In-Sit Testing of Gas Orifice Meters By: Ron Beaty, PE David Simpson, MSME both of Amoco Prodction Co. Overview Why do we do in-sit testing Test Eqipment Design Considerations Smmary 1
Elements of a Differential Prodcer Primary Element--mostly steel; sbjected to pressre, temperatre, and flow velocity, bt no electronics Sensing Element--electronics and exotic metals; sbjected to pressre and temperatre, bt not flow; generates electronic signals or pen movements. Recording Element--Electronics or pen & ink; only sees electronic signals or pen movements Differential Prodcers are Inferential Devices Assming all conditions match reference conditions yo can se Bernolli s Eqation to infer a flow rate from a differential pressre across a known restriction Assmptions in Calclations Constant density at each snapshot Flid properties known and constant for each snapshot No friction Tbe is straight and level Gas does no work Flow profile matches power law Single phase flow Atmospheric pressre known precisely Enogh friction to dampen swirl Press and Temp known Tbe roghness in narrow range Plate condition meets specs Plate bore concentric to tbe bore 2
Measrement System Accracy Each element of a measrement system is: manfactred to close tolerances inspected by the manfactrer and prchaser installed to precise specifications After all that, what can still be wrong? What can be wrong with a new installation? Flow profile Backwards plate Beta Ratio Incorrect station parameters Non-rigid monting 3
Flow Profile Problem 25% of flow not in center 20% of pipe Error can be low by 5% or more How to prevent it Upstream piping Flow Conditioners and/or Straightening vanes How to detect it In Sit Testing Backwards Plate Problem Leading edge shaped like a ventri Reading can be as mch as 26% low How to prevent it Attention to details How to detect it In Sit testing 4
Beta Ratio Problem: Uncertainty in gas measrement is controlled to a large extent by beta ratio» Less than 0.3, ncertainty is a fnction of edge sharpness and concentricity» Greater than abot 0.62, it increases very qickly Combined error can be 7% How do yo prevent it Proper station design How do yo detect how ncertainty eqates to error In Sit Testing Incorrect station parameters Problem Any of the parameters from gas analysis to Tbe ID will affect the conversion of dp, P, and T to flow rate Combined error can be over 10% What can yo do to prevent errors Doble check original inpt (two people) What can yo do to detect errors Verify all parameters each calibration In Sit Testing 5
Non-Rigid Monting Problem The gas does work (redcing dp) p to 15% error How to prevent Independent braces Use hold-down bolts How to detect Inspection after installation In-Sit testing Installation Verification To insre that a new station meets its potential: Install all the parts at their final location Use normal gas at normal temperatre and pressre Design a test manifold that ensres that all gas goes throgh the test skid after the new station Use a certified test skid 6
When to Test As part of station commissioning process Reglarly on very large volme stations Reglarly on particlarly erosive/corrosive or dirty applications When rotine plate inspections point to a problem Test Eqipment Uncertainty Calibrated Differential Prodcer Proportional devices with provers Trbine meters Vortex-shedding meters Inferential vs. Proportional devices Calibration methods 7
Uncertainty Uncertainty means yo don t know No conclsions can be drawn from data beyond the ncertainty range, for example: If a calibration device is stated to be ±0.5% Yor tested device is off by 0.6% Actal reslts are 0.1-1.1% and the meter has a bias high Conseqently: a calibration device can only certify a calibrated device to within twice its ncertainty (i.e., a 0.25% device mst be sed to calibrate a 0.5% device) Midpoint of a confidence range mst fall within the dead-zone arond zero (i.e., the center of a 90% confidence range from a 0.5% skid mst be in the range -0.5-0.5%) Calibrated Differential Prodcer Skid As long as: The test manifold has positive isolation between inlet and otlet Sensing elements on test skid are calibrated before every test Orifice plates on test skid are inspected before and after every test Tbe on test skid is inspected qarterly Recording elements are verified against a calclation standard freqently The skid is certified against a primary standard annally to be within ±0.5% A properly designed Differential Prodcer Skid can be sed to certify a station is ±1.0% 8
