DCN: 89-232-011.034-OS DRAFT TEST REPORT PILOT-SCALE ESP AND HYDRO-SON1C SCRUBBER PARAMETRIC TESTS FOR PARTICULATE, METALS AND HC1 EMISSIONS JOHN ZINK COMPANY RESEARCH FACILITY • TULSA. OKLAHOMA Prepared for: Mr, Shiva Carg Office of Solid Waste U.S. Environmental Protection Agency .Washington, O.C. 20460 Prepared by: . Radian Corporation Progress Center 3200 Cast Chapel Hill Road Research Triangle Park. Kerch Carolina 27709 June 1989 RR3038U3
Transcript
DCN: 89-232-011.034-OS
DRAFT TEST REPORT
PILOT-SCALE ESP AND HYDRO-SON1C SCRUBBER PARAMETRICTESTS FOR PARTICULATE, METALS AND HC1 EMISSIONS
JOHN ZINK COMPANY RESEARCH FACILITY• TULSA. OKLAHOMA
Prepared for:
Mr, Shiva CargOffice of Solid Waste
U.S. Environmental Protection Agency.Washington, O.C. 20460
Prepared by:
. Radian CorporationProgress Center
3200 Cast Chapel Hill RoadResearch Triangle Park. Kerch Carolina 27709
June 1989
RR3038U3
TABLE OF CONTENTS . . ..• - • '-; • ' •Section , i .
2.0 PROCESS DESCRIPTION AND OPERATION.......................... 2-12.1 Process Description........................ .......... 2-12.1.1 Beleran Two-Stage ESP;................................ 2-32.1.2 Hydro-Sonic Scrubber................................ 2-32.2 Process Operation..................................... 2-7
3.1 Particulate Matter Results............................ 3-13.1.1 Control Device Inlet Results................... 3-13.1.2 ESP Outlet Results............................. 3-33.1.3 Hydro-Sonic Scrubber Outlet Results............ 3-53.1.4 Gas Sampling Parameter and Conditions..;....... 3-8
3.2 Metals Results........................................ 3-10. 3.2.1 Inlet to the Control Device.................... 3*10^3.2.2 Outlet to the Control Device................... 3*133.2.3 Metals Removal Efficiencies.................... 3*133.2.4 Ratio of Metals to Particulate................. 3*17
«!o QUALITY ASSURANCE AND QUALITY CONTROL. ..................... 6-1S.I QA/QC Program Objective*............... ....,...;../ 6-16.2 M«thod-Sp«clfic QA/QC... ............... ............... 6-3
6.2.1 Central Quality Assurance and Quality Control , "Procedures ..................................... 6- 3
6.2.2 Particulate and Metal Sampling and Analysis.... 6-46.2.3 HC1 Sampling and Analysis.. .................... 6-7
1111 .-• ' • • ' • '
JBSOS7
LIST OF FIGURES
. ' • ' • - . • • • . . . " . • - Page
2-1 v Process Flow Diagram with Sampling Locations for theUet-ESP Trials at the John Zink Company ResearchFacility.;..................................."..........,.,, 2*2
2-2 Schematic Diagram of the Beltran Two-Stage WetElectrostatic Precipitacor................................. 2*4
2*3 Process Flow Diagram with Sampling Locations for theHydro-Sonic Scrubber Trials at the John Zink CompanyResearch Facility............................................ 2-5
3-1 -• PSD Composite Curves Generated from Inlet Testing Data atLow Grain Loading Conditions, John Zink Research Co..Tulsa, OK (April 1989)..................................... 3-21
3-2 PSD Composite Curves Generated from Inlet Testing Data atHigh Grain Loading Conditions, John Zink Research Co;,Tulia. OK (April 1989),.................................... 3-22
4-1 Inlet PSD and Particulate/Metals Sampling Locations.John Zink Research Co.. Tulsa, OK (April 1989)............. 4-2
4-2 Inlet PSD Sampling Configuration Using the Anderson HCSSHigh Grainlbading latpactor, John Zink Research Co.,Tulas, OK (April 1989)..................................... 4-3
4-3 Outlet Particulate/Ketals Sampling Location for the VenturiScrubber/Wet ESP Performance trials. John Zink Research Co.,Tulsa, OK (April 1989)..................................... 4-5
4*4 Outlet Metals/Particulate Sampling Location for theHydro-Sonic Scrubber Performance Trials, John Zink ResearchCo., Tuisa, OK (April 1989)................................ 4-6
4-5 Schematic of the Toxic Metals Sampling Train............... 4.84*6 Components of the Particulate/HCl Sampling Train........... 4-14
5-1 Digestion and Analysis Scheme for Toxic Metal TrainComponents - Front Half.................................... 5-3
5-2 Digestion and Analysis Scheme for Toxic Metal TrainComponents - Back Half...........*.......».>.....••'•.••••* 5-4
. . - , : ; . i v
JBS057 . '.-- • ',
flR3038t*6
LIST OF TABLES
JBS037
2-1 Test Material Feed Rates. .................................. 2-92*2 Summary of SEM Results for Radian * 200 Mesh Fly -Ash •
Sampl*.. .......... ......................................... 2-112*3 Metals Concentration in MSW ESP Ash. ....................... 2-122-4 Summary of Experimental Test Conditions. ................... 2*' V
3-2 Summary of ESP Outlet Test Result*. . ....................... 3*43.3 * Summary of Hydro-Sonie Outlet Test Results..... ............ 3*63*4 .Summary of Hydro* Sonic Outlet Test Results Comparing
Scrubber Steaa and Pressure; Drop Parameters. .. .............. 3-73-5 Summary of Gas Sampling Parameters and Conditions for
Particulate/Ketals Sampling. ............................... 3-93-6 Summary of Inlet Metals Mass Rates for the John Zink Test
