HAPTER VI FURNACE DRAFT CONTROLS Host large utility boflers used to be designed with pressurized furnaces. The justification for such designs was the lower installed equiptent cost (one set of forced draft fans) and reduced fan maintenance since the FD fans handle clean air. However, problens with maintaining easing pressure integrity, and concern for personne) safety have sade the fndustry return to use of atanced draft units, Concurrently, clean air standards requiring use of scrubbers, ete., intro- duce higher draft losses and the need for higher-head induced draft fans. nite the problem of regulating the fans to supply the required airflow and.at the sone time, maintain the desired furnace draft during normal operation is elenentary, special control requirerents ‘are associated with Tiniting transient pressure excursions during major upsets such as main fuel ‘trips. A discussion of the factors imolved will be based on studies of a 600 IW off-fired unit reported in reference (6.1). Although this untt has unusual requirenents, the example serves to itlustrate the nature of contro? and override logic involved in furnace draft controls [As shown in Figure 6.1, conbustion air 4s supplied to the boiler by two constant speed rotor driven forces draft fans, with inlet vanes control. There are tho sets of Induced Draft fan trains: ‘each train consisting of a prinary 1.0. fan discharging to a secondery 1.0. fan inlet. The secondary 1.D. fans discharge through a scrubber systen to the stack. The primary I.D. fans are constant speeé, rotor driven, with control on fnlet vanes. The secondary 1.0. fans are speed controfled through hy- draulic couplings. Because of the head Toss requirenents created by high draft Toss boiler design, particulate scrubbers and future sulphur renoval scrubber, these Induced Draft fans are considerably aversized for a unit of this capacity. Each of the fans are driven by 7000 HP motors and each set ‘ean develop a total head of 76 inches WC Conbustion air flow neasurenents are made by piezoneter rings located in the inlet of the two forced draft fans. Furnace draft taps are Just below the furnace roof, with each of three transmitters connected to its own dedicated tap. This otl-fired high-draft toss boiler design has crested severe problens for the electric operating conpanes, Of initial concern vas the Targe induced draft fans, with their high inertia ané e long coast-dowm tine, continuing to pull a significant draft after the forced draft fans had stopped ro- tating following the tripping of all fansBOILER 377 100" Fe ‘SCRUBBERS AIR HEATERS | 6 \ C) i FORCED DRAFT FANS PRineary. SECONDARY INDUCED DRAFT FANS INDUCED DRAFT FANS FIGURE 6.1 STACKon further consideration, the bofler implosion potential fron sudden Toss of fire appeared even ore criticals especially because of the use of sertes sets of two (2) induced draft fans discharging to four (4) flue gas scrubbers The most critical elenent in miniatzing an inplosion potentfaT was found to be that of tine. Studies disclosed that saxisun negative pressure oceurred approximately 1.7 seconds after Toss of Mane and loss of Flane occurs at an estimated one second after loss of fuel. This allows approximately 2.7 seconds to take corrective control action to counter the anticipated high negative pressure unless ‘the time could be extended. PROCESS CHARACTERISTICS ‘The dynamic behavior of the boiler air-gas path was simulated ty a digital computer. The sinu- Tation study had to objectives: First, to indicate the pressure excursion profiles that could be expected following a master fuel trip disturbance. Second, to indicate the relative merits of alternative F. 0. and I, 0. automatic contro? arrangements. ‘The sieulation model accounts for the following effects: = gas flow head loss according to the square Taw, P = Cpl. = variation of gas tenperature through the boiler due to conbustion heat release and to convective and radiative heat transfer. = variation of gas specific volume according to the gas law, P = BRT = variations in the mass of gos residing in furnace sections = variations in fan characteristics with gas density, fan speed, and damper position. Before discussing model resvits, it 1s useful to review the fundamental requtrenents of furnace air flow control.AIR FLOW CONTROL REQUIREMENTS For the furnace and fan configuration shown schenatically in Figure 6.1, air and gas flow through the furnace and ducting 1s maintained by the combined action of forced draft and induced Graft fans. Opposition to this flow is contributed by air heaters, tube sections, furnace dampers, uct losses, and scrubbers, and ts distributed along the entire length of the flow path. It fs useful to visualize the pressure distribution along