Capacitance and Radio Frequency (RF) Admittance

Point level control (using contact and proximity sensors) and continuous level transmission for liquids, granular solids, and liquid–liquid interface Design Pressure Routinely to 1000 PSI (7 MPa), others to 5000 PSI (35 MPa), specialized applications to 20,000 PSI (140 MPa) Design Temperature 500°F (260°C) maximum with insulated sensors; 1000°F (540°C) bare metal, sealed to 200 PSI (30 kPa); 2000°F (1100°C) bare metal at atmospheric pressure Excitation Less than 10 V @ 10 kHz to 1 MHz Wetted Materials Type 316 SS and TFE for common models, with options for CPVC, FEP, PE, PEEK, PFA, PP, PVDF, urethane, Hastelloy, Inconel, Monel, nickel, titanium Span 2 to 3 in. (50 to 75 mm) of insulating liquid to 1000 ft. (300 m) for immersion probes and 0.1 in. (2.5 mm) to 10 in. (250 mm) with proximity sensors

3.3 Capacitance and Radio Frequency (RF) Admittance LT CA To Continuous Receiver To On-off Receiver D S KAYSER (1982) B G LIPTÁK (1969, 1995) J B ROEDE (2003) LS CA Flow Sheet Symbol 430 © 2003 by Béla Lipták Service Point level control (using contact and proximity sensors) and continuous level transmission for liquids, granular solids, and liquid–liquid interface Design Pressure Routinely to 1000 PSI (7 MPa), others to 5000 PSI (35 MPa), specialized applications to 20,000 PSI (140 MPa) Design Temperature 500°F (260°C) maximum with insulated sensors; 1000°F (540°C) bare metal, sealed to 200 PSI (30 kPa); 2000°F (1100°C) bare metal at atmospheric pressure Excitation Less than 10 V @ 10 kHz to MHz Wetted Materials Type 316 SS and TFE for common models, with options for CPVC, FEP, PE, PEEK,    PFA, PP, PVDF, urethane, Hastelloy , Inconel , Monel , nickel, titanium Span to in (50 to 75 mm) of insulating liquid to 1000 ft (300 m) for immersion probes and 0.1 in (2.5 mm) to 10 in (250 mm) with proximity sensors Inaccuracy Horizontal, less than the diameter of the probe rod Vertical, less than 0.1 in (2.5 mm) for bare single points in conducting material, roughly 1% of maximum active length for all insulated probes in conducting or interface service, 3% in insulating liquids (or 0.5% with dielectric compensation), roughly 5% for granular insulating solids with constant density and composition, and to 5% for conducting granular solids Dead Band Unmeasurable with analog instruments and horizontal probes; dependent on A/D resolution with digital continuous instruments; small and application dependent on vertical single points, but optional dead band adjustment is available Temperature Coefficient Extremely variable depending on (a) probe insulation and degree of probe-to-sheath bonding in conducting materials, and (b) composition and density variation in insulating liquids Damping and Time Delay Adjustable time delay of to 30 sec is included on most single-point controls; adjustable time constants up to 30 sec are available on most analog transmitters, and digital instruments offer zero to several minutes Cost $200 to $800 for single-point controls; $500 to 1500 for two-wire level transmitters; all with type 316 SS and TFE wetted parts; increased cost with exotic metals, hermetic seals, flange mounting, longer insertion length, digital output, dielectric compensation, extended press and temp, and longer inactives Vendors (partial list) ABB Process Automation Instrumentation Div (www.abb.com/us) AMETEK Drexelbrook (www.drexelbrook.com) Arjay Engineering Ltd (www.arjayeng.com) Babbitt International Inc (www.babbittlevel.com) Bindicator (www.bindicator.com) 3.3 Capacitance and Radio Frequency (RF) Admittance 431 BinMaster (www.binmaster.com) Delavan Inc Delta Controls Corp (www.deltacnt.com) Endress+Hauser Inc (www.us.endress.com) FMC Invalco (www.fmcinvalco.com) GLI International (www.gliint.com) HiTech Technologies Inc (www.hitechtech.com) K-Tek Corp (www.ktekcorp.com) Lumenite Control Technology Inc (www.lumenite.com) Magnetrol International (www.magnetrol.com) Monitor Technologies LLC (www.monitortech.com) Monitrol Manufacturing Co (www.monitrolmfg.com) Omega Engineering Inc (www.omega.com) Penberthy (www.penberthy-online.com) Princo Instruments Inc (www.princoinstruments.com) Robertshaw Industrial Products Div., an Invensis Co (www.robertshawindustrial.com) Scientific Technology Inc (www.automationsensors.com) Systematic Controls (www.systematiccontrols.com) Vega Messtechnik AG (www.vega-g.de) INTRODUCTION Characteristic of probe-type sensors, the RF probes operate by applying a constant voltage to a metallic rod and monitoring the current that flows This current is proportional to the admittance or capacitance (if conductivity is absent) from the metallic rod to a second electrode Because the tank wall is the most convenient second electrode, most instruments monitor current to ground (which is usually connected through the probe mounting) The obvious difference between conductance and RF probes is the frequency of that constant voltage Whereas conductance types use DC or low-frequency AC, the RF items usually operate in the range of 0.1 to 1.0 MHz (although special applications can operate at 15 kHz or even lower frequencies) RF probes are connected to their associated electronic units with coaxial cable except when the electronics are integrally mounted on the head of the probe The classic shortcoming of capacitance probes is false HI level indications caused by conductive coatings that connect the above-level sensing element to ground (or the actual process level) Since 1970, solutions to this problem have existed in all but the heaviest, high-conductance process situations In the case of single-point level switches, the answer is electrogeometric In the continuous level transmitters, the approach is purely electrical Because of the solid, no-moving-parts construction, there is very little to deteriorate or fail once an RF probe is installed Compatibility with process liquids is the most obvious obstacle to a satisfactory life span This is no problem with common, well-documented reagents at temperatures below 150°F (65°C) At higher temperatures, the increased chemical activity