Prograa. ........ j.. ........................................ 3*113*7 Comparison of Metals Feeds Versus Actual Mass Rates i
Measured at the Inlet... ................................... 3*123-8 Summary of the- Outlet Metals Flue Gas Concentration for the J
John Zink Test Prograa. ..................................... 3-143-9 Summary of the Outlet Mass Raess foe the John Zink Test
Program...................../......................"..... ... 3*153* 0 Metals Removal Efficiency Across the Control Device for the
John Zink Test Prograa. .................................... 3-163-11 Ratio of Metals to Partleulates at the Control Device Inlet
for the John Zink Performance Trials. ...................... 3-183-12 Summary of Inlet Particle Sizs Distribution Results........ 3*203-13 Summary of HC1 Sampling Results. ........................... 3-24
4*1 Particulate/Metals Glassware Cleaning Procedures. . ......... 4*104*2 Staple Recovery Components for the Particulate/Metals
6-4 Method and Reagent Blank Results for Metal SampleAnalyses • Front Half Samples... ......,.....••••*•••••••••« 6*9
6-5 Method and Reagent Blank Results for Metal SampleAnalyses • Back Half Impingers 1, 2. 3. 4.................. 6*10
6*6 Method Blank Results for Metal Staple Analyses - VastsFeed. .... ..........,.........,..............••••••-• «V" •••• *'*t
v . I
flft3038U7
SECTION I
-- - • , INTRODUCTION
the Environmental Protection Agency's Office of Solid Waste (EPA/OSU) iscurrently developing additional regulations to control emissions ofparticulate matter, toxic metals and hydrochloric acid (HCl) from hazardousvasts Incinerators. Radian Corporation, under subcontract to Versar, Inc..was contracted by the EPA to collect emissions data to support thesei . . - .regulations. The testing was conducted on two pilot-scale air pollution.control systems installed and operated at the John Zink Company ResearchFacility in Tulsa, Oklahoma. The test program took place between March 28 andApril 4, 1989. This report presents the results of the testing program anddescribes the processes involved as veil as the testing and analytical methodsused. • • . • • ' . • ' • , ' " .
•' The primary objective of the test program vas to collect particulate andtoxic metals removal efficiency and emissions data for two pilot-scale airpollution control systems:
* " • ' ' .,
o A Hydro-Sonic* vet scrubber; and
o A Beltran two-stage electrostatic precipitator (ESP).
The toxic metals examined in this test program wars arsenic (As), cadmium(Cd), chromium (Cr). lead (Pb) and mercury (Kg). The control devices weretested using a flue gas stream generated by injecting an ash containing thetoxic metals into a natural gas fired horizontal incinerator. An
organo-Atssnie* compound (monosodlua methylarsonic acid) and a salt (Al (SOsolution were also injected into the incinerator to generate subaicron aetsfume and inert subaicron partlculates, respectively. Carbon tttrachloride wasfired to produce HC1 and thus simulate ths flue gas conditions of a hazardousvasts incinerator firing chlorinated vasts.
A second objective of the test prograa was to evaluate the potential useof a surrogate metal mixture as a means of assessing control efficiencies,establishing allowable vasts feed metals concentrations and stack emissionlevels for the toxic metals. To accomplish this objective, four selectedmetals, copper (Cu), bismuth (Bi). strontium <Sr), and'aegnaslua (Mg), werealso fed into the incinerator during the testing. In combination, thesesurrogate metals are expected to represent all of the toxic metals withrespect to volatility and behavior in the combustion system, e.g. likelihood' . tof fume formation or entrsinment. The surrogate metals mixture was fed'to the
•*
incinerator to provide s draft comparison of removal efficiencies and emissionlevels for. the surrogate and toxic metals.
Tl.C. Burton, V.O. Clark and tf.R. Seeker, EEft, Prafe Topteel Reaorc *Surrofecie Metele Mixture. £P^ Contract 68*03*3363, Engineering Analysis ofHazardous Vaste Thermal Destruction, October 1981..
JBS057 1-2
AR3038l»9
o
. SECTION 2
PROCESS'DESCRIPTION AND OPERATION
2.1 PROCESS DESCRIPTION• • . , . /
: . ' ' •
The pilot-scale test system consisted of a natural gas-fired incineratorfollowed by the emission control equipment. The horizontal fired incineratorhas a design heat value of 4,3 million BTU/hour, with a gas residence time ofapproximately two seconds. The incinerator was fired at temperatures near1600°F and the MSU fly ash and other materials were injected into theincinerator as appropriate to achieve desired particulate loadings andparticle size distributions. Details of the test conditions are discussed inSection 2.2. •:';
The flue ges was directed from the incinerator to a conditioning tow«rwhere the temperature was lowered to approximately 550°F by direct contactwith water. Gas flow rates at this point ranged from 650 to 950 dscfm. Fromthe conditioning tower, the flue gas was passed through a quench pot wheretemperatures were lowered to approximately 180°F. The flow was then directedto the air pollution control system to be tested.
