TERMINOLOGY USED IN INSTRUMENT ACCURACY

1 2002 PROCEEDINGSPAGE 271 AMERICAN SCHOOL OF GAS MEASUREMENT TECHNOLOGYTERMINOLOGY used IN INSTRUMENT ACCURACYRick WilliamsRawson & Co., Box 924288, Houston, TX 77292 The purpose of this paper is to offer a brief explanationand discussion of many key terms used in describinginstrument ACCURACY . The terms included within thisdiscussion are most commonly used for definingperformance standards with primary sensing elementstypically used in the measurement of flow, level pressureand temperature instruments . Many of the terms usedmay apply to controllers, recorders and final controlelements. However, the focus provided herein is theprimary element device. The specific devices includetransmitters (differential pressure and temperature) andflow meters ( , magnetic, vortex, turbine, variable areaand positive displacement).This paper is written for the benefit of the typical user ofinstrumentation products to include INSTRUMENT engineersand technicians.

2 A comprehensive discussion ofprecision measurement must address calibration andtraceability issues. The scope of this paper will be limitedto discussing the terms associated with the applicationof instruments rather than addressing the issues ofmaintaining ACCURACY . Calibration is a key issue indetermining the continuous performance of aninstrument and is worthy of discussion. However, it iscomplex enough to warrant a separate fundamental understanding of gas laws and the effectsof compressibility are necessary for selecting andapplying instruments used for gas measurement. This isobvious since the majority of the flow measurementdevices used in industry are volumetric, and volumechanges under actual conditions. We will begin ourdiscussion with the assumption that this baseunderstanding exists. There will be no debate regardingthe merits of direct or indirect mass measurementdevices nor a comparison the advantages of massmeasurement over volumetric.

3 The decision to use anyspecific technology for a solution must be weighed uponthe need for mass measurement vs. volume, theaccuracy requirements from the measurement, theoverall installed cost of the solution while taking intoconsideration user and industry standards and theadditional ownership costs affected by maintenance andlife expectancy of the of the concepts discussed herein will be painfullyobvious to the experienced INSTRUMENT user. The goal isto provide a logical discussion of most key terms andthereby offer a condensed reference guide for any userof instrumentation products. These terms cross theboundaries of ISA, ASME, AGA, ANSI, API and variousmanufacturers TERMSA ccuracy is a composite statement of performance thatdefines the quality of the INSTRUMENT measurement. It isthe difference between the reading of an INSTRUMENT andthe true value of what is actually being is normally expressed as plus or minus apercentage of either reading, calibrated span or the fullscale of the INSTRUMENT .

4 ACCURACY is one of the mostcritical factors to consider when applying an instrumentfor a given application. ACCURACY is also a term that ismost misunderstood by the typical user. The reason forthis misunderstanding is due to the myriad of effects oninstrument ACCURACY and the ability of manufacturers tooffer varying interpretations for the expression ofinstrument ACCURACY . In the world of marketing this iscommonly known as creative specmanship. ACCURACY STATEMENTS % Reading (% Rate) % Calibrated Span % Full ScaleAs stated in the discussion on ACCURACY , the specificationmay be expressed in terms of a percentage of reading,calibrated span or full scale. As illustrated in Figure 1,the comparison is enlightening. An INSTRUMENT accuracystated as a percent of reading maintains a constantwindow of error throughout the measurement range. Adevice with an ACCURACY expressed as a percent of spanor full scale possesses an ever-widening envelope as% InputFIGURE 2722002 PROCEEDINGSAMERICAN SCHOOL OF GAS MEASUREMENT TECHNOLOGYthe measurement reading moves farther downscale.

5 Thismakes it imperative that the user operates the instrumentas high as practical in the span in order to ACCURACY is a term that defines how theperformance of an INSTRUMENT relates to a ACCURACY includes the total effect of ACCURACY ,linearity and repeatability on an INSTRUMENT at ACCURACY is the baseline ACCURACY for manyinstruments. This specification does not include someof the effects that cause INSTRUMENT error such astemperature and static pressure effects. Referenceaccuracy is the percentage of error associated with theinstrument operating within designed constraints underreference conditions. This is the most liberal of accuracystatements and is commonly misinterpreted as abenchmark for evaluating one INSTRUMENT against further confuse the user, this ACCURACY statement maynot take into consideration errors induced by the outputmode, such as the error associated with digital to analogconversion necessary to obtain a current is an ISA term that is the combination oflinearity, repeatability, hysteresis and ACCURACY is a term used to define the overallaccuracy of a process measurement involving more thanone component used in series or parallel.