and accelerated permeation experienced by polymers can produce unexpected results Abrasion of metals and insulators is another cause of shortened life that must be anticipated Baffles to protect the sensor from high-velocity © 2003 by Béla Lipták solids, combined with judicious location, can minimize this danger Probe failure in heavily agitated tanks can be avoided by attention to structural considerations In the case of rigid probes, the most likely cause of breakage is fatigue failure caused by eddies rapidly pushing the probe in one direction and then in the opposite A support, with an insulated bushing, near the tip of the probe greatly reduces the possibility of such a failure Flexible cable types can wrap around an agitator and fail in minutes if not adequately anchored to the tank structure If they are anchored without removing slack, they can whip back and forth, causing insulation failure and eventual breakage Intermediate supports are possible in highly agitated service using insulated bushings Beware of thick, conductive coatings that can cause substantial errors at each support point These “shortcuts” to ground can defeat the electrical coating rejection in the worst cases Correct structural design is the responsibility of the system designer, not the probe supplier Most suppliers can give rules of thumb for their probes and provide structural details of the construction It may require a consultant who is skilled in fluid dynamics and mechanics to arrive at a sure configuration in a highly agitated vessel Process instruments that depend on electrical characteristics of the process material are at a disadvantage, given that the electrical character is of little interest to most instrument users Fortunately, exact values of conductivity (g) are never required, and changes in relative dielectric constant (K) are more important than the precise value In most cases, classification as conducting or insulating goes a long way toward successful application Within the conducting category, it is sufficient to know that, except for completely deionized water, aqueous solutions will be conductive On the insulating side, it is generally sufficient to understand that most liquids will have a K of or greater The main exceptions are liquids that would be gases at room temperature (with the notable 432 Level Measurement except for ammonia, for which K > 15) and atmospheric pressure Not only these fluids tend to have K less than 2, they tend to have much higher temperature coefficients of K than insulators that are normally liquids This section is divided into single-point and continuous transmitter categories to reduce confusion The RF sensors are somewhat unique in that single-point switches have a substantial performance advantage over the continuous type in terms of accuracy, temperature capability, coating rejection, self-checking, and reliability It seems that, in many cases, the transmitters are thought to be the superior approach to control, and they obviously are necessary to obtain a proportional band In many cases, strategically placed singlepoint units are a superior route to precise and reliable process control Considering that any instrument can fail on occasion, single-point probes provide highly reliable backup, with selftest capability, and are superior to any transmitter Center Rod Connection Process Seal Inactive Length Insertion Length Weld Mounting Flange Grounded Inactive Section Metal Rod Polymer Insulation Polymer Plug FIG 3.3b An insulated, two-terminal probe with grounded inactive section TYPES OF PROBES The most basic probe configuration is a metal rod The rod is insulated from a metal mounting element that connects it to the process vessel via threads into a half coupling or a flange that mates with one on a tank nozzle The insulated junction of rod and mounting includes whatever seal is required between the process and outside world The next step in complexity includes an insulating coating on the rod, (Figure 3.3a) that isolates it chemically as well as electrically from the process Insulated probes should have the insulation securely bonded to the metal rod over the entire range of service temperature This bonding ensures that process pressure changes Center Rod Connection Process Seal Mounting Flange Metal Rod Insertion Length Polymer Insulation Polymer Plug FIG 3.3a Insulated two-terminal probe © 2003 by Béla Lipták will not compress air space and change the calibration Bonding is also important to minimize permeation, which is present to some slight degree with all polymers The addition of a tight-fitting metal tube over part of the insulation, welded to the mounting element (Figure 3.3b), will make the covered section of the probe “inactive,” because it will always see the same impedance to ground, regardless of the process material on the outside A more sophisticated way to “inactivate” a section of the probe is to use the widely known electronic guard principal A tight-fitting metal tube, insulated both from the rod and the mounting element (Figure 3.3c), with a voltage identical to that on the rod, will not only deactivate that section of the probe but will also negate the rod to mounting capacitance These are sometimes referred to as three-terminal probes, because there is now a rod connection, a ground connection, and a guard (sometimes called a shield) connection, as opposed to the two-terminal rod and ground style connections An additional variation is the probe that carries its own intrinsic ground reference This can be a larger concentric tube welded to the mounting element (Figure 3.3d), with bleed holes at the top to avoid compressing gas as the liquid rises in a closed chamber Perforated tubes, insulated and bare ground rods, as well as structural cages (Figure 3.3e) are also available for various conditions of agitation, temperature, and chemical compatibility The sensors, which use a ground wire wrapped in a helix directly on the probe insulation, are unstable, unreliable, and facilitate coating A three-terminal probe, with a plate welded on for greater capacitance (Figure 3.3f), makes an excellent proximity sensor The field from the guard can even be used to direct or focus the field from the sensing element and determine the region of sensitivity In proximity measurements, the 3.3 Capacitance and Radio Frequency (RF) Admittance Center Rod Connection Guard Connection Center Rod Connection Process Seal Process Seal Mounting Flange Polymer Insulation Guard Length Guard Element Polymer Insulation Insertion Length Metal Rod FIG 3.3c A three-terminal probe that employs the “electrical guard.” Center Rod Connection Process Seal Weld Bleed Holes Insertion Length Mounting Flange Concentric Ground Tube Metal Rod Polymer Insulation 433 Mounting Flange Welds Insertion Length Grounding Cage Metal Rod Polymer Insulation Polymer Plug FIG 3.3e A two-terminal probe with a cage, for grounding in viscous liquids Center Rod Connection Guard Connection Process Seal Mounting Flange Polymer Insulation Guard Element Polymer Insulation Proximity Plate FIG 3.3d A two-terminal probe with intrinsic, concentric ground reference FIG 3.3f A three-terminal probe with plate for proximity sensing capacitance change (proportional to sensor area) is typically very small but stable, because it is effectively looking at the air capacitance between sensor and process material Even minute capacitance changes due to temperature or stress in the mounting area could be catastrophic with an unguarded probe The distance between the probe and process material should be as short as feasible The capacitance produced is inversely proportional to this distance This means that beyond 10 in (250 mm), the capacitance change becomes extremely hard to detect The error produced by splashing and condensation buildup on the plate is fairly benign—never more than the actual coating thickness, and usually less Spans of or picofarads (pF) have been used successfully Such small ranges are the result of level change in shallow pans, with limited space for proximity plates The electronic guard principle can be employed in an infinite variety of geometries A typical example is the flat plate probe (Figure 3.3g), which mounts flush with the tank wall This is very effective in reducing abrasion, eliminating bending of low-level sensors under heavy mechanical loads, and eliminating sparks from static discharge The monitored current flows from the center plate, and the guard surrounds © 2003 by Béla Lipták 434 Level Measurement Connection Enclosure Electronics or Connection Enclosure Mounting Holes Polymer Insulator Sensing Plate Added Tee Guard Element Vent Inactive Section Gasket Surface FIG 3.3g A three-terminal flush mounting probe Probe FIG 3.3i Vent pipe serving dual function for tank access Hermetic Seal Flange Facing Probe Insulation FIG 3.3h Probe with plastic-welded process seal (Courtesy AMETEKDrexelbrook.) it, interrupting a path to ground through any coating on the face of the probe For the ultimate in seal reliability and hermeticity, plastic welding offers unique benefits By plastic welding a polymer flange facing to the probe insulation (Figure 3.3h), a highpressure seal is formed (and has been used to 5000 PSI [35 MPa] with the correct flange rating), with the normal process seal acting as a backup This type of weld is not possible using TFE, which does not melt, but most other polymers are candidates It also precludes the use of thin probe insulation The flange facing should be used as the only gasket, because it must be pinched between the flange faces Only raised face flanges provide the correct sealing, so RTJ and flat-faced flanges are not applicable impossible, because the tank is pressure coded, requires arduous inerting, or cannot be taken out of service One very simpleminded and geometrically demanding approach is to use a bent probe inserted through a side entry so as to get its active area to the desired measurement location A flange mounting is usually mandatory to avoid “screwing,” which would cause the bent probe to rotate inside the tank like an airplane propeller Its angular location, once the thread is tightened, would also be a question If a suitable nozzle is available, it must be short enough to get the bend in the probe “around the corner” before it binds If the measuring leg of the probe is relatively long, it must not hit the opposite wall of the vessel before the bend allows it to pivot Another drawback to the bent probe is the requirement that any insulation be flexible enough to tolerate the bending process without tearing or splitting This limits the selection to soft, thick insulation and therefore relatively low capacitance (60 to 100 pF/ft) Many strategies allow us to avoid the bent-probe trap by using a little bit of imagination Every tank or bin should have a vent or pressure relief pipe It is usually possible to tee that pipe so that the vent or relief goes off to the side and a probe has straight access into the process (Figure 3.3i) This does not affect the pressure code of the vessel, since it is external plumbing Other ways to avoid the “pretzel probe” include the following: • • • MOUNTING AND TANK ENTRY • One of the most frustrating barriers to good RF probe application is inadequate or misplaced tank access on existing tanks Often, adding a nozzle or even a half-coupling is © 2003 by Béla Lipták • Angle mount a straight probe from the side Install the probe from the bottom up Use a grounded inactive section to get through various impediments “Dog-leg” a pressure gauge to give the probe a straight shot Cut a hole in the building’s roof or ceiling in lieu of headroom 3.3 Capacitance and Radio Frequency (RF) Admittance 435 Isolation Valves Probe Cage FIG 3.3j Side arm (cage) mounting, when there’s no other way (Courtesy of Robertshaw Controls Co.) • • • Use “rear mount” piping to plumb through a dust collector or a side of the tank Add a stand pipe (sidearm) parallel to the tank (Figure 3.3j) Tee into a fill pipe (this needs detailed application analysis) Ingenuity will always trump bent probes for performance, cost, and efficiency ELECTRONIC UNITS Electronic units for single-point