For tests on the two-stage ESP system, a clip stream of approximately130 dscfm (218 aefa at 145°F) was diverted to the ESP inlet, as shown inFigure 2*1. A conventional variable throat venturi was in-line upstream ofthe ESP, but was operated with the throat fully open (two inches of vaterpressure drop) and used only for scrubbing HCl from the gas stream. Duringthe Hydro-Sonic scrubber performance tests, the venturi scrubber vas bypassedand the entire flow of flue gas vas directed to the scrubber (Figure 2*3).Flue gas vas pulled through the control system by an Induced draft fan, and
JBS057'' ' - 2*1
AR303850
1'1]
t1
rJ
11 li
I
s
£
o2-2 flH30385l
then directed through'a secondary afterburner-and to ths atmosphere via a' stack. Each of the air pollution control systems is described in detail in the
sections below. .
2.1.1 Belcran Tve-Staye ESP .• ,' - • . • i
The pilot-scale Beltran two-stage ESP was a tubular preclpltator designedfor particulate and acid gas emission control. A schematic diagram of theBeltran two-stage ESP is shown in Figure 2-2. Flue gas entering the ESP firstpasses.through a small pre-cleaning section intended to saturate the gas withmoisture and remove large particles. The flue gas then passes into the firststage of the ESP where particulate is charged and subsequently* collected.During this test program, the ESP vas operated dry, with no vater sprayed tothe pre-cleaning section. The Beltran operator claimed that sufficient .moisture WAS present in the gas stream.
•2.1.2 . Hvdro-Sente Scrubber '
'].. • -{ J' • Upon completion of the CSP trlaL*. the venturi scrubber vas removed and
the entire gas flov vas directed to the Hydro-Sonic scrubber as shown inFigure 2-3. The pilot-scale Hydro-Sonic scrubber tested was a SupersubFan/Ejector Drive Tandem Nozzle, System. A schematic for the Hydro-Sonic• ' . , • • - ' •system is shown in Figure 2*4. The flue gas exiting the quench pot passesthrough a venturi section where the gas stream is saturated at approximately173°F. Flue gases then pass into the first of the tandem nozzle systems. Theflue gas, wat-r and steaa first pass through a supersonic ejector nozzlecontained within a subsonic ejector nozzle. The area surrounding the subsonicnozzle is operated under negative pressure, creating a cone shaped standingvave consisting of high localized pressure gradients at the nozzle exit. Inaddition, water is injected through vee-jet spray tips around the nozzle exit(countercurrcnt to the gas flov). Water encountering the gas exiting thenozzle is atomized forming small irregularly shaped droplets which serve asimpact targets onto vhich ptrticulatt, metal salt and aetal oxides collide and
art captured. The cone shaped area where the atomization and impaction occursis called the mixing zone. The mixture then passes into a mixing tube sectionwhere agglomeration of water droplets occurs. ,
i
The flue gas containing these water droplets passes through'stage ofsupersonic -subsonic nozzles, a nixing zone and a mixing tube identical tothe first stage except that steam is not used in the supersonic nozzle. Theflue gas then passes through a cyclonic separator where the agglomerated waterdroplets are collected and removed. •
2,2 PROCESS OPERATION , .
The purpose of this test program WAS to evaluate the performance of twoair pollution control systems for use in particulate and toxic*aetals emissioncontrol for hazardous waste incinerators. The test program was designed toevaluate the air pollution control system performance at two levels of totalparticulate loading and with two particle size distributions. In order tohave better control of the flue gas composition, a horizontal naturalgas-fired incinerator was used to develop a simulated hazardous waste fluegas. The desired particulate loading, particle size distribution (PSD), andmetals content of the flue gas was controlled by injecting appropriatequantities of the following materials•
e MSW 'ash* and bismuth nitrate were -co-fired into the incinerator withnatural gas. A predetermined quantity of the mix vas continuouslyfed into a screv auger feed mechanism which conveyed the materialinto the incinerator burner; '
*The ash vas collected as a composite simple froa a aultiple field ESP on aaass burn municipal vast* incinerator. The ash contains material collectedin all fields of the ESP.
JBS057 • . . ' , 2 - 7
AR303856
* Aa aqueous solution containing monosodiua nethylarsonic acid (KSMA jstrontiua nitrate, cuprle nitrats, and aagnesiua nitrat* vas ^ ^
. injtetad through a lance into the flame at the exit of the burner;
o A 30 percent aqueous solution of aluainua sulfate vas injectedthrough a lance and into the flaae at the exit of the burner, and
• A carbon tetrachleride solution vas injectedthrough a lance into the flaae at the exit of the burner (to produceHCl).
During this prograa, tests were conducted at four flue gas grain loadingconditions:
. *
1. 2 gr/dsef and a lov percentage of sub-eicroa particulate (hereafterreferred to as low/low); I
•' ' ' • ' * '2. 2 gr/dsef and a high percentage of sub-micron particulate
(hereafter referred to as low/high);3. 10 gr/dscf and a lov percentage of sub-micron particulate
(hereafter referred to as high/low); and4. 10 gr/dscf and a high percentage of sub-aicroa particulate
- (hereafter referred to as high/high).