the boiler air gas path as shown in Figure 6.25. This shows positive pressure upstream of the furnace, negative pressure downstrean of the furnace and then positive pressure again to force the flue gases through the scrubber to the stuck The scrubber is an adjustable orffice device designed to hold a constant head Toss in steady ‘operation, but the head losses through the reseinder of the flow path vary as the square of air/gas Flow. Thus, if Figure 6.28 applies for full load, the pressure proffte at part load should be as shown in Figure 6.25. Note that, to maintain balanced pressure in the furnace, the net head developed by both F.D. and 1.0. fans must be reduced. Now consider the case where the F.0. and 1.D. fan controls are inperfectly manipulated to reduce ‘ir flow to a part Toad value. The pressure profile will be as shown in Figure 6.2¢; the slope of the pressure profile and the total head loss will be essentially unchanged from Figure 6.20, but the whole profile will be biased up er down, Examination of Figure 6.2 shows that: = The net head of F.0, and 1.0. fans must be adjusted in unison to regulate air flow while maintaining furnace pressure at a given value. = The net head of the F.0. and I.0, fans mist be adjusted én oppasition to regulate furnace draft while holding a given air flow. In accordance with these characteristics, reduction of unit load under normal operating conditions involves sinsTtaneous gradual reduction of the net head of both F.D, and 1.0. fans. This fs achteved by closing inlet vanes on the F.D, fans, closing inlet dampers on the prisary 1.0. fans, and reducing ‘the speed of the secondary 1.0. fans.It fs to be noted now that the above contro} rules apply for normal operation where the system is undergoing relatively gradual changes. The contro} requirenents following sudden loss of fire are significantly different, The nature of these different requirenents is best iTlustrated by reference to simulations of fuel trip disturbances as presented in the next section. PRIMARY SCRUBBER FD. FAN FURNACE 1D. FAN EXIT SECONDARY 1D. FAN Ain ncaten, BACKPASS, AIR FURNACE DANPERS HEATER, DUCTING SCRUBBER FD, ner — HEAD 1D. ner HEAD aS FIGURE 6.22 Full Load, Afr Flow BI = > FIGURE 6.2 Reduced Load and Air Flow, Properly Coordinated Fan Adjustnent Funwace PRESSURE ERROR FIGURE 6.2¢ Reduced Air Flow, Inproperly Coordinated Fan Adjustment,‘SUDDEN DISTURBANCE BENAVTOR Figure 6.3 shows a simulation of the bofler's response to» sudden conplete shut-off of fuel flow in the condition where a1 air flow and furnace draft controls are locked. There is an imediate and rapid decTine in the temperature of the gas within the furnace. The pressure in the furnace 4s governed by the ges law: Pv = ni where voluse, v, and the gas constant, R, are constant. The rate of change of the furnace mass inventory is dependent on the air and gas floxs which are governed by fan and head Toss characteristics. The mass inventory adjustment rate does not immediately match the rate of decay of tenperature and there 4s, following from the gas Tax, a rapid fall in pressure inside the furnace. Figure 6.3 shows an increase in furnace air inflow as furnace pressure and temperature fal. ‘There 4s a corresponding decrease in gas outflow. Goth of these flow changes aggravate the distur bance since they create decreased net F.0. head and increased net 1.0. head, hence depressing the already low furnace pressure. The furnace inventory adjusteent "catches up" with the temperature change process after about 2 seconds ond the furnace pressure then commences its return to remind] value. ‘This behavior shows thet initial movenent of the T.D. fan controls in the closing direction wil? ultinately be needed to regulate air flow at the destred post-disturbance velue, and will tend to counteract the transient dip {n furnace pressure. Nore importantly it shows that, aTthough they will ultimately be required to close in correspondence with the 1.0. fan controls to maintain the new air flow, the F.D. fan controls should be moved in the opposite (opening) direction during the initial transient following @ sudden firing rate reduction. This requirenent for different directions of notion of the F.. fan controls during sudden Firing rate transients, as opposed to gradual load changes, is a key factor in selecting between the two commonly used control arrangaments described below.somnwnc. 