instruments are available in line-powered configurations for 24 VDC; 120 V, 50 to 60 Hz AC; and 230 V, 50 to 60 Hz AC They also can be obtained with “universal power” capability that allows them to use any of these plus 130 VDC The output from these units is generally a set of double-pole, double-throw relay contacts Some of the DC-powered instruments use an NPN or PNP output transistor to effect the switching There are also two-wire, loop-powered versions that offer intrinsic safety and automatic self-checking The signal from these instruments is a high or low current within the 4- to 20-mA range Various types of automatic calibration are available for these instruments, but they all encounter certain conditions that prevent them from being calibrated properly under every possible condition Regardless, the calibration of single-point instruments is hardly rocket science © 2003 by Béla Lipták Level transmitters are primarily loop powered, although line-powered items are available from some suppliers The traditional analog 4- to 20-mA instruments have been giving way to those with microprocessors on board Most of the digital instruments are capable of communication using one of the HART, Honeywell, or fieldbus protocols This allows interrogation and modification of the instrument by means of a digital communicator One of the most popular features of these instruments is their ability to calibrate on any two points in the range It is possible to enter the correct output (in milliamps or level units) at an existing low level and enter a high point days later The instrument locks the input/output curve into those two points The signal from the instrument is either analog 4- to 20-mA or digital The digital mode allows more than one transmitter to use a single loop The limited power available at the transmitter, combined with the multitude of calculations being executed, causes digital instruments to be considerably slower than the analog ones Response times of to sec are possible, whereas analog instruments are capable of responses within 100 to 300 ms For small tanks with high fill and drain rates, the digital instruments might not be an option A hybrid instrument is the “multipoint” control It is essentially an analog instrument with internal, adjustable pick-offs and multiple relay outputs It is useful for sump control where several pumps might be involved, but the absence of an analog output makes calibration lengthy Each pick-off must be adjusted with the level at the desired point on the probe This type of instrument is specified for HI, HI-HI, 436 Level Measurement and HI-HI-HI points at times This doesn’t make much sense, because one failure kills the whole operation probes for switch points close to the bottom and top of a tank rather than a vertical or in (50 or 75 mm) Conducting Process Materials SINGLE-POINT SWITCHES The typical capacitance switch employs a vertical or horizontal probe, either polymer insulated or bare metal, projecting some distance into the process vessel with the center rod insulated from the mounting (Figure 3.3k) With any gas or vacuum (K = 1.0) on the probe, there will be a minute RF current flowing from the center rod to the metallic tank wall When a liquid or granular solid covers the center rod, the current from there to ground will increase This change in current is detected by the electronic unit, which switches the output When the process material drops below the probe, the RF current decreases and, hopefully, returns to its previous low-level state If the process material is thick and conductive, any coating remaining on the insulator between probe and mounting will maintain a higher RF current level and can cause the switch to continue signaling a high level The answer to false high-level indications is based on the classic electronic guard principle By interposing a third metallic element (shield) with an identical RF voltage between the center rod and the mounting (Figure 3.3c), no current can flow from center rod to mounting The shield element supplies whatever RF current the coating demands, but this current is not included in the measurement “Length is strength” in capacitance probes Longer active lengths translate to more substantial capacitance changes in insulating processes, and longer interelement insulators allow the instrument to reject higher conductance coatings This means using horizontal Center Rod Connection Process Seal Mounting Flange Polymer Insulation Insertion Length © 2003 by Béla Lipták Insulating Process Materials The insulating materials (oils, solvents, resins) are subtler in their effect on the switching circuit A horizontal probe produces a sharp capacitance change over the thickness of the center rod Any portion that is in a nozzle should be inactive, either with the guard or a grounded inactive With a vertical probe, capacitance increases gradually as more of the active length is covered The switching point is adjustable even if the probe is bare metal Good practice requires that the probe be at least in (50 mm) into the process material at the desired switching point Attempting to get switching at the tip of the probe will lead to unreliable performance, because the slightest change in probe mounting or electronics can cause a constant high-level signal Dielectric constant (K) is relative to the absolute dielectric of a vacuum (K of gases barely differs) This means that the capacitance (and the proportional RF current) in air will be doubled in gasoline (K = 2) It will be multiplied by 20 in ethanol (K = 20) It will only be raised by 60% in liquid carbon dioxide (K = 1.6) Insulating coatings are a very minor problem Just avoid bridging to a grounded part Plastic, Concrete, or Fiberglass Tanks and Lined Metal Metal Rod FIG 3.3k Bare metal, two-terminal probe Conducting materials (aqueous, metals, and most forms of carbon) carry the ground potential of the tank walls right to the probe If bare metal, the probe will signal “high level” the instant it contacts the process If it is insulated, the switching point will depend on the capacitance setting of