Particulate loadings of 2 gr/dscf and 10 gr/dsef vere selected fortesting to represent the lower loadings expected froa liquid*firedincinerators and the higher loadings expected froa rotary kilns treatingcontainerized wastes and contaminated soils and debris. Effective removal ofsub-micron particulate have been purported to be essential in achieving loveroverall total particulate concentrations (i.e., less than 0.08 grains/dscf).The fraction of sub-microa particuiate present in the gas streaa vas varied bychanging the ratio of HStf ash to aluainua sulfate injected into the gas streaato obtain the desired particulate loading. Feed races for each of theaattrials injected into the afterburner during the four df**««nt test
Injection of th« MStf fly ash provided larje particles, as veil asthe desired toxic metals. A mass distribution determined for the KStf flyby scanning tleetroa microscope (SEN) is shown in Table 2*2. It can be seenthat approxiaatsly 33 percent (by veight) of the ash is smaller than 40 eo 8<Xmicrons. Ths concentration of As. Cd, Cr, Pb, and Hg present in the MSW f lyash is shown in Tabl* 2*3.
The concentration of sub-aleron particulate and aeid gases vas controlledby adjusting the feed rates of aluainua sulfate, KSHA, and carbontetrachioride. The aluainua sulfate (AMSf )-) provided moderate tosub- micron size particulate as Al.Oj and aeid gases as SO./SO-. The MSHA(NaCH4AsOj) vas chosen to provide a sub -micron fume as As.O. and sub -micronparticulate as NaCl. The carbon tetrachioride provided HCl.
\ • •
The initial test plan specified that the surrogate aetals (bismuth*copper, aagnesiua and strontium) be injected into the Incinerator as an 'i. • . • \ 4aqueous solution. Hovever, after conducting saall scale nixing tests iff che _field, it vas determined that the bismuth nitrate did not dissolve complete!] jinto solution vita the other non-toxie metals. This vas apparently due topoor water solubility of bismuth nitrate. The bismuth nitrate vas, therefore.blended with the MSW ash and fed through a screv auger into the incineratorburner. - "
A summary of the feed conditions tested for the two -stage ESP and for theHydro -Sonic scrubber is presented '-i Table 2-4. Original plans called forconducting s total of four test conditions with the ESP. Hovever,insufficient time and resources vers available to complete all four tests, dueto difficulties in obtaining desired gas flov rates through the ESP and delaysin installing a functional drain systea for the pilot unit. Three tests werecompleted: hovever, the first test run vas invalidated due to * number ofprocess and sampling problems. These problems included a high gas flovthrough the ESP. vatsr in the duet leading to the ESP, and in the ESP,improper input rates for the feed materials and discoloring of the outletparticulars filter by vatsr froa the ESP. ,r . * i -./ : • • ' • . . • • : • ' • " • ' • ' • •JBS037 2-10
AR3Q3859
TABLE2-2. SuWMARY OF S|M RESUUS FOR
Particle Size Range (urn) Kass Distribution (Wt. %)
TABLE 2-4. SUMMARY OF EXPERIMENTAL TEST CONDITIONS
' '
Air PollutionControl System
Two -Stage ESP
Hydro -Sonic*Scrubber
Test Run
v ' ic ; .-. •'2
; ' 3 ' .•: . '•• •456 -: ' '•7 A8
*V
Particulate4Loading
LovLovHigh
LovLovHigh.HighLovLov ;
Particle Size6Distribution
LowHighHigh
LowHighLowHighLowLow
*Lov particulate loading » 2 gr/dscf; high particulate loading * 10 gr/dscfLov and high particle size conditions represent lov and high fractions ofsubmicron particles.eThls test run vas invalidated due to a number of process and samplingproblems. . .Replicate tests to determine combined process, sampling, and analyticalprecision.
JBS037 2-13
fl«3Q3862
- • . - - ' " ' •• iA total of ten test runs vere performed with the Hydro-Sonic0 scrubb*rS-
Six-of these tests vere a part of the original EPA test prograa. These testsincluded one test run at each of the four test conditions, plus two additionalreplicate tests at the lov grain loading/low fraction ef sub-aieron particles.The replicate tests vere performed to provide a aeasure ef the combined 'process, sampling and analytical precision. The final four test runs whichvere conducted at the low/low test condition vere performed for the John ZinkCompany and are reported in this EPA report as supplemental data. These tests .vere performed to assess the effect of scrubber pressure drop an<i steaa use onparticulate removal effectiveness.
'
JBS057 ' - 2-14
SECTION 3
; SUMMARY OF RESULTSi . . ' '
Final results for particulate matter, metals and HC1 emissions andparticle size distribution testing conducted at the John Zink facility on *pilot-scale 2- stage ESP and a Hydro-Sonic scrubber ere presented in thissection. Total particulate results are presented in Section 3.1, met el*results are presented in Section 3.2, particle size data are discussed InSection 3.3 »nd the HCi results are presented in Section 3.4.
3.1 PARTICULATE MATTER RESULTS
• " • " • - . ' " • ' . ' \ .Sampling for flue gas partieulate matter/aetals emissions vas conducted
simultaneously *t the control device inlet and outlet according to a draft .EFA/EMSL method entitled "Methodology for the Determination of Trace MetalEmissions in Exhaust Gases from Stationary Source Combustion Processes.*Partieulate matter emissions vere measured during two valid test runsconducted on the 2-stagc ESP (Runs 2 and 3) and during 10 test runsconducted on the Hydro -Sonic scrubber (Runs 4 through 13). Test Run 1 onthe ESP vas invalidated due to both process and sampling problems. Inletand outlet partieulate sampling results are discussed separately in thesections belov.
