Prceedings f the 17th Biennial Waste Prcessing Cnference ASME 1996 REDUCTION OF CARBON MONOXIDE EMISSIONS WITH REGENERATIVE THERMAL OXIDIZERS Sctt M. Finnin. Sheldn Lipke, P.E. Passaic Valley Sewerage Cmmissiners Allen Baturay, P.E. Carlsn Assciates Abstract Regenerative thennal xidizers (RTOs) have been extensively used fr the cntrl f vlatile rganic cmpund (VOC) emissins frm varius surces. Hwever, very little infnnatin is available n the ability f RTOs t cntrl carbn mnxide (CO) emissins. This paper presents the results f extensive tests cnducted n tw RTOs t detennine their VOC and CO cntrl efficiencies. treated sludge are drus and cntain high cncentratins f VOCs, CO2 and CO. The riginal dr cntrl system included a carbn adsrptin system which did nt effectively cntrl the Zimpr drs. In 199 1 PVSC purchased and installed tw identical RTOs t cntrl drs and VOC emissins frm the Zimpr prcess and the filter press building. One RTO perates at all times with the secnd unit acting as a stand-by. The inlet gas stream t the RTOs includes VOC and CO cncentratins as high as 2,000 ppm and 3,600 ppm, respectfully. The testing demnstrated that bth RTOs were capable f cntrlling greater than 98% f bth inlet VOCs and CO. While the destructin efficiencies within the cmbustin chambers exceeded 99.9%, direct leakage past valves accunted fr the lwer cntrl efficiencies. The tests indicated that the verall VOC and CO cntrl efficiencies f the RTOs may be limited by valve leakage. The design and pennitting f a RTO shuld include cnservative cntrl estimates which accunt fr pssible valve leakage. Intrductin Each RTO cnsists f three canisters cntaining ceramic media which captures thennal energy frm the exhaust stream during an utlet cycle and transfers this energy t the inlet gas stream during an inlet cycle. The flw is cntrlled by butterfly valves which are pened and clsed by linkage arms cnnected t a drive shaft which cmpletely rtates nce every 3-4 minutes. A cmmn cmbustin chamber is lcated abve the canisters. Tw natural gas fired burners, lcated at ppsite ends f the cmbustin chamber, maintain a temperature f I,500 F. Prir t an utlet valve pening (an utlet cycle) a purge valve is pened t evacuate any ful gas within the canister and valve area t prevent excessive levels f cntaminants frm being discharged. A simplified diagram f the system and the valve psitins during each cycle are shwn in Figures (I) and (2). The Passaic Valley Sewerage Cmmissiners (PVSC) wn and perate a 330 millin galln per day secndary wastewater treatment facility lcated in Newark, NJ. The area serviced by PVSC is highly industrial, with a ppulatin f apprximately 1.5 millin peple. The wastewater treatment prcess includes primary treatment and a pure xygen activated sludge prcess fllwed by secndary clarificatin. Sludge generated in the treatment f the wastewater is thickened in gravity thickeners prir t treatment thrugh the Zimpr Wet Air Oxidatin Prcess. The Zimpr treated sludge is then thickened in decant tanks and dewatered in recessed plate filter presses. The current sludge prductin is clse t 250 wet tns per day. The Zimpr prcess utilizes high temperature (420 F) and pressure (650 psi) with the additin f air t xidize apprximately 65 percent f the vlatile matter in the liquid sludge. Bund water within the sludge is released thrugh the Zimpr prcess, which results in the imprved cagulatin f the slids. The vaprs released during the decanting f the The inlet gas stream cnsists f apprximately 30,000 cfm f highly drus ff-gas frm the Zimpr prcess with a VOC cncentratin as high as 2,000 ppm and an xygen cncentratin between 15% and 9%. The inlet gas stream als has a high cncentratin f CO ranging frm 1,800 ppm t 3,600 ppm. The cncentratin f VOCs, CO, and the percent xygen entering the RTOs varies with the amunt f sludge prcessed thrugh the Zimpr facility. Bth the VOC and CO cncentratins increase with the amunt f sludge prcessed by the Zimpr units, while the percent xygen decreases as mre sludge is prcessed by the Zimpr units. The riginal pennit t perate required a VOC destructin efficiency f 98%. The permit als allwed a maximum CO utlet cncentratin f 50 ppm r 100 ppm crrected t 7% O2 when the xygen cncentratin was belw 14%. This permit cnditin is typical f a cmmn cmbustin surce such as a biler. and indicates efficient cmbustin f the prcess ff-gas and any auxiliary fuel within the cmbustin chamber. The 437