the instrument With a vertical, insulated probe, it is possible to adjust the level at which the switching takes place by varying the capacitance setting of the instrument A horizontal probe of either type allows no level adjustment other than by relocating the mounting Coating is a serious consideration in all conducting materials, and the guard-type construction is usually required With bare metallic probes mounted vertically, the plain capacitance probe will as long as the process never reaches the mounting area and no crystallization or heavy condensation occurs When using the guarded probes, the guard should be long enough to project well into the vessel, beyond nozzles and potential wall buildup The absence of metallic contact for a ground reference is seldom a problem using single-point RF probes in conducting process media The probe will indicate high level whenever it touches the process Unlike metallic tanks (or those with metal pipes, pumps, or grounding rods), it must be tuned to a value less than the capacitance-to-ground value of the tank Most tanks have at least 10 pF to ground, which is a perfectly adequate level for a reliable measurement Probes with a 3.3 Capacitance and Radio Frequency (RF) Admittance guard element should not be mounted horizontally in ungrounded tanks with conductive contents The guard can drive the entire process at the same voltage as the center rod, hence there is no current flow and no high-level signal In the case of an insulating medium in a fiberglass tank, even metal pipes offer little help to the miniscule capacitances The only sure answer is a vertically mounted probe with its own metal ground rods or concentric tube This is also a good precaution against RFI from walkie-talkies and other sources, which can cause false high-level signals Metal tanks that are rubber or plastic lined and steelreinforced concrete vessels, on the other hand, represent excellent RF grounds The capacitance from process to metal structure is very high as compared with the capacitance produced by an insulating medium This makes it look like a short circuit to the RF current It is excellent for conducting media, too, but it still requires the avoidance of horizontal probes with driven guards Underground concrete or fiberglass tanks (except in desert conditions) present a similar ground configuration Concrete structures of cinder block with no vertical steel reinforcing bars are completely ungrounded and perform exactly the same as fiberglass tanks Interface Electrical sensing is the premier method of detecting the interface between an insulating and a conducting process medium The typical margin between the conductivity of organic and aqueous phases is greater than 1000:1 The measurement is completely independent of temperature and density variation Probes may be mounted vertically or horizontally Vertical probes should be inactive down to about in above the desired interface control point This can be accomplished by use of the electronic guard, a grounded inactive, or a short probe mounted from the rear (rear mount) on the end of a suitable pipe The instrument will indicate high level as soon as conductive material contacts the tip of the bare probe In the case of heavy oil separators, it may be desirable to use a probe with sharp edges machined into the tip (Figure 3.3l) This will ensure good contact between water and steel in spite of oil coating An interface detector with bare metal rod should be tuned to the maximum capacitance level of which it is capable Horizontal probes should be bare metal with relatively long insulators between probe and mounting (ground) Typical proportions, to maximize the margin between insulating and conducting phases, would be 12 in (300 mm) overall length with a 10-in (250-mm) long insulator The use of sharp edges machined on the tip of the probe is also advantageous in this orientation This is one situation in which the electrical guard is of no advantage (coatings are usually insulating rather than conducting) and can actually be a drawback if the guard drives the entire conducting phase at the same potential as the probe This phenomenon is usually a product of insufficient grounding, but it has been observed in metal tanks © 2003 by Béla Lipták 437 Metal Rod Sharp Edges FIG 3.3l Oil shedding tip for interface with viscous organic phase GRANULAR SOLIDS The best approach for establishing the high level in large silos is usually via a flexible cable probe with a long, thin weight at the bottom It can be mounted close to the fill point (Figure 3.3m) and negate most of the angle of repose that side-mounted sensors encounter The flexible aspect allows it to swing out of the way when struck by incoming solids The actual switching should take place on the weight that will be least vulnerable to abrasion In insulating materials, at least in of weight should be covered at the desired switching point Conducting materials, of course, will switch as soon as they touch the tip of the probe Incoming material will not cause false high-level indication as long as it is in free fall This is because there is a very small air capacitor between each grain and all the others All this series air capacitance means that any deviation from normal air capacitance is a negligible quantity If the incoming material is compressed (especially conducting materials), it IS possible that it could cause a trip, so the probe should be located accordingly Rigid probes, located near the fill, are suitable for smaller bins and lighter duty Fill Top of Silo Electronics or Connection Enclosure Half Coupling Cable Insertion Length Process Granular Weight FIG 3.3m High-level probe for heavy granulars in large silos 438 Level Measurement Horizontal probes mounted in the bin wall should be of the “guarded” variety to avoid false high-level alarms— especially for low-level service The primary concern is bending or abrasion, which becomes more acute as the depth and density of the material increases Short, fat probes have much better lifespan, and it is possible to use a flat-plate probe (Figure 