3.1.1 otgrol Peviee Inlee Results
Partieulate concentrations and ease emission rates for samplingconducted at the quench tower inlet (inlet) are summarized in Table 3*1.The inlet results show particulars loadings ranging from O.S7 to1.70 gr/dscf for the "lev loading1' cast condition and partieulate loadingsranging froa 3.06 to 4.60 for the high loading tests. The measured
I t JBS057 ' . . -3*1"X - X ' - " • ' - , - • • - ' - ' . , , . '
AR303861*
ww
5 3
"5 t
33 S<X
..* 4
» M * Al
aa i
«• —
5 e> «• 3
S j 2 5 5
2 3 - I
3-2
I - '• ' • w14 • .• • • . • •
4R303865 *
partieulate loadings represent approximately 30 to 70 percent of theI j' , ' loadings targeted. The difference between the amount of material injected
and'the aass loadings found in the gas stream is believed to be du'e to lossor drop out of material In the incinerator or ducting prior to the inletsampling point. Some ash vas found in the incinerator upon completion ofthe testing. Comparison of the mass rate of aetals fed to the incineratorcompared to the Bass rate of metals determined at the inlet sampling point
. shovs similar results and is discussed further in Section 3.2.. / ' • •, - * •
3.1.2 ESP Outlet Reaulea
.; ' Particulate sampling results for the ESP outlet arc summarized alongwith ESP particulate removal efficiencies in Table 3*2. Oxygen data
. presented in Table 3-2 for the ESP^tests shov that a significant amount ofair In-leakage occurred between the quench tower Inlet end ESP outlet .sampling points. The oxygen concentration at the inlet vas 7.3 and 6.1percent for the two tests, and the corresponding ESP outlet oxygenconcentrations vere 17.8 and 18.0 percent. Static pressure through the ESP
\^J Was approximately -35 inches of vaeer and the air in-leakage is believed to> • ~ • •have occurred through a flange at which a restriction plate vas installedjust prior to Runs 2 and 3 (see Figure 4-4) and through two half-inch holesopened in ports on the side ef the ESP by the Beltran operator. The portsvere installed to provide purge air for the electronic insulators and therestriction plate vas installed to reduce flov through the ESP to thedesired 200 acfa. ' .
ESP outlet partieulate concentrations were 0.0072 gr/dscf at 7 percent0. for the lov partieulate loading/high fraction of fine particles(low/high) condition and 0.0279 gr/dsef at 7 percent 02 for the highparticulate loading/high fraction of fine particles (high/high) condition.The higher outlet concentration for the high/high condition vas apparentlycaused by the higher partieulate loading entering the ESP.
JBS057 3-3
AR303866
3
3d
w
M
* - - .
3d1
•• -II
-i?a
e
*cf
!wi
* , *
* •
*
3* .
3 3
3 3
* •H 114 4
« «•
Al 1*
i! Iw «•iIpill
• X«• u S
Jii3
AR3038673-4
3.1.3 Hvdro-Sonie Scrubber Outlet Resujcy .:'. ' ' - I H 1
A J '"" f' • ' •N-^ .Tests conducted on the Hydro-Sonic scrubber vere performed at-differentscrubber pressure drops, with and without steam injection, and at different
, particulate loadings and particle size distributions.
. Results for test Runs 4'through 9 are presented in Table 3-3. These: test runs vere all conducted with an overall scrubber pressure drop of
approximately 31 inches of weter and with steam Injection. Runs 4. 8 andjv 9 represent replicate tests performed at the lev/lev test condition. These\ ' - ' ' • ' . , ' • - •I test runs provide a measure of the combined process, sampling and analytical. precision for the test program. Outlet particulate concentrations measured• _ for Runs 4, 8 and 9 ranged from 0.0023 to 0.0056 gr/dsef at 7 percent 0.,j with a standard deviation of p.0017 gr/dscf.
f - . ' . • ; ' : - . - " ' ;- ' 'j> . -. .i , The replicate test runs (4t 8 and 9) and Runs 3, 6 and 7 vere conducted: at the four different particulate;loadings and particle size distributionsi ' *( discussed in Section 3.2. Statistical comparison of outlet particulate
\ j concentrations determined for the replicate tests with outlet concentrations. determined for Run* 3, 6 and 7 shows that the results for all six test runs
are statistically the same. This means that variations in particulate/ . -loading and the fraction of fine partieulates did not have a statistically
i: significant effect on the performance 6f the Hydro-Sonic scrubber in termsI of the achievable outlet particulate emission concentration. The averagei; outlet particulate concentration for the six test runs (Runs 4 through 9) isj 0.0047 gr/dscf at 7 percent 0..
. Partieulate concentrations for test Runs 10, 11, 12 and 13 are: presented in Table 3-4, along with the average concentration for Runs 4,j 8 and 9. Test Runs 10, 11, and 12 were conducted at a lover pressure drop,I ' • • •• approximately 36 inches ef water, without steam injection. Test Run 13 vas' also conducted at a pressure drop of 36 inches ef vater, but with steamI injection. All of the data presented in Table 3-4 is for tests conducted atS-. . •' •• ' • . .: the lev/lev test condition. -
, .' ' , ./'.'• ' '• '' ' s ' ' •
JBS037 , 3-5
AR303868
- '
,.
