Document NRzyegB0n7DdpXx50qJm5BL8

relatively easy process to control since no time logs are present. A more diffi cult process requires careful adjustment of the instrument to attain stability. Thermal lop. Assume the left-hand tank represents heat inflow or steam in put to an air-heating coil. Water flow from the tank now represents heat flow ing through the coil wall. By imposing -a'restriction to water flow from the first lank we can simulate resistance to heat flow through the pipe wall, (thermal lag). To provide this restriction, a small hole is drilled in gate of valve be tween the two tanka. With valve closed, water flow is restricted to the small opening. Right-hand tank represent* temperature levs) of surrounding air, and it is this level we wish to control. Closing the gate valve introduces time lag In ayetem, and left-hand tank lavel begins to rise because of restricted outflow. We now visualise continued inflow of steam to the coll (first tank), hut because of thermal resistance C Two opposed bellows connecting to nozzle arm gives both proportioning end reset v action If restricting valve to bellows Y Is adjusted to motch the process nmb (small hole in valve), heat is delivered slowly to surrounding air (second tank). With continued addition of steam, temperature level of aJr (water in second lank) rites enough to causa tho controller to close the supply valve. Level in left-hand tank is atill abnorm ally high and because of the time lag, it recedes slowly. Excess water in this tank corresponds to heal stored in the walls of a heating coll when supplied with too much steam. As air temperature groduolly drops (level in second tank), the controller again comes into action to open the valve. Cycle then repeats because the valve tends to overshoot desired flow rate necessary for stable operation. The cyclic condition may continue indefi nitely and at beat requires considerable time to dampen out We can conclude that the throttling range of the instru ment must be increased to stabilize the system. - Effect* on Offset. Increasing the throttling range also increases amount of offset from the control point experi enced with a load change, so we need automatic load-change compensation. Before designing such automatic com pensation, let's Investigate operation of a proportional controller having 100% throttling range when the load changes. Moking necessary adjustments, we set the pelcocka for S0% load. Output pressure settles out ot 9 psi end the pro cess reaches stable condition- With 9-pal output pressure, lever T, Fig. 4, is in its midposilion since movement of bellows apd spring meehanltm is pro- portionoi over a range of 3- to 15-piS output pressure. Remember when the pen is at midpoint, the baffle is at its midpoint, and output is at midpoint of its range. Consequently, lever T is at its midpoint of travel, and thus the nozzle is correctly positioned In relation to baffle. With a change in pen position, all the foregoing parts move In proportion to magnitude of change. II we change the process load to 75%, the -float drops, pan moves up scale, and as the controller mechanism gradu ally reaches a new point ol balance, cutput pressure reaches 12 pal. Control valve now passes a flow equivalent to 75% load and the process is again in balance. Examination shows that lever T has moved to the left and now oceupies a position corresponding to 75% of its total travel. Seeking some way to return the pro cess to its original control point and still maintain balance at 75% load, we first analyze the condition of the pro cess. It is evident that tank level is be low the desired control point, so the valve must be opened farther to admit more woter. To do this we mast in crease the output air pressure. By disturbing the nozzle-baffle clear ance the amplifying action of the relay produces the desired increase. Com pressing bellows X to cause such a dis turbance returns lever T to its original midposilion. Output air pressure im mediately increases, controlled valve opens, and tank level rises to its origi nal position. Pen has now relumed to the desired control point ff repositioning of lever T nftdfd the application of on external fans la the bellows and spring unit In a ditto lion to further oppose bellows sctlas, and the load change experienced erased an increase in output from 9 to 12 an auxiliary bellows, so placed to oppose bellows X, would supply the needed as ternal force if the 3-psl change wen utilized as the restoring power. Let's investigate replacing the spring with a bellows. Referring to Fig. 5, bellows Y t> places the spring and opposes bellow X. 'If Y is to furnish an opposing fotce. equal to the spring a given amount of air must be trapped within iu At a starting point, let's charge the bellow with normal output pressure of 9 pd required by the controller to ostnlais stable control at 50% load and ekss valve R. Alter adjusting link S to compeoaats for a slight variation between^ spring ond inflated bellows, the controller rw produces the 100% proportional action obtained with the spriag-opp**" bellows. This action can be cheeked by moving pen. Fig. 3, across ch`rt . observing the proportional change output pressure. IHminoHag Offset- Allowing process to reach stable control at S0% load, we now increase the load to 75% As with the previous controller, as tW pen moves to 75, output pressure p* ally reaches 12 psi and lever T aan* to the 75% position. We now want w retum lever T to its midposilion * thereby upset the nozzle-baSe relau" 90 (696) POWER Novs'ohw 1,4 