3.3g) mounted flush with the wall for optimal life CONTINUOUS TRANSMITTERS Probes are most commonly mounted vertically from the top of the tank Angle mounting through the side of a tank is possible, as is mounting up from the bottom When tank penetrations are at a premium, it is often possible to “tee” into a vent or drain line for probe mounting There are even cases in which probes have been successfully “teed” into fill nozzles, but this requires considerable application analysis to predict the effect of incoming material The insertion length of the probe should not extend beyond the desired measuring range by more than 5% An active probe that is not producing signal can still produce errors Conducting Liquids The main concept required to understand this class of applications is called saturation Assume that an insulated probe is immersed in deionized water with minimal conductivity (Figure 3.3n) The addition of drops of hydrochloric acid to the vessel will gradually increase the conductivity The output will also rise as more RF current flows from probe to ground Eventually, the conductivity will be high enough that the resistance (R) from probe to tank wall is negligible compared to the capacitive impedance of the probe insulation (C5) Further increases in conductivity will make no observable difference in the RF current and, therefore, the output (i.e., the RF current has reached its saturation point) This is the concept that makes the RF transmitter an instrument rather than a lab curiosity A probe that is not saturated will have a calibration that is a function of two variables (conductivity and level), just as a d/p transmitter calibration is a function of density and level It is possible to adjust the threshold of saturation by adjusting insulation thickness (hence capacitance) and excitation frequency Raising the impedance of the insulation lowers the threshold, and vice versa Saturation thresholds from 0.1 to 1000 µS/cm can be accomplished A reasonable question might be, “Why not just use the 0.1 µS/cm threshold at all times?” The answer is “coating rejection.” The higher impedance that correlates with a low saturation threshold is less able to ignore high-conductivity coatings In fact, a large mismatch may allow the coating to saturate the probe so that the output reflects the actual level plus coating length In general, the path to best conducting coating rejection lies in the highest probe capacitance and excitation frequency that will allow saturation by the process material A1 A2 Ce B C1 Ce C1 C4 C2 C5 C3 Ka C4 C2 L R = Any Value C5 I C3 Ce = C1 + Kp C2 C4 C2 + C4 + C3 C5 C3 + C5 R FIG 3.3n Electrical representation of an insulated probe in conductive liquid (Courtesy of The Foxboro Co.) © 2003 by Béla Lipták 3.3 Capacitance and Radio Frequency (RF) Admittance being measured A special case for coating rejection is one in which the process liquid is conductive but can dry on the probe, forming an impervious, insulating coating An example is latex paint The liquid is quite conductive, but the dried paint adds to the insulation thickness of the probe, decreasing its capacitance and changing the calibration The solution is to use a thick (low-capacitance) probe insulation so that thin additions will have negligible effect on calibration By using the highest possible frequency, conductive coating rejection will also be maximized, and the resulting accuracy will be the best possible compromise Complex impedance (or its inverse, admittance) measurement allows the electronic unit to measure two variables: the capacitive component of RF current and the resistive component of RF current The actual liquid level that saturates the probe produces a pure capacitive current The conductive coating, because it is relatively thin, has a much higher resistance than the bulk liquid, so it will produce both capacitive and resistive phase current By subtracting the resistive component from the capacitive, the effect of conductive coating on the output will be decreased In fact, once the coating is longer than a nominal value, the two RF current phases equate, and the effect of coating is precisely zero Maximum coating error occurs when the coating length is relatively short and its resistance therefore is low The maximum error is a fixed, predictable number of inches for a given probe capacitance, excitation frequency, coating thickness, and conductivity This means that percent inaccuracy due to coating will be greater on short ranges than on long ones In other words, length is strength Insulating Liquids Probes for measuring insulating liquids should generally include their own parallel ground reference to guarantee a uniform distance to ground through the process liquid Linearity, sensitivity, and immunity to RFI are enhanced by a concentric ground tube (Figure 3.3d) or other construction In some cases, it is efficacious to use a metallic tank wall, baffle, or ladder to furnish the required parallel ground reference This is frequently the case in pharmaceutical or beverage applications where a concentric tube interferes with any clean-in-place function It is also common in tall tanks that require flexible cable sensors and with slurries, which can accumulate solids and plug the ground reference Changes in dielectric constant will cause a change in calibration If composition is constant, temperature will be the only concern for variation of K Because K is proportional to the number of molecules between probe and ground, output (as with a d/p transmitter) will be proportional to density and, hence, the weight of process material For cases in which constant composition and a reasonably narrow temperature range are not possible, a transmitter with dielectric compensation is available By using such an instrument, it is possible to obtain a level signal that is independent of dielectric, and © 2003 by Béla Lipták 439 Concentric Shield (Ground Reference) Level Sensor Insertion Length (I.L.) Measured Level Reference (Composition) Sensor 5.5" FIG 3.3o A probe with a second element for dielectric compensation (Courtesy of