2000°R 2000 4/SEC, FURNACE PRESS. FURNACE TEMP, ‘AIR FLOW 49 FURWACE AIR MASS FLOW (=/SEC.) FURNACE TEMPERATURE ("R) FURNACE PRESSURE 20) FUEL FLOW F (W/SEC) 4 to aaa 1 L L 20 Time (SEC) 70000 Rapid fuel shut-off, fan dampers locked, FIGURE 6.3AIR FLOW CONTROL ALTERNATIVES Two feedback control arrangenents that have comonly been used to meet the control objectives are shown in Figure 6.4. Goth achieve the desired coordination of fan contro} actions by assigning fone set of fans to regulate air flow and the other to regulate furnace draft. The difference is ‘that they have reversed responsibilities for the F.0. and 1.0. fan contro} loops. Simulation runs covering several different botlers have shown that both arrangenents ean give satisfactory regulation of air flow and furnace draft when following load ramps as rapid as 100% per minute. ‘The behavior of the two contro? arrangenents following sudden disturbances of the Toad reference te quite different, however. Scheme A, which makes the F.0. fan control sensitive to Toad reference, WwiN) tend to close the F.0. fan inlet vanes imediately following a sudden downward disturbance of load reference and fuel flow, hence aggravating the negative furnace pressure excursion. In contrast, Schene 8 wiTl respond to the sudden reduction of load reference by closing the 1.0. fan dampers, but the assectated negative furnace pressure caused by the fuel flow reduction will cause it to initially open the F.D. fan inlet vanes. It will close the F.0. vanes only after the negative furnace pressure ‘excursion has passed. ‘Thus, Schene 6 4s regarded as being preferable to Schene A because it inherently gives the correct initial response to potentially danaging fuel trip disturbances while Schene & would have to depend fon override logic to prevent inappropriate closure of the F.0. fan vanes. ‘The other factor affecting the selection of Scheme & is that the F.D. fen control, usually has ‘the more direct effect on furnace pressure since there is less buffering volune and head loss in the ductwork between the F.0. fan and the furnace than between the furnace and the 1.0. fans. Now, furnace pressure control is the highest priority requirenent; it should be assigned to 2 fast-acting ‘nner contro} oop, with the air flow controt befng handled by an outer control loop. Since stable contro? design requires the’ inner Toop to be faster acting than the cuter loop, it is desirable to assign the more direct contro) means, the F.D. fan vanes in this case, to furnace pressure control. The recomended contra} arrengenent incorporates the essential features of Schene 8 for the regulation of furnace draft and afr flow, together with high gain overrides to drive both F.D. and 1,D. fan controls in the compensating directions indicated ty Figure 6.2 whenever the deviation of furnace draft from setpoint exceeds preset emergency Timits-1- ‘ORIGINAL covrROL_ s¥STEH ‘The origine? control system configuration as it was installed, before any specific requirenents {for coping with furnace imptosion were imposed on it, is shown on Figures 6.5 and 6.6. The original air flow and furnace draft controls on this unit were inplesented with pneumatic. control hardware throughout. The forced draft inlet dampers are each powered by one 10" x 16" pneu- atic piston drive. The to primary I.D. fan inlet vanes required, for each fan, two 10" x 16" piston drives, set up in 8 naster/slave configuration. Operation of each of the secondary 1.0: ‘fans hydraulic coupling control Tevers was accomplished with one 8" x 16" piston drive to each lever. ‘The original system assigned the duty of maintaining air flow to its required set point to all three sets of fans in parallel (the forced draft H/A stations could be considered air flow master Stations). The furnace draft controller had plus/minus tris adjustment on the prinary I.D. fan fons were made inlet danpers available to {t to maintain balance between fans. No specific provi in this syster to cope with the implosion possibilities inherent with a main fuel trip. Kith no F.D. fan running, an automatically set maxicun 1.0, inlet vane position equivelent to about 30: air flow is established. Initial checks on the original pneunatic draft control system disclosed that there was © tine lapse of eighteen (18) seconds from the tine the furnace pressure transmitter felt a change unti? the prinary induced draft fans inlet vanes, used for furnace draft control, began to mave. The vane drives required thirty (30) seconds to go full trevel Booster relays were installed in the inlet vanes érives and the tine for full travel, without oad, was reduced to approximately 5-6 seconds. Installation of electronic transnitters and electric transmission to the pneumatic control systen and fron the contro? systen to the drives reduced the overal? loop time to 3.5 seconds. To gain more tine, sequential elevation tripping of fuel was utilized as recomended in Reference 6.2.