Figure (1) - Simplified RTO Diagram TEST loca lions BURNER COMBUSTION CHAMBER : a 0 r::; BURNER STACK -0. from ZIMPRO DECANT TANK PURGE BLOWER Figure (2) - Valve Psitin During Each Cycle INDUCED DRAFT FAN CANIS R CANIS R CANISTER - IN OUT HEATING PURGE COOLING IN OUT PURGE PU RGING 438
penn it als included mass emissin rates fr these cntaminants based upn these cncentratins and design flw rates. This penn it cnditin did nt recgnize the RTOs as a cntrl device fr inlet CO. Bth units effectively cntrlled drs frm the Zimpr prcess since start-up, and stack tests in 1992 indicated a VOC cntrl efficiency in excess f 98%. Hwever, neither RTO was able t cnsistently cmply with the pennit cnditins fr CO emissins. Bth RTOs emitted higher levels than the pennitted mass emissin rate fr CO during the riginal stack tests. With the inlet cncentratin f CO as high as 3,600 ppm and the inability f either RTO t regularly cmply with the penn it t perate, the prblem f CO cntrl was addressed. PVSC first attempted t mdify the pennit t reflect the high inlet cncentratin and a cntrl efficiency fr CO similar t the 98% cntrl efficiency fr VOCs. The New Jersey Department f Envirnmental Prtectin (NJDEP) requested that PVSC. cntinue t investigate alternate methds f cmpliance with the existing pennit cnditins. Suggested areas f study included baffles t increase mixing, increased temperatures in the cmbustin chamber, and valve sequencing. Early investigatins revealed that the utlet CO cncentratin fllwed a cyclic pattern which repeated with each valve cycle f apprximately 3-4 minutes, as shwn in Figures (3) and (4). The amplitude f the "peaks" varied prprtinally with the inlet CO cncentratin. At this pint, PVSC cntracted with Carlsn Assciates t investigate the perfnnance f the RTOs and pssible methds f cmpliance with the existing pennit I Testing Prcedures Until the testing prcedures were devised, PVSC had studied the RTOs as an entire system. sampling nly the inlet and utlet streams. The testing prcedures implemented were designed t mnitr CO cncentratins entering each canister, within the cmbustin chamber, and exiting each canister. The lcatins f the test prts utilized in the testing are shwn in Figure (1). The testing prcedures included the fllwing: 1. The ttal destructin efficiency fr bth CO and VOCs was evaluated by sampling the inlet and the utlet streams with cntinuus analyzers. 2. Temperature and CO prfiles in the cmbustin chamber, between each f the canisters, were mnitred. This prtin f the testing prgram was designed t detennine whether shrtcircuiting within the cmbustin chamber r the lcatin f the burners were factrs cntributing t the high utlet CO cncentratins. 3. The effect f temperature n CO destructin efficiency was mnitred. The temperature in the cmbustin chamber was increased frm l,520 F t l,720 F, in increments f 50 F. 4. Sampling belw all three media canisters detennined the cncentratin f CO during bth the inlet and utlet cycles. During the inlet cycle, the cncentratin f CO entering the canister was mnitred. During the utlet cycle, the utlet cncentratin f CO frm the canister was mnitred prir t flwing past the clsed inlet valve. By mnitring the CO cncentratin ver a series f inlet and utlet cycles the destructin efficiency within the cmbustin- chamber was determined. 5. CO levels in the purge ductwrk were mnitred t determine the success f the purge cycle prir t the pening f the utlet valve. The purge cycle time was increased t determine if the purge time was adequate. If the purge cycle is nt cmplete, "peaks" f CO wuld result as the slug f bad gas was released int the utlet stream. 