6 This is anexcellent means to determine the total error induced ona process variable that will be used for recording orcontrolling purposes. Individual INSTRUMENT errors mustbe calculated as an error of reading (not a percentageerror) for a given measurement point. The combinationof errors then is averaged using a statistical approachsuch as taking the square root of the sum of the squarederrors. For example, a flow INSTRUMENT has an error ofplus or minus one percent of calibrated full scale. Thescale is 0-100 SCFH. The reading is 50 SCFH. The erroris plus or minus 1 SCFH. The recording or controllingdevice has an error of plus or minus one half percent ofrange. The range is 0-100 SCFH. The reading is 50 error is plus or minus .5 SCFH. Using the formula:(12 + .5 2)1/2 = a system ACCURACY of plus or minus ASSOCIATED WITH ACCURACYD amping does not affect INSTRUMENT ACCURACY but doeshave an effect on the quality of the control loop.

7 Dampingis used to slow the response of the INSTRUMENT to processchanges. It is necessary to use damping when there isprocess noise and/or input fluctuations that areundesirable for control. Damping may be implementedby mechanical or electrical conditioning. Damping is ameans to average the process variable over time in orderto stabilize the output to the controller. For that reason(particularly with electronic means) the adjustment isexpressed in terms of seconds of damping. Dampingmay be applied to an INSTRUMENT with poor installationtechnique or inappropriate application of the instrumentfor the process. For this reason, it is important torecognize when applying damping that the need is : All instruments exhibit a point where somechange in process data cannot be measured. This erroris commonly expressed as a percent of is simply the ability of the INSTRUMENT to reactto small process variable input changes.

8 Deadband maybe associated with resolution because it does affect theability of an INSTRUMENT to measure continuous , resolution addresses the ability to continuouslymeasure the process variable. Deadband only addressesthe ability to monitor minor changes from a given processvariable reading. An INSTRUMENT with a deadband of plusor minus one tenth of a percent and a span of 100 psihas the ability to not read small changes of plus or minus1 psi at any reading point. If a resolution of plus or minus1 psi is necessary, a tighter span or an alternatetechnology is advised. Deadband has an effect oninstrument ACCURACY when very small changes is sometimes confused with damping. Filteringis actually the ability of an INSTRUMENT to accept or rejectan input signal. There is more than one interpretation forfiltering an INSTRUMENT signal. Filtering is commonly usedto ignore intermittent spike input signals to the instrumentthat may not be actual changes to the process example, noise may be measured by vortex flowmeters or magnetic flow meters from hard solids in aliquid flow application.

9 The solids interfere with thecontinuous measurement of the flow signal byintroducing an error normally interpreted as a flow of the signal may help to alleviate the , this type of signal spike is irrelevant to themeasurement of the process. Therefore, the ability toignore the spike is advantageous. Another common typeof filtering is the use of capacitors on a DC power supplyto minimize the effects of AC ripple : A true and accurate measurement of aprocess variable shall yield a beginning measurementand an ending measurement cycle that is equal. That is,the upward curve from zero to one hundred percent andthe downward curve from one hundred percent back tozero are identical. Any deviation from these two curvesis defined as hysteresis. Hysteresis is also linked todeadband. Since deadband affects the ability of theinstrument to react to small INSTRUMENT input signals, thiseffect amplifies the hysteretic is described in ISA Standard as thedeviation from the calibration curve of an INSTRUMENT froma straight-line relationship between zero input and 100%input.

10 Ideally this is a forty-five degree slope. Mostinstruments do not possess a linear output with respectto input under reference or actual PROCEEDINGSPAGE 273 AMERICAN SCHOOL OF GAS MEASUREMENT TECHNOLOGYD epending upon the INSTRUMENT , there are many variablesthat can affect linearity. For example, most differentialpressure transmitters use a sensor technology that isinherently nonlinear under reference conditions. Asprocess and ambient temperature and static pressurechange during actual conditions, there are effects onlinearity. Pressure transmitters vary by class. A baselinefor a pressure measurement is the common pressuretransducer. The transducer converts a pressure signalinput into a voltage signal output. A pressure transmitterextends the pressure input range and typically providesfor a current output in order to extend the transmitteddistance. There is the added value of compensation forthese transmitters to stabilize the output for the negativeeffect of temperature on the linearity of the output.