“13+ REVISED CONTROL SYSTEM Based on computer model studies, it nas found advisable to éo most of the control using prinary Induced draft fans and to use the secondary 1.0, fans to supply the additional head to push the exhaust gases through the serubber to the stack. The studies led to the control revisions described below: a) Primary 1.0. Fane Control (Figure 6.7) ‘The air flow denand and the afr flow seasurenent, with C, correction, are developed just as fn the origina? system to index the afr flow controller, However, in this revision, the fair Flow denand 4s a1s0 fed-forward as @ denand for total primary I.D. fans inlet danper position. This reduces the duty of the integral component of the air flow controller to a calibrating trim adjustzent, since the load watching duty in positioning the vanes {s carried by the feed-forward denand, Utilizing average vane position demand (fran below the H/A statfons) af a negative feedback against this cenand, a much faster and more accurately tined systen Hinearization for one-tyo fan operation is obtained ‘This same averaged primary 1.0. vanes signal acts as the feed-forward denand to the F.0. fans control. Separately, each prisary I.D. fen position denend indexes {ts corresponding secondary 1.0. fan speed contro}. Thus, air flow control 4s applied, in parallel to all three pairs of fans. Back-up safety features, and special action taken soecificaTly to help contein furnace pressure within bounds on a H.F.T., cone into the systen below the primary 1.D. fans W/A stations, so they will be effective even if the fan loops are on manual) If furnace pressure increases to 2 value higher than the normal contro} range (set at #3 fan increasing signal from the furnace pressure dezdband controller (discussed later in the F.D, fan contra! sub-Toop) wiTl take over and increase the opening of the vanes at the high sfgnal auctioneers. On an out-of-renge Tow furnace pressure (set at -{") 2 runback tokeover at the Tow signal aueStoneers will close the vanes. Entering the sumers below the auctioneers, is 2 transfent negative signal thot fs triggered by the sane logic that initiates 2 M.F.T. This negative stgnal, promptly reduces the primary 1.0, fan vanes position by 35%, ané then wipes itself out over @ 30 second perfod.” a “Be Secondary I.. Fans Control (Figure 6.8) Since the prirary and secondary 1.0. fans form physically, and operationally, sertes pairs, ‘the contro} system is set up on that basis. The "A" primary 1.0, fan vanes positfoning signa} becomes the Feedformard, through a function generator, to the "AY secondary fan hydraulic coupling controt. The pair of "8" fans are treated in the seme way. Empirically determined from operation®| ¢ata, a load referenced set point for the duct pressure between the two fans {5 compared with the actual duct pressure to provide a balancing trim to keep the two fant working together properly. This duct pressure corrective action also provides the destvable Feature of a self-tining derivative type action to the high trertia, ow responding secondary fan speed loop. "Qutrof-range” furnace pressure overrides are apolied to the secondary 1.9. fans also. Forced Draft Fan Control (Figure 6.9) ‘The averaged prinary 1.0, fan vane positioning signal acts as the feedforward to the F.0. ‘fan vane position control. Calibration of this feedforvard based on the steady state relationship between 1.0. and F.D. vanes 1s provided. Because of the importance of the furngce pressure messurenent, it was decided to use three transnitters feeding into a median selecting circuit, In thts way, the oss of any one ‘eraramftter can be accepted si thout loss of automatic control. This verified furnace pressure stone} from the selecting circuit 1s used to index the furnace pressure controller, Because lof the Toad related feecforuard fron the primary 1.D. fan vane positions the furnace draft controller has only a trim duty to perform under normal conditions. In the deadbané controtler, furnace pressure 4s conpared against both high and Tow pressure set points. Should @ sufficfant upset occur to drive furnace pressure to one or the other of these Tintts, the appropriate override action will be taken at the Tow and high signal selectors below the F.0. fans H/A statfon, These sare signals, inverted, are used tn the prinary end secondary 1.0. fan cfreuits, as previously discussed. ‘Taig ‘qutcof-range’ furnace draft controller 4s of prime importance as a safety backup to ‘the normal control. For this reeson 1 cones into the system below the Hand/Automat ic stations. Introducing this backup control loop below the H/A statfons, however, forces an levatustion of faiTure modes that could occur. The two extremes would be foilures that create 100% positive or 100% necative outputs from the sackup" controller. Either case would be disastrous, since the 1.D, end F.0. contrels would be