6. Sampling between the inlet and utlet valves was perfnned t detennine if leakage past the valves cntributed t the elevated CO emissins. This testing was perfnned n RTO # 1 which cnsistently exhibited higher utlet CO cncentratins than RTO #2. The testing prgram lasted several days. Initial Testing Results The initial testing f RTO # 1 indicated that the system cntrl efficiency fr bth CO and VOCs was apprximately 95%, as shwn in Table (1). Hwever, sampling within the cmbustin chamber and belw the media beds indicated that the CO and VOC destructin efficiency within the cmbustin chamber exceeded 99.9010. N significant stratificatin f temperature r CO cncentratin was seen within the cmbustin chamber. The cncentratin f CO in the utlet stream did nt decrease as the tempera was increased.2 Testing within the purge ductwrk shwed that the cncentratin f CO prir t the utlet cycle was reduced t levels belw 600 ppm. and did nt imprve as the purge cycle time was increased. Due t the small vlume f gas cntained in this area, the purge cycle did nt seem t cntribute a significant amunt f CO t the utlet stream. The cncentratin f CO exiting the cmbustin chamber and canisters, as measured belw the media bed and in the cmbustin chamber, ranged between 0 and 10 ppm. Testing just beynd the inlet valve during an utlet cycle (inlet valve clsed) revealed CO cncentratins as high as 200 ppm.2 The lw cncentratins f CO belw the media bed and significantly higher cncentratins beynd the inlet valve during an utlet cycle were bserved repeatedly in each f the three canisters. The results f this testing shwed that the RTO achieved nearly 99.9% cmbustin f bth CO and VOCs within the cmbustin chamber, with an average residence time f clse t 1 secnd. The testing suggested that almst n shrt circuiting 439
Figure (3) - Outlet CO (ppm) Prir t Valve Replacement Outlet CO (ppm) - RTO #1 Prir t Valve Replacement 160,----, _140 &. 120 Q. - 100 80 1) 'S 60 40 20 ---- ---- ---- 200 400 600 Time (sec.) I Inlet CO 2.000 ppm J Figure (4) - Outlet CO (ppm) Fllwing Valve Replacement 160 _ 140 &. 120 f- a 100 0 80 0 -G) 60 r- = 40 r- 0 Outlet CO (ppm) - RTO #1 Fllwing Valve Replacement -- - 20 0 0 200 400 600 Time (sec.) I Inlet CO 2.660 ppm) 440
was ccurring between each canister within the cmbustin chamber, the perating temperature f 1,520 F was sufficient, and the lcatin f the burners was satisfactry. The purge cycle seemed t be adequately evacuating high cncentratins f CO prir t the utlet cycles and was nt a great cntributr t the excessive levels f CO in the exhaust stream. In summary, there are three surces f CO (and any ther cntaminant in the inlet stream) in the utlet gas stream: I. Residual frm the cmbustin chamber 2. Any residual frm the purge cycle prir t an utlet cycle 3. Leakage directly past clsed valves The residual prtin frm cmbustin was less than 0. 1 % f the inlet, and given adequate purge times and flws. the residual frm the purge cycle was small. The results indicate that the valve leakage is respnsible fr the majrity f any cntaminant in the utlet stream. Clsed inlet valves were allwing between 5% and 10% f the inlet gas t leak past the valve directly int the utlet stream. This leakage allwed the inlet gas t bypass the cmbustin chamber and exit the RTO thrugh the utlet stack. Depending n the amunt f sludge prcessed by the Zimpr units (varying the inlet CO cncentratin), the average utlet CO cncentratin ranged between 50 and 200 ppm. Rather than using the term destructin efficiency t describe afterburner perfrmance, the term cntrl efficiency wuld be mre apprpriate since the destructin efficiency is respnsible fr nly a small amunt f the utlet cncentratins. The mst startling result f the testing was that the RTO was nt achieving the required 98% cntrl efficiency f VOCs. If CO emissins had nt been a cnstant prblem with the RTOs, due t the high inlet cncentratin f CO, PVSC wuld have assumed prper VOC cntrl based n the 1992 stack test. The reduced cntrl efficiency wuld have been undetc:cted until the next stack test was perfrmed 5 years after the riginal test. Withut CO in the inlet stream, mnitring the utlet CO cncentratin will nt give an indicatin f the leakage past valves which accunted fr the majrity f bth VOCs and CO in the utlet. Valve Replacement After recgnizing the reduced cntrl efficiency, PVSC prpsed the replacement f the inlet, utlet, and purge valves t the NJDEP. The new valves were state f the art. designed t seal mre effectively than the existing valves. Thc valve bdies were als designed t resist crrsin, which may have accunted fr the increased leakage and the decreased cntrl -.