Proportional Devices All have some moving part that is proportional to mass flow rate (motion can be gross like a trbine rotor or microscopic like the vibration freqency of a vortex) They are all prone to: Fol in dirty flid (20 mils of paraffin on a trbine blade cased the reading to be 2% low in one test against a primary standard) Have bearing or vibration-sensor wear that is not obvios on recording element A Proportional Skid mst be calibrated against a primary standard before and after each station test Inferential vs. Proportional Inferential Proportional Pros Pros No moving parts Well nderstood in field Very rgged Useable in dirtier flids Cons Narrow rangability Flow measrement nrelated to flid density Good rangability Flow measrement is directly related to mass flow rate Cons Moving parts prone to foling and wear Mst be recalibrated before and after each se Shold not be sed with dirty flids 9
Calibration Methods Piston Prover Sonic nozzle Primary standards not discssed in this paper: Weigh tanks High pressre Bell Prover Piston Prover S S Trbine Meter Piston sed to qickly develop k-factor for trbine meter Trbine meter is sed to prove station being tested After test, trbine can be re-proven 10
Piston Prover Design Considerations Piston rn times related to flow by: Precise displacement volme Exact dration of travel between detector switches Exact cont of piston rond trips Obstacles to se in gas No U.S. standards exist for piston provers in compressible flow (some Eropean contries have adopted procedres and ISO is evalating) Gas pressre mst increase to overcome piston inertia Fairly small volme capacity Techniqes to overcome obstacles New correlations based on eqations of state Bi-directional pistons Fast-acting, electronically operated diverter valves Acceleration/deceleration regions before/after measred volme Use Piston to prove trbine and trbine to prove installed station Sonic Nozzle Under most conditions, a compressible flid is limited to speeds less than or eqal to the speed of sond Sonic (or Critical Flow) nozzles se a sbstantial pressre drop to force the gas to the speed of sond With an open-ended pipe sonic velocity is reached when downstream pressre (psia) is less than 1/2 pstream pressre Properly designed convergent/divergent nozzles can redce this to abot 20% pressre drop With carefl measrement of pstream pressre, temperatre, and gas composition; very accrate (abot ±0.25%) mass flow measrement is possible so volmes can be accrately determined 11
Sonic Nozzle Limitations They reqire a large differential pressre, so: It can be difficlt to get the gas back into the line If hydrocarbon gas is discharged to atmosphere, care mst be taken to avoid an explosive atmosphere Liqids can drop ot of gas Temperatre drop can case hydrates to freeze They are not effective in mlti-phase flow Calclations reqire carefl evalation of: Atmospheric pressre Upstream temperatre and pressre Density Compressibility Sonic Nozzle Use in In-Sit Testing Considerations Use dry air or non-flammable gas as test flid Ensre that operating personnel are carefl, competent, and observant Verify that atmospheric pressre and gas composition are the same in EFM nit as in test-facility compter Verify that EFM nit ses the same Temp and Pressre base as the test facility Verify that EFM nit ses the same compressibility calclation as test facility With proper diligence, a sonic nozzle can be sed to certify that the differential prodcers on a test skid are accrate to ±0.5% of volmetric flow rate 12
Test Skid Design Considerations Flow conditioning Capacity and Rangeability Operating pressres and temperatres Flexibility Data Captre Transportability Why do we need Flow Conditioning? Any device sed on a test skid will be sensitive to flow profile problems Confined-space piping on test skid will contribte to poor flow profiles Swirl and asymmetry case flow-rate errors that are related to compressibility and density Flow conditioners with straightening vanes are reqired to get repeatable reslts that can be transferred from one type of gas to another 13