AMETEK-Drexelbrook.) even conductivity, when tanks are not dedicated This technology demands two conditions: A short inactive section at the bottom of the probe, approximately in (150 mm) Homogeneity of the process liquid (no stratification) The compensation is accomplished by making two independent measurements The first measurement is made with a short sensor (composition probe) below the tip of the level probe (Figure 3.3o) This segment is assumed always to be covered by the process liquid Its capacitance is proportional to the electrical character of the process material The second measurement uses the level probe Its output is proportional to the level times the electrical character of the process fluid By dividing the output of the level probe by the output of the composition probe, the transmitter output becomes independent of the electrical properties Please note that conductive coatings can cause large inaccuracies in these particular transmitters Even so, conducting liquids that not coat will be measured with accuracy equal to the insulating ones For the sake of utility, both probe segments are usually provided, coaxially, in the same assembly It is also possible to use two completely separate probes, with the composition probe mounted horizontally below the tip of the level probe Inaccuracy can be limited to 0.5% over a wide range of electrical values and a wide temperature range Continuous Liquid–Liquid Interface Two elements are key to successful interface level measurement: immunity to total level variation and maximum margin between the signal contribution of the organic and aqueous 440 Level Measurement phases Making the probe inactive down to a point that always will be covered by liquid can negate variation in total level Of course, packed vessels and separators on overflow not exhibit the problem The primary item to minimize the signal from organics is to maximize the distance to ground Low K is another positive item The only practical way it can be varied is by better separation to exclude the high K aqueous content Concentric ground references, small-diameter stilling wells, and external standpipes, by bringing the ground closer, make the measurement more sensitive to changes in the organic phase The ideal mounting is in the center of the tank, with no metal nearby Maximizing the aqueous contribution means using a probe insulation with the highest feasible capacitance A good margin would be one in which the aqueous contributes 300 pF/ft and the organic pF/ft Regardless, many very satisfactory separators in the heavy oil fields operate with an 80to 30-pF/ft margin The limitations on maximum probe capacitance include fragility (thinner coatings have higher capacitance), temperature (TFE and PFA have the lowest K of common polymers and highest temperature capability), and chemical compatibility (again, the low-K insulators are most inert chemically) The question of an intermediate emulsion or rag layer adds a bit of confusion to the measurement Users often desire a measurement at the bottom of the rag, but that area is generally very close to the aqueous in conductivity The point sensed by the continuous probe is always that at which the emulsion reverses from organic continuous (aqueous drops in organic matrix) to aqueous continuous (organic drops surrounded by aqueous liquid) In liquids with distinctly different colors, users often believe that the instrument is not functioning correctly The visual interface typically does not coincide with the electrical interface except in the rare case of perfect separation This means that sight glasses are useless for calibration of interface transmitters Sample taps are more valuable, but only if the electrical conductivity of the sample, rather than color and viscosity, is observed Users often lose sight of the fact that the instruments cannot separate the components The first criterion for accurate interface control is good separation This may entail more or better emulsion-breaking chemicals, longer residence time (less throughput), higher temperature, or additional mechanical or electrical aid to coalescence The idea that certain instruments can produce lower organic in the aqueous stream, less water in the organic stream, and raise throughput is total baloney In many cases, separators are run above the design throughput, so something needs to be done to accelerate separation if product quality is to be maintained For best results, the probe should be located as far from the inlet as possible, where the best separation will have occurred The most frequent use of interface transmitters is in horizontal separators, where the total level is usually maintained by overflow, and the interface is maintained by controlling the aqueous dump valve A whole other class of application has been largely ignored because of the difficulty in getting © 2003 by Béla Lipták the probe to the interface area That application, sometimes called water bottoms, measures the water level beneath organic products (usually fuels) in large storage tanks Typically, this involves measuring a few inches of water at the bottom of 40-ft (15-m) or taller tanks Dropping a short probe, attached by a cable or pipe, down from the top of the tank is pretty straightforward unless the tank has a floating roof The most successful floating-roof applications have used a 45° hot-tapped nozzle near the bottom of the tank Because the tank is seldom empty, it generally requires a probe to be inserted through a sliding seal under pressure Granular Solids The most common problem in granular measurements is caused by moisture variation in insulating materials, which represent the majority of granular solids Since water has a nominal K = 80, and the typical solids are in the K = area, small variations in water content can produce large percentage changes in K – 1.0 and, hence, the slope of the input/output curve If the particular granular produces a similar change in conductivity, subtracting the resistive phase of RF current from the capacitive component can compensate moisture variation over a particular range of