driven inversely to extrene positionsFeo} fi a al eee FOXES] po] = "a > > Ficure 6.8@ yi) To protect against such a catastrophic faflure situation, the desdbended furnace draft. controllers are triplicated and the three controTler outputs are auctioneered in a median signa) selecting circuit. In this manner a fefTure in any one control Toop, or furnace pressure transuitter, can be accepted without system upset, Monftors alarm to the operator any such falure, and alert. hin to the fact that one of the channels of control is faulty. Operational Characteristics In terms of the overs] contro? phitorcphy, the major change made in this revision ts to assign furnace pressure supervisory contro) to the F.D. vanes, instead of to the prinery 1.0. fan vanes as originally designed. Air flow contro? is stil? assigned in paratel, to all three sets of fans. The operational characteristics of this revision under normal operating conditions are apparent from the systen's description Just dis- cussed. We can postulate a H.F.T. and follow, oualitatively, the contro! action: of the system Two immediate actions are triggered from, and coincident with, the M.F.T. contact: 1, The master decand (and thus the air flow demand) is locked up at tts existing value 2, The transient negative signal fs inserted in the primary I.D. fan vanes positioning signal, closing these vanes, at their fastest rate, about 35% ‘to offset the anticipated furnace negative pressure In spite of this kicker action to the primary 1.0. fans, the furnace pressure wil? start going strongly negative az the furnace tenperature collapses. This starts the forced draft danpers openina by normal furnace pressure control action. ‘The combination of lever furnace pressure and further open F.D. fan vanes increases the ‘air flow as measured on the suction side of the F.D. fans The incresse in air flow starts the air flow controler driving 217 fans downward. ‘The feedforverd action to the F.D. fan loops in the reducing direction starts countering the opening action of the furnace draft controller, and continues as the reduction in the furnace draft requires progressively Tess FD. fan.19. linen measured afr flow has been reduced to match the "locked-in" demand, at? 1.0. fans wilT be operating at a lower level and the forced draft fan danpers will be back at essentially the sane position they started fron at the time M.F.T. was triggered. (CONTROL RESPONSE SIMULATIONS 1k sinulation study was made with controtTer models and looicel statenents added to the basic Furnace model to represent the proposed contro? system in detail. The controls were tuned by making an init series of small disturbance runs and adjusting the controller proportions) end inteoral eins for good darping and response in normal operation. Simulations of major bofler trip distur- ances were then made by suddenly stepping the Toad reference to zero and reducing fuel flow accord- ‘ng to the desired shut-off sequence. ne representative set of runs showed the effect of controls on furnace pressure transfents. ANV-runs shown here used the inftiel operating condition where the flue gas scrubbers are fully ‘opened to give minimum head Joss; both fan trains are operating, the secondary 1.0. fan motors are {in operation, and the primary 1.0. fan motors are de-energtzed. A sinulation of 2 full load trip fron this initia! operating condition where al fuel flow ts shut off in two seconds with aT1 controls locked showed @ maximus furnace draft excursion of 20 finches of water, One typical series of simulation runs was made to show the extent to which contro? action could reduce this excursion, Figures 6.102 and b show the response to this disturbance when the F. = (draft) control fs active in its norma] mode and the 1.0. control is in manual at fixed damper position. The negative furnace pressure excursion of 18 inches w.g. {5 slightly Tess than thet observed with no contro? faction and the initial negative pressure excursion 1s followed by a significant positive excursion, A Vinitation of the baste contro! schene as used to yield the results in Fig, 6.10 fs that it relies on the regulating controller whose gain settings are optimized for normal overation, and oes not provide the strongest possible contra] action in energency situations. Gain values that vould give maxinun rate control action in bofter trip transients would result in poorly damped response to normal small disturbances.s2IN.we, 000"R FURNACE PRESS, FURNACE TEMP. -20- Sono iste. FURNACE PRESSURE, IN.W.G. 4 AIR MASS FLOW, =/SEC. —. 