- efficiency f the RTOs in the past. In rder t determine the leakage rates f each valve, additinal test prts were installed directly abve each utlet valve. The CO cncentratin exiting the cmbustin chamber culd be determined frm the test pn belw each canister. During an utlet cycle, any increase culd be attributed t leakage past the inlet valve as shwn in Table (2). A secnd series f tests were cnducted fllwing the valve replacement. The testing demnstrated that the RTOs were achieving the required 98% cntrl efficiency f VOCs and the prpsed cntrl efficiency f 98% fr CO, as shwn in Tables (3) and (4). The permit t perate was mdified t allw fr a mass emissin rate fr CO based upn a 98% cntrl efficiency f the inlet CO, uncrrected. This testing indicated that the VOC and CO cntrl efficiencies were similar, since the majrity f lsses were due t leakage past the valves. The testing was als used t determine leakage past each individual valve. Since leakage accunts fr the majrity f the utlet cncentratin f cntaminants, this is the area that allws fr the mst imprvement f cntrl efficiency. Findings The mst imprtant finding was that bth units were capable f cmbusting 99.9% f bth VOCs and CO within the cmbustin chamber. The pssible cmbustin f VOCs in the lwer prtin f the media beds (600-800 F) and CO in the upper prtin f the media beds (1,200-1,550 F) may cntribute t the bserved destructin efficiencies. A pressure drp f 8 inches f water clumn acrss each media bed seems t prvide adequate mixing f the gas stream resulting in little stratificatin r shrt circuiting in the cmbustin chamber. Despite the destructin efficiency bserved within the cmbustin chamber, leakage past inlet and utlet valves may limit the ultimate perfrmance f a RTO system. Due t the regenerative nature f the system and the number f valves, the verall system efficiency is limited by the valve design. Cnsideratin shuld be given t the required cntrl efficiency and alternate methds f valve arrangement, such as in series r with a psitive air seal. The use f a system with nly single valves similar t thse installed at PVSC may limit a system t a 98% cntrl efficiency fr permitting purpses. Cnsideratin must als be given t pssible valve perfrmance decline ver their lifetime, with cmfrt built int the permit. In rder t mnitr any future deteriratin in-the perfrmance f the new valves, PVSC will cntinue t mnitr the cntrl efficiency f each RTO. It is anticipated that the RTOs will be tested by PVSC persnnel n a quarterly basis. The leakage rates past each individual inlet and utlet valve will be mnitred using prcedures similar thse emplyed during these tests. Any decreased perfrmance will hpefully be detected early, and pssible slutins explred befre the diminished efficiency results in any permit vilatins. With regards t leakage past valves, test results suggest that the installatin f the valves is critical fr effective perfrmance. Butterfly valves similar t thse installed at PVSC must rest n a true and level surface. Any irregularity in the supprting surface may result in distrtin f the valve bdy and seat, 441