I*2m2iwi13;i»m
*
i
i:• <M *
1!•i.•» u
«*
*•*M
*
j<MM
"5wM
- "1
. X
Stfm*.*
1w"ub4
t
"j««1•
!- i*mm+ii2•a
31
»•M
i
•—
**•
*
• **> »isr «*I >| M
k
•»»i «« '1 M
I•
£Is**;*<•i1•i*t«ciS
•siii
a•A • A* AJ • «
Ib AJ « M Al • m
• ! ' '
$ e» $ $ S $i
M *• . *A 2 9 2
• • •;>A> Al - •« ^ , » •^ ^ . * * M. . ' *• * * *» ib iy A ^ ik»
IA *• Al • •• Mi i i i i i« » * * 9 •
.,
I i i 1 M, « O « * • «
'
• ' . ' - .a a 5 3 2 a•» • M « • » . » •
1 J J 1 1 i1 I 1 1 ! !
' ~ '
« * •* *• « - e*.
**
/' ' ' 4- - • . '
1 . 'J} ' , ;
',. A
|
WisJ1*• .«* • ,-
53• • . '
* 4•• il]j i " •* is .* •i!iji ;
« ' *•s is • , •*•* •• »
1
AR303889
^
• .
,
*ii" MWf«* •) ,** IH
Si.w 5«i «
i|w w
i!W £
f«S *I S• «wSBA:»> U*; I5 wA M
•*'«2«••
4
'<
-'
ii• «e i
i!s i— u•b
iJ;3
•» M; *!' »t**
• *^1 ><*« fc-• *.
• >-• - i2 S-«w e u wb * «b. -
•*•-*•
i*w«H•k*in*8K
|
•. ' b
i1w*•*•
' u: **
b*&
'
Outlet
ClM/fcr)
••"c
«•
i
•e
{e>*r1*2
1u1 ts I
: *
|
**b«
1**•*0e<e
iX»
«
it***+ •V$i«*!-b» '
£f*••i
bi
-'' 'A* ; • '•!'*• ' tf* '•«- .*> * .• * ... . ,A. A. 4) M. 4> ,«
'.i '
! *" '
• ' •' ** • . ~ * '4> IA tft M* * • £
i . i• ' ".- j
« ' .»'.'• « » 'A
* 5 S * 8s
* AI 3 t- fi o-S ? s 5 s c• *> . .5 . • o« * • •*
1 ei •''••" f* • 9«• - — A, ' •• « M
e«* » * «* «* *
Al ' AJ V« IA Al ^^ -•» A* N •- »*••-••: . ? = : *• - e o -.
... i .
^ e> e* ' -* •* , A.2 AJ K M ££S , S e e o 5* • m o *
- -' ' - .
41 • tf« M •> •« Af Al m A» M. . « . . • *^ ••> ^ •
^
41 « e» » « A** * . . « .^ IA IA lA «A ' 41IA m in M n **
' " ' , ' •
1 2 22 2 S
e> • • ' •. .V * ' 'II - r• i s = 2 • t 2-5 « ><* - - ,•* *
- -
~
•
i ** w
!!M• *,sj
3-7 '
The average outlet particulate concentration for RUM 10, II, and 12was 0.0123 gr/dsef at 7 percent Oj, Comparison of the average particulateconcentrations for Runs 10, 11 and 12 with average concentration for Runs 4.a and 9 shove a statistically Ugntficant increase ia outlet particulateconcentration for the lover pressure drop, without s teats test.
The outlet partieulate concentration measured for Run 13 vas0.0091 gr/dsef at 7 percent Oj. Comparison of this concentration with cheaverage concentration for Runs 10. 11 and 12 shows a slight, butstatistically significant, reduction in the outlet partieulate emissionsi . . .vith the addition ef steaa. Thus, data presented in Table 3-4 show theexpected result of reduced partieulate emissions vith 1) higher scrubberpressure drop and 2) steaa injection.
Cas sampling parameters and conditions, including test dates, samplingtines, flue gas parameters and isokinatie calculation results are presentedin Table 3*3. Comparison of inlet and outlet gas flov rates for the ESP ona dry standard basis (dsefa) shows the lover slip streaa flow through theESP. -Inlet and outlet flov rates for the Hydro- Sonics shov acceptableagreement, when changes in oxygen content due to air in* leakage areconsidered. •
Results for isokinetie sampling rate calculations are within the QCobjective of 100 i 10 percent for all of the inlet test runs and for the ESPoutlet tests. Hydro-Sonie outlet isokinetie sampling rates are reported as arange due) to uncertainty In the actual moisture content of flue gas at thissample point. These isokinetie results are discussed in detail vith otherQA/QC result! In Section 6.0.
JBS037 . 3-S
4ft30387l
x IA a a£ — c o • &5 - - « f
** AJ 9 9
x '• .9 2M <*• 6 e
II III. 8 8
> - *» *»
£ S- £ 4> ; O
5 5
M M S aW Ml O M
'b •: i s S «sl A i A g g S g
ff»0 •* > ;w» 41
Al 0 * M 0 *• 0* AJ . • *» * • tft m" « <* IA *> ** 0•• Ai iA,m
IA AJ 41 4> M I4> K S S *i I
tO M O «J
| " ' ' ' ' "" ' ?