water content The same instrument that provides electronic coating rejection, for example, compensates grains of wheat, corn, and rice These whole grains produce capacitive and conductive RF current components that are affected equally by moisture variation in the to 20% water range By subtracting the resistive component from the capacitive, the effect of these changes is eliminated At the opposite end of the moisture spectrum, very dry (600°F), it is difficult to fabricate, impossible to plastic weld, and exhibits a high degree of microporosity Can be destroyed by butadiene and styrene monomer TECHNOLOGY The purpose of this section is to help the more technically inclined to understand just what is happening in a particular application It is not necessary to personally make good use of this information Your friendly vendor will be glad to take the necessary steps to ensure a good result with the instruments you purchase Analysis of RF probe circuits requires little more than a knowledge of AC circuits The simplest situation is the bare probe mounted in the center of a vertical, cylindrical tank (Figure 3.3q) The expression for C2 and C3 is the formula for a concentric capacitor and can also be used to calculate the capacitance of probe insulation (To so, use K = for TFE, PFA, FEP, PE, and PP Use K = for PVDF.) The length units are inches, and the diameters can be any units as long as both are the same A good approximation of C1 is 20 pF For rough estimates of air capacitance in tanks that are not cylindrical and/or not concentric, 0.35 pF/in is a good rule of thumb If the probe is within ft of the wall, it will be higher A probe in the center of the tank is least sensitive to position Capacitance increases as the distance to ground is reduced In the center of the tank, motion toward one wall increases the distance to the opposite wall, so the capacitance is nearly constant Close to the wall, the capacitance increases rapidly with any reduction in the distance The capacitance produced by any insulating liquid will be its K times the air capacitance The insulated probe complicates the calculation because of the nonlinear combination of series capacitors (Figure 3.3n) The equation shown is correct for insulating liquids For conducting liquids (assuming saturation), the third term is reduced to C5 , because the low resistance represents a short circuit to ground In the case of a liquid–liquid interface, the organic phase would substitute for the air space shown The K of the organic phase must be measured based on an upper phase sample at operating temperature The amount of aqueous phase held in the organic will be a function of temperature as well as throughput Table 3.3p is a list of nominal dielectric constants, which can be used to make rough estimates of what capacitance will result from a particular probe geometry 444 Level Measurement A B Probe Ce Insulator C1 Ce Vessel C2 C1 C3 Ka R∼∞ C2 L Ce = C1 + C2 + C3 = I Kp C3 = C1 + 0.614 Ka (L-I) log10 A/B + 0.614 Kp I log10 A/B R∼∞ FIG 3.3q Electrical representation of a bare probe in an insulating liquid (Courtesy of The Foxboro Co.) CONCLUSION Early capacitance probes had the advantages of simplicity, relatively low cost, corrosion resistance, and a lack of moving parts In addition, the proximity design required no contact with the process fluid On the other hand, these early capacitance probe designs were subject to errors resulting from changes in the dielectric constant of insulating process fluids, to errors resulting from the tank geometry and fiberglass construction, and to conductive coating buildup on the probes In the newer designs, these problems have been largely solved by increasing the operating frequency, incorporating a phase detector component in the electronic circuits, and modifying the design of the sensors and their guards These improvements have made the capacitance/RF admittance type probes a powerful means of level detection The only types of measurements that are still excluded with these instruments are interfaces between two conductive liquids or the liquid–solid interface A major maintenance problem is water in the head assembly that results from incorrect conduit routing and venting The head may be potted with two-component RTV to eliminate this problem Bibliography Andreiev, N., Survey and guide to liquid and solid level sensing, Control Eng., May 1973 © 2003 by Béla Lipták API Recommended Practice 550, Manual on Installation of Refinery Instruments and Control Systems, Part I, Process Instrumentation and Control, Section 2, Level, American Petroleum Institute, Washington, DC Belsterling, C A., A look at level measurement methods, Instrum Control Syst., April 1981 Capacitance method for liquid-depth measurement, Electron Power, December 1967 Dinkel, J A., Universal capacitance probe liquid level measuring system, Rev Sci Instrum., November 1966 Duncan, J and Dutton, W., Capacitance probe confirms presence of liquid NH3 when unloading, CIM Bull., January 1978 Hall, J., Measuring interface levels, Instrum Control Syst., October 1981 Herbster, E J and Roth, J H., How to gage by capacitance, ISA J., June 1965 Lawford, V N., How to select liquid-level instruments, Chemical Eng., October 15, 1973 Lerner, J., Continuous level measurement: an introduction to 16 basic types, Control, November 1990 Mital, P K., Capacitor sensor monitors stored liquid level, Electronics, October 30, 1967 Morris, H M., Level instrumentation from soup to nuts, Control Eng., March 1978 Preshkov, V P., Capacitance liquid helium level indicator, Cryogenics, April 1969 Proximity sensors, capacitive, Meas Control, February 1991 Ritz, G., Choosing the right solids level sensor, Control, January 1994 Schonfeld, S., Capacitance gaging checks spacecraft fuel level, Hydraulics and Pneumatics, April 1967 Schuler, E., A Practical Guide to RF Level Controls, Drexelbrook Engineering Co., Horsham, PA, 1989 Tavis, R., How to mount a capacitance level-sensing probe in your vessel, Powder and Bulk Solids, 19, April 2000 Weiss, W I., Capacitance level control, ISA J., November 1966