently to make the paper searchable in Google Scholar.