*• 4T At* > M
<M M *
sr j ;s s s f $ r
Z fe "*• "*^ tn • .IA
M . . » M AtAJ A. • f ff S
R AJ ' .
23 S •. * 5 | S-A; ~-SS*TS
4» 0 e* A. 4> •• «•«> AJ * . S ^» 7* * """S g « 7.^
?? »^^SS A.
~ m m • • m 0 AJ• • •* AJ St "*• <••. aZl -j ^» JL« " ~ I — " "* ™ ™ ^* *^
xp AJ AJ . . *« AJ 0 m fb0 (ft A*
Al AJ . . m v« AJ 0 m <« * *••* IA AI - .
A. » C Ib 41 AJ Wttf« AJ IA . . (M <* .
M ^ « A I A f ^ «• •• AJ o»
S» «« » * * «^ f, * . •• A *.. .. ^ AI 0 At «• y»»M «• AJ •*
- Is a s s - - - - - -88 S S««•«•»*•u, M* m «M a
= 52,,2
•
* *
ection
prec
3-9 AR303872
3.2 METALS RESULTS* •
. * Hue gas particulate/necals samples froa Runs 2 through 9 vere analyzedat Radian's Laboratories for the following aetals: arsenie, blsauch,cadaiua. chromium, copper, lead, aagnesiua, mercury, and strontiua. ,Test results for inlet samples, outlet samples, removal efficiencies andmetals-to-particulate ratios are discussed separately in the sections below,
3.2.1 Inlet to che Control Pevlee
x - . '. . '''The mass rate of the aetals aeasured at the eontrel device inlet Is
presented in Table 3-6. The aass rates .were determined froa the aetalsconcentration in the flue gas and the flue gas volumetric flow, rate. ASexpected, the aetals aass rates vary with the alx and input rate of the *various feed aaterials. 5
rMetals feed rates calculated froa feed material inputs are compared
with aeasured aetals mass rates in Table 3-7. Comparison of these dataterms of percent recovery of the aass spiked indicates the following:
Recovery of Cd, Cr. Fb, and Hg, which were present only inthe MSW ash appears to be relatively high, ranging froa 73 ee
. 100+ percent for three of the four test conditions.
Recovery of Bt. Cu, and As froa the aetals spikes wasaoderate, ranging froa 23 te 53 percent.
Recovery of strontiua (from the surrogate aetals spike) andaagnesiua (froa both the MStf ash and the surrogate aetalsspike) vas poor, ranging froa about 6 to 17 percent forstrontiua and 1 to 6 percent for magnesium.
JBS037 - . 3-10RR3Q3873
8M
2I
«*
§
i
A> M , • M «D M ^ M* « *. t •"•*: « *w ' ~ • *" ' * • 4f AJ M
V .
s'-r •.
0
28 2
c S f fc SI - i i l l 1 - 1-
i i.2 k -S e
™ ™ •» . •» i5
i«K .^ w • '0 <• , • »
2 = - « ^ »
— 0
• . S
« j ft - •* 0 » * 0 •J " 5 J . * ^ . f i ^ . J w * ~ t* s ^ - 5 2 5 ^ ~ A: fi
I*
AR30387U3-11
JI1,5
I
1
II
ill
ii
m f 11 t_ 5 — i *
2
8
• • » « * » *• ; • • ? - *- J *
IS
s s - a- . * > ar --i I a 5 s a - a j
i s j j » a - i
a s ". j • s " s •. . • * . * • « . . »* - ! • - • -
• * .s i- '-'??*.•
IS!: s s - : •
f ? : 2 •• ; 5 ? : «
. s ; . 9 1 9 1 9
-1 '*
-!'! sin: innil
I S - . , i ipijl:»?!!ijil
3l »
1,Afl303875
The poor recoveries observed for some of:the spiked metals are believed, , to be due to loss or drop out of metals in the incinerator or the duct work>-' prior to the inlet sampling point. John Zink personnel reported some
buildup of slag and particulate in the incinerator upon completion of thetesting.
3.2.2 -Quclee to cha Contra! Device
The outlet metals results are presented as concentrations adjusted to7 percent 0- in Table 3-8 and as mass rates in Table 3-9. For Runs 2 and 3.it should be noted that the aass rates are based on the slipstream flovrate.
Test results presented in Table 3-6 generally show similarcontrol system outlet concentrations for each of the target metals for alleight test runs. Any observed differences in meeels concentrations .betweentest runs and between operating conditions are not statisticallysignificant. The highest concentrations were observed for lead (30.3 to
i j 243 ug/dsca) and the lowest were for strontium (2.6 to 12 ug/dsca) andmagnesium (not detected). The single apparent outlier in the data presentedin Table 3-8 is the Run 8 outlet concentration of 312 ug/dsca for chromium.[The cause for this unreasonably high value is being investigated.]
3.2.3 Meeala Removal Efficiencies
Met*Is removal efficiencies are presented in Table 3-10. The removalefficiencies were calculated based on mass rates except for Runs 2and 3 which were calculated using inlet and outlet concentrations adjustedto 7 percent oxygen. For the scrubber, the removal efficiencies ranged from98 to 99"percent for all of the target metals, except mercury. Removalefficiencies for mercury ranged from 47 to 86 percent.
JBS057 3-13* . - •
AR303876
iI
4"S
3
a '" 4 ?