the cootroller is greatly accelerated. glowing Follow-*? Aerie*. Evidently, nozzle follow-up cannot be entirely eliminated il proportional response la lo be retained. But iT noale movement is momentarily interrupted as the baffle assumes o new position, output from the controller momentarily changes at a rapid rote. If follow-up action could then be restored, the instrument would again perform as a proportional con troller. Recalling that follow-up la produced by air-pressure impulse delivered to bellows X, desired momentary Interrup tion is achieved by installing a restrict ing needle valve P In the air lines feed ing bellows X, Fig. S. Controller now operates thus: A alight pen disturbance causes baffle to change position In relation to nozzle, and thus generate e rapid change in out 6NosiIe-boffio system mounted on W Nozzle position, determined by lover vglve positions tho stem accurately I T, controls output pressure. Table 1 put pressure. Since needle valve P restricts air flow, proportional action of bellows X Is greatly retarded, and this holding back allows output pressure to hip to rebalance the system. By open ing valve R, Fig. 5, we con admit out put air at 12 psi into bellows Y and force lever T to its midposilion. At this psiol both bellows Y and X are charged with 12-psi pressure and teach a posi tion of equilibrium just as (bey were when equally charged with 9-pai pres sure. In both cases, lever T la et its midposilion. Table 11 shows this stepvise met action. nozzle accordingly. As air slowly bleeds into X, lever T Is grsdnslly forced to its original position preceding the losd shift. fn demonstrating (his action, let's ad just valve R to admit air gradually to bellows Y, Fig. 5. In practice, stability of the process governs this bleed rote; otherwise, o cyclic condition may be set up. Imposing s 50% losd we observe that increase or decrease rapidly. Such acceleration provides the increased rate of controller response we require. The term rate response is generally applied lo such controller action. As air continues to bleed through needle valve P, the proportioning bellows eventually causes nozzle to assume its correct position In relation to bafflie and thus preserve proportional action. Let's make use of this response by The process has rebalanced during the pen levels out and holds at the 50 partly dosing needle valve P and re this procedure end the pen has returned mark. Suppose we change the load to peating the. load change previously to the control point within the color 15%. Pen swings about 20 chart units shown with proportional plus reset re baad. As we observe the process lo right of control band,, stops, and as sponse. Load on the process is set ot action, controller appears to behave ths reset action continues to function, 50%, throttling range remain* el like an on-off unit Process has fallen slowly return* to the control point. 100%,. while reset rate (a Increased Into a cyclic state and pen continues to Controller has thus compensated for a slightly by opening needle. valve R (urge first above and then below the de load change, ff the load fa now re farther. These three adjustments cor sired control point (color band). turned to 50% the pen swings about 20 respond to the three adjusting knobs on Reason for the on-off behavior ia obvi units to left of control band, then a conventional Instrument and are set ous as we wotch opposing bellow*. Y slowly returns to original position. in accordance with process'stability. work, Fig. 5. With valve R fully open, While the controller now compensates Load-Change Action. If we increase both bellows receive 'an equal Impulse for load changes, its recovery speed is the load to 75% the pen swings but four boot the output line. So no motion la too slow. Increasing the reset rate may or five chart units away from the control Imparted to lever T to cause proportion- help at times but at others a higher bond and quickly returns. Decreasing jpi or follow.up action of (he noale. reset rate may give unstable control. losd to 50% shows the same action in 'fees it remains siatfonary, the baffla'a Wbst we need is an auxiliary device to the opposite direction. Deviation from slightest movement causes a rapid Increase recovery speed without impair the control bond Is reduced from 20 thsnge in output pressure- This action ing reset or proportional aetion. units to about five, and recovery time " Wcatical to an on-off controller's- Reviewing the action of the on-off is reduced considerably. To achieve automatic reset, yet pro- controller we find that any alight mo While this test shows an accelerated **** `lability, we have to thtottle valve tion transmitted to the baffle by the pen recovery following an inatantnneods *80 *'r bleeds slowly into bellows Y couses a rapid change of output pres load change from 50 to 75%, the major wnever a load change occurs. Thus, sure. Because of lack of follow-up ity of processes do not receive such on ' 8 proportioning action of bellows X octJon by the nozzle, e minute movement abrupt upaei in eelual practice. It la *** l5* retained because any load of (he bafflie caused an Instantaneous more likely that tbe load change will < ?*e c#n|ln8 a baffle movement lm- change In nozzle backpressure. If this occur ol o far slower rate. Rote re uaiely produces a change in output action can be Incorporated in our pro sponse will assist greatly in stabilizing Pressure, sod bellows X repetition* tho portional system, response rate of the.control in such coses since it causes Novwnbar 1941 (697) 91