4 FURNACE TEMPERATURE oR ‘AIR FLOW 1000 go 0 Tine (SEC) 75.000 Rapid fuel shut-off F.D. vanes-auto LD. vanes-locked FIGURE 6.108100% VANE OPENING -a- 1D. VANES: 0 Time (SEC) Rapid fuel shut-off F.D. vanes-auto .D. vanes-locked FIGURE 6.10 25000This Tisitation of the normal reguTating contro? is overcome by the use of high gein overrides whch becone active when the furnace pressure deviates from set point by more than 4 inches w.9. The effect of these overriding controls 1s evisent fron Figure 11 which shows the response to the save transient as in Figure 10 but with the high gain overrides set to drive the F.D. fan vanes strongly in the opening direction and 1.0. fan dampers strongly in the closing direction when the furnace pressure falls below -4 inches v.g. The overriding controls give a stgntficant reduction of the initial transient, and are able to return the furnace pressure rapidly to the -4 inch w.g- value. As the rate of change of furnace temperature decreases the normal regulating Toop 1s able ‘to regain control and return the furnace draft smoothly to its set point value. Figure 6.11a and b shoved that the use of fast acting override contro} action provides 3 ‘strong reduction of the furnace draft excursion, but that the first very rapid fa1l in furnace temperature following complete fuel shutoff in two seconds cannot be compensated by contro? action because of the stroking speed limitation of the danver and vane actuators. Tf the prinary 1.0. fans had been in operation, giving a greater maxinun available 1.0. head, the initia? negetive furnace pressure excursion shown in Figure 11 would have been signfficently erester. ‘The most effective method of limiting the first dowmard swing of furnace pressure has been found to be the use of programed reduction of fuel flow to reduce the initia? rate of fal! of Furnace tenperature. Figures 6.120 and b shows the response to a Master Fuel Trip when both F.D. and 1.0. controls are in autonatic mode and the energency overrides are activated. In this case the "hydranotor fuel valves on the four Tevels of tio. 6 of1 guns were closed in succession at 0, 2, 4, and 6 seconds after receipt of the MT signal. The staged fuel shutoff fs shown to reduce the rate of f011 of furnace tenperature to 2 value where the overriding controls can hold the initial furnace pressure excursion very close to the -# inch w.s. Tevel.a8 ne 0 2mm 3000" R FURNACE PRESS. FURNACE TEMP, =e 3o00 #/sec, FURNACE PRESSURE, "HO AIR Mass FLOW #ISEC AIR FLOW, FURNACE TEMPERATURE “R 3000 1 pop . . a Tite ECT Rapid fuel shut-off over-ride controls active FIGURE 6.118 5000100% VANE POSITION. =m FD. VANES (% OPEN) UD. DAMPER (% OPEN) po Tie GEC Rapid fuel shut-off over-ride controls active FIGURE 6.118 2500005 - samwe, 3o00°R 3000 #/sec. FURNACE PRESSURE "#20 FURNACE PRESS, FURNAGE TEMP. AIR FLOW FURNACE TEMPERATURE “a f as | 16 1000 [asmesel ES#s HOSE Sembee HEEEOE | HOEEOS RESIS EEEET 00 Tite (SEC) 700,000, Staged fuel shut-off full automatic fan controls FIGURE 6.128100% VANE POSITION ~~ FD. VANES (% OPEN) 1D. VANES (% OPEN} po Tie (SEC) 700.000 Staged fuel shut-off full automatic fan controls FIGURE 6.12-- WPLOSTON/EXELOSTON It might be argued that the contro} system, advocated here could increase the possibitity of boiler explosions. The rationale for this argunent seens to be grounded in N.F.P.A. standards for prevention of furnace explosions; specifically in the statenents contained in paragraph $63 of standard 850 quoting directly: 562, Recomended Procedure For Purging After an Energency Trip: “If the fans are operating after the trip, continue in service. Do not imediately increase the sir flow by deliberate manual o autonatic control action. If the afr flow 4s above 25% of full Toad air flow, it may gradually be decreased to this value for 4 post firing purge of at least five minutes. If the atr flow is below 25% of full Toad volumetric air flow at the tine of trip, it shall be continued at the existing rate for five minutes and then oradustiy increased £0 25% of full load air flow and held fat this value for a post Firing purge of five minutes." ‘This paragraph has been widely interpreted and implenented as a requirement to lack the forced draft dampers in the position existing at the tise of trip, and allow the induced draft danpers. (or speed) to go toward shut on furnace pressure control. Some interpretations have called for Jocking the 1.0, fan actuation also. To us, this seens completely at odds with the problem to be solved, The overall intent of this paragraph, as stated in its heading and several tines in the body of the text, is to establish a procedure for post firing purge. Earlier,in this same NFPA publication, purge’ is defined in Chapter 3 by: “A flow fair through the furnace, boiler gas passages and assocfated fives and ducts wnich will effectively remove any gaseous conbustibles and replace with air. ‘and on page 33 by: “Purge air flow shal] be