Table (I) RTO # I Cntrl Efficiency Test Results 2 Prir t Valve Replacement Inlet VOC (ppm) 1511.28 Outlet VOC (ppm) VOC Cntrl EfI. (%) Inlet CO (ppm) 76.69 94.9 2864 Outlet CO (ppm) CO Cntrl EfT. (%) 139.3 95.1 Table (2) Inlet Valve Leakage Testing Results ) Fllwing Valve Replacement RTO #2 Canister Inlet CO Cncentratin CO Cncentratin Belw CO Cncentratin At Outlet Average CO Cnc. At I 2 3 (ppm) 2000 2000 2000 Canister (ppm) Valve (ppm) -0 70-0 30-0 25 Exhaust Stack (ppm) 70 40 20 Table (3) RTO # I Cntrl Efficiency Test Results ) Fllwing Valve Replacement Test I 2 3 Inlet VOC (PP!1l) 1440.8 1432.7 1291.8 Outlet VOC (ppm) VOC Cntrl Eff. Inlet CO (ppm) 21.1 22.8 20.8 (%) 98.5 1870.3 98.4 1831.2 98.4 1956.3 Outlet CO (ppm) CO Cntrl Eff. (%) 34.9 98.1 34.1 98.1 36 98.2 Table (4) RTO #2 Cntrl Efficiency Test Results ) Fllwing Valve Replacement Test I 2 3 Inlet VOC (ppm) 1511.2 1561.3 1833.5 Outlet VOC (ppm) VOC Cntrl Eff. Inlet CO (ppm) (%) 17.3 98.9 2496 17.5 98.9 2533.7 21.S 98.8 3018.4 Outlet CO (ppm) CO Cntrl EfI. (%) 31.2 98.8 31.3 98.8 28 99.1 442
affecting the ability f the valve t seat prperly. It has als been bserved that stresses assciated with the supprt f additinal weight, such as any maniflds abve the valves, shuld be translated directly t the supprt f the valve and nt t the valve itself. Additinal time and care during the installatin f the valves may be rewarded by an imprved cntrl efficiency f inlet cntaminants by minimizing. direct. leakage past the valves. Withut CO present in the inlet stream, mnitring the utlet CO cncentratin as an surrgate fr VOC destructin may nt indicate the actual perfrmance f a RIO. The destructin efficiency f any cntaminant within the cmbustin chamber may be within acceptable levels. Hwever, VOCs bypassing the cmbustin chamber by leaking past valves will nt be detected withut an n-line ttal hydrcarbn analyzer. Mre frequent mnitring r testing f an RIO may be required t assure cntinued cmpliance with any permit cnditins. A cntinuus CO mnitr may be effectively used as an indicatin f bth VOC and CO cntrl if CO is present in the inlet stream. If the permit is based upn a cntrl efficiency fr bth VOCs and CO. and mass emissin rates based upn histrical influent cncentratins and flws, cmpliance with the permit can be mnitred using a cntinus CO analyzer n the utlet stream. time due t deteriratin in valve perfrmance shuld als be cnsidered. After installing any system, cntinued mnitring and testing by the wner will play a valuable rle in prper surce cntrl and cntinued cmpliance with the permit t perate. Acknwledgments The authrs wish t acknwledge Mr. Richard Sharp and Mr. Curtnie Lgan f Smith Engineering fr their wrk in testing and data interpretatin. Thankyu t Mr. Victr Fnseca and Mr. Dnald Stacy f PVSC. amng thers. fr their invaluable assistance. References I. Baturay, A., Regenerative Thermal Catharpin, VA. 1994, "Perfrmance Oxidizers (RIO)", Evaluatin fr Carlsn Assc., 2. Sharp, R. C., 1994, "Engineering Study" Smith Prject N. E-4493, Smith Engineering C., Brmall, PA. 3. Sharp, R.C., 1995, "Engineering Study fr Carlsn Assciates" Smith Prject N. E-4507, Smith Engineering C., Brmall, P A. Finally, the tests undertaken by PVSC indicate that an inlet xygen cncentratin as lw as 9% may be sufficient fr adequate cmbustin f bth CO and VOCs. Originally, the average xygen cncentratin f 12% t 13% was cnsidered a pssible factr in the inability f the RIOs t effectively cntrl CO. N increase r decrease in cntrl efficiency has been detected as the xygen cncentratin has varied hetween 15% and 9%. Cnclusin When cntrlling CO r VOC emissins, cnsideratin must be given t the verall cntrl efficiency f the system. While an xidizer with indirect heat exchangers may be capable f cnstantly exceeding 99% cntrl efficiency, a regenerative system is limited by valve perfrmance. Depending n the desired cntrl efficiency f the system, apprpriate valves r valve cnfiguratin must be included in the design. Ihe use f a cntinuus CO analyzer t mnitr the VOC cntrl efficiency f a RIO system withut CO in the inlet stream may nt give an accurate representatin f the actual perfrmance f the RIO. While mnitring CO may assure that the destructin f VOCs in the cmbustin chamber is within acceptable limits, it will nt detect an inadequate purge cycle r leakage past valves. The permitting f any surce shuld include cnservative estimates f cntrl efficiency which allw fr pssible leakage past valves. The pssible reductin in cntrl efficiency ver 443