Changes in Compressibility 1.00 0.95 0.90 0.85 0.80 0 500 1,000 1,500 psig Air 89% C1H3, 11% CO2 Capacity and Rangeability Skid capacity shold be Large enogh to handle max expected volme Small enogh to be within design conditions for min volme Appropriate to keep sensing elements within 10-90% of calibrated range Example A differential prodcer has a range of abot 6:1 (assming arbitrary limitations on beta-ratio of 0.3-0.6 and dp of 45-105 ) A skid with one 3-inch and one 6-inch tbe (and the piping to rn them in parallel) has a range of almost 39:1 (with the same limitations) 14
Operating Pressres & Temperatres The test skid (exclding hoses) shold be designed to handle 150% of design pressre on highest design-pressre station in the operation Increase hydrostatic test pressre to compensate for the lowest ambient temperatre expected Hoses shold be: Designed for most-likely pressres and temperatres Labeled with MAOP Tested to 150% of MAOP annally (hold test for 24 hors) Flexibility The test skid shold be able to: Use one meter to evalate one stream Use some combination of skid meters in parallel for larger streams Simltaneosly evalate mltiple flow streams Re-measre a given stream mltiple times Possible reasons to re-measre Compare one meter to another to qickly evalate calibration Evalate impact of damaged plates Streams mst be positively isolated from each other Doble-Block-and-Bleed valves Spectacle blinds 15
Flexible Flow Normal Flow Parallel Flow 2-Stream Serial Flow Data Captre Types of data Prover reports Flow Parameter reports Volme reports Data shold be available on both paper and ASCII files Data shold be able to accmlate/smmarize data over flexible time frames For each time frame, captre at least: Cmlative volme for period Average flow rate dring period Instantaneos flow rate at end of period Average and snapshot temperatre, dp, and static pressre for each calc 16
Data Analysis At least 30 time periods mst be sed, any nmber greater than 30 is acceptable It is better to se cmlative gas for each time period than to se flow rates Calclate: Error = Skid Volme Station Volme 100 Skid Volme Total error shold be less than 1% Error for individal time periods shold be compted and checked for:» Standard deviation of errors» Mean error» Cont of periods above and below zero (worse than 60-40% distribtion indicates a bias when n>30) 90%Confidence = 1.65 Standard Deviation = CONFIDENCE(0.10,Std Dev,n) Sample Cont Data Analysis Example Acceptable Range Meter 1 Meter 2 Total Error -1% to 1% -0.88% -0.32% Mean Error -0.99% -0.07% Std Deviation 0.58% 0.42% 90% confidence -1% to 1% -1.10% to -0.88% -0.15% to 0.0% Cont < 0 32-48 75 44 Cont > 0 32-48 5 36 Conclsion Bias low Accrate Meas Notes: Total error and mean error vales -0.5 to 0.5% mean accrate vales no conclsions can be drawn abot where a nmber falls in that range Midpoint of 90% confidence range mst fall between -0.5 to 0.5% on passing tests Rle of thmb cont range» 30 <n<100 ==> acceptable range = 0.4n to 0.6n» n>100 ==> acceptable range =0.45n to 0.55 n 17
Transportability Considerations Weight Width (limited to abot 8-feet) Fitting all piping/eqipment onto skid Non-negotiable items Isolation between streams mst be positive (either block and bleed or spectacle blinds) Differential prodcers mst have dal-chamber fittings Flow profile-isolating conditioner mst be installed pstream of each station Transportability Options Trck Monting Larger weight-carrying capacity Considerable flexibility in skid size Leveling can be tricky Transportation for driver can be a problem after hook-p Bed height can be a problem with operations personnel Trailer monting Many weight/size compromises needed Axles and tonge (or 5th wheel) need to be matched to weight Less expensive than trck-monted Cargo Least expensive Least convenient for moving 18
Smmary In-Sit testing is both necessary and practical It ensres The measrement system works together Large stations contine to work Stations in dirty/corrosive/erosive streams contine to work Skid certification: Proportional skids shold have a bilt-in prover Differential-prodcer skids shold be calibrated periodically The skid mst be designed to be transported, leveled, and connected to the process gas allow a wide variety of flows work with expected flid pressres, temperatres, and flid qalities captre data consistent with installed eqipment 19