equal to or greater than 25 percent of full oad volumetric air Flow for a period of 5 minutes= 26 - Studying these direct quotations raises the question: Just what “air Flow" does paragraph $63 recommend not be increased? In the control system, when we refer to "air flow", we mean the air delivered by the forced draft fans to the burner windbox. After a fuel trip, "purge air flow" is rot equsl to this "forced draft air flow” until furnace draft 1s very nearly restored to its set value. Referring to Reference 2 we find reported on page 10 the situation after a fuel trip, with a11 danpers Tocked. The plots of furnace pressure and T.D, suction pressure following the trip show ‘that from about 2 seconds to 12 seconds after the trip, furnace pressure was reduced to very nearly equal 1.0. fan suction pressure. In other words, for about 10 critical seconds there was no purge air flow, The “forced raft afr flow" entering the furnace during this time was all used to restore volunetric inventory Tost when the fuel was tripped off. An additional piece of inforsation {1lus~ ‘trated by this sane set of response curves is that, even with alll fan danpers Tocked, the "forced Graft air flow’ increases as the furnace pressure plunges to its mininun value. This 1s recognized {in paragraph 563 by the phrase "deliberate automatic or manual control actfon." While the suggested control system tenporarily increases “forced draft air flow" on a controTled basis following @ fuel trip, a control philosophy that Tocks up the forced draft damper differs fron this only in degree, since both approaches result in a tenporary increase in “forced draft air flow" while Tost inventory 4s being restored. Only when the volumetric unbalance 1s restored can “forced draft air flow" be considered as "purge air flow. The only controt action that would resuit tn no change in F.D. air Flow would be a closure of the FD dampers. This appears to be a sure invitation for a boiler implosion. Another reason sonetines stated for not increasing air flow is that en increase in air flow wiTl pick up conbustible material that has settled out in low-veloctty areas, and that this can create an explosive mixture. Our proposed solution incresses, transiently, only "forced draft afr flow’ in an effort to restore air flow through the furnace boiler gas passes, fTues and ducts (the defined "purge air flow"). There 1s no sudden increase in flow rate through those parts of the botler where conbus~ tible eateria? could collects the increase occurs only in the clean air side of the botler. In facts Af we do not take the steps we do to keep the furnace pressure from plunging to minus 15 to 17 inches wes, we can visualize more combustibles being stirred up by the inrush of air through observation doors and any other points of leakage, These paths are not the normal flow path, and there would be more tendency for accumulation pickup by these random air flow paths. In summary, the suggested control configuration and override logic will reduce the probability lof implosion following M.F.T., without Increasing the possibility of « furnace explosion. The actions ‘taken positively account for the tenporary inbatance created by the rapid inventory upset in the=. furnace, giving eptinun short-term protection. Contr? action quickly returns the forced draft dampers to the position they were in prior to the trip, thus restoring and assuring the continuous post-trip surge air flow as required concuustons 1A careful review of furnace control and performance by both sinuletion and test has shown that, Jorge furnace draft excursions can be contained by: a) Tined fuel shutoff over a period of 6 to 8 seconds following initiation of bofler trip. b) fast-acting actuators for a11 fan control vanes, c) Assigning the furnace draft contro! duty to the fans (FD or 10) having the pronpter impact on furnace drafts in most boilers this wiTl be the forced éraft fans. 4) Direct acting override controls responding to out-of-range furnace pressure, applied to both F.D. and 1.0. fans, combined with 2 transient closing bias of 1.D. damper position initiated by master fuel trip. ‘The contro} syste, properly designed and tuned to handle master fuel trip situations, e150 provided improved contro} under norm] operation REFERENCES, 6.1 Volleer, H.D., UndriTl, d.Mh., Crim, H.6., dr-, "Draft Controt System Design, Simulation and Test for a High Inploston Potential Softer", Aaerican Power Conference, Chicago, IL.» Apri, 1977. 16.2. "Recommendations for Furnace Enclosure Damage Prevention” - Combustion Engineering Co. position paper 6.3 Euehner, P.C., UndriT1, d.M., "Furnace InpTosions, An Analysis and Proposed Protective systens", American Power Conference, Caicago, IL., April. 1975.