• A. A>« 0 0 A > H »
3 1 f 3 * * .*'« rf • » • » • - • *
; -^^ *-m • n
» s i a
« e J h 0W M IJA __ ^ —W* j ^* *^ m *0> 0V A • mMa s s 5 i « -a 3- 5
i
i « 2 " * . i s s *: s* a ft a «..-» s a 2
at•• ••
3
1
4
1 1S S ."! ^ *. 3 3 3 fi^j 3 s s 2 a 2 3 *
•* —
iti J15
41
nxi«• ••J!ii11
i 2 ! 1 . I J! 1 1 ! I i ! fi s J a J i l i
i ' f •'»4 « • •«.
I3-14
J I
ft > i i
• ' «
!
jwwW 41
i!
tI I* I « £ « « - * S " - 0 I _ ,
2 «
i j •'=>-^ III I " I V 5 I J
• t "t S *.• * 2 - *s - s * s « 3 a 2
"5 v S - 1 • ' •!s s s r s
"! 'S SS S- a
S . B - - 2 S 5 S | = 3
«. •« "•»" M> A> 0 ' • • M m* " ' 0 jf * Al •• AJ . O-M . A. 0 * * • |A IA0* 2 0* 0* ^ * Al *" 0
, *• M
& £*M f V f> MM 0f c ^ " * . 2 " • * * • » * • « *"• • '4> « . . * M• £"* .' « ^ * A. .• • C . • " £ •
I i J f 1u -
i?WO««» '
e »S I• *
b
C -
2 IS i
'II1 *•11I iw-i :
b __0 _»
3.15 AR303878
^
laii
303
33i
^3 w9 3
i
I
aI S .8 ft S * .ft g- £; * * *: * * »• 2 5S
$32 $ * $ s j• • • • • • • i* 3 S 3 3 * £ 2
' *~ ; ,3 S S 3 5 3 $ &* ' fi• • • • • • . . »S S 3 3 3 S * 3 2
c f^ s s - * % - r r * ** —S 5 ' S 2 225 • *2 2 2 3 2 2 2 * 2
Kctals removal efficiencies across the ESP were greater than'99 percentfor all of the target metals, agein with the exception of mercury. The dacefor mercury shows removal efficiencies ranging from essentially zero toapproximately 12 percent.
3.2.4 Raeto of Meeals to Partieulece
The metals-to-particulate ratio is defined as the mass of a particularmetal present per unit mass of particulace matter <mg of metal per g ofparciculate). The ratios are calculated by dividing .the metalsconcentration (ug/dscm) by the particulate concentration (mg/dscm). In1 • . i . -general, this ratio will be higher for smaller particles because small
> particles have higher surface area per unit mass and metals that have beenvaporized in the combustion cone will condense on a particle in proportionco the available surface area. If a control device is removing largeparticles and allowing the smaller particulate to escape, the
1 raecals-to-particulate ratio will generally increase across the controlsystem. ' - r • - '
L 7 ' . " " . - . . " . • • • • • • .^— MetalS'to-particulate ratios for inlet and outlet samples are presented
in Table 3-11. Comparison of inlet and outlet metals-co-particulace raclosfor the ESP (Runs 2 and 3) generally shows enrichment of mercury across checontrol device. The remaining metals, however, show either an increase orno significant change in metals*to*partieulate ratio across the ESP. Thedata showing little or no change in raetals-co-particuleee ratio suggest chatche ESP'was removing both large and fine particles equally well.
Comparison ef mecals-to-parctculece ratios in Table 3-11 for theHydro-Sonic scrubber shows significant enrichment of lead, mercury, copper.and bismuth, as well as slight enrichment of cadmium and chromium across checontrol device. These results suggest that particles escaping control bythe scrubber are largely the finer or sub-micron particles.
JBS057 3-17
flfi303880
^ 0
2s:i4?"
*3i
§III'S«I
S3i
i
I
1
? * s
33M133H
3 22512323
33533381s!* A ( A . » « » 4 f * .«•
£ 5 } * 3 5 S £• « 4> * . • .0*» •• 3 • HI M » . •
» 2
a 5 s s $
.33
• • .5 • 53• m • ~ oo
COCOCDoo0=
»* * e . -= >. *j«ii I? 1 11« 0 u 3 w -i i i 0 , 4! 0 3 3 £
. The objective of the particle size distribution (PSD) analy*1) determine the effect of changes in material feed rates and cooparticle size and 2) document the size distribution of particlesthe air pollution control system.'
Due to the high grain loading at the quench poc inlet samplilocation, an Andersen high capacity stack sampler (KCSS) heavy grimpaetor was used. This sampler separates particles ineo four siEach size range is typically represented by the SO percent cue po(Dp..). This value represents che aerodynamic diameter of a parthas been empirically determined co be removed with SO percenc effchic respective PSD stage at the given gas conditions (temperaturviscosity, flow, tec.). Cut points are determined for the firststages and the remaining smaller particles are removed by the bac'or fourth stage. ..
Table 3-12 shows the results of the PSD test runs conducted «inlet. Size range data for each run are presented in Table 3*12.the mass median particle diameter (HMD). The HMD is the particlehalf che parciculace mass is associated wich larger particles andmass'is associated with smaller size particles. This value is decershown in Appendix A.3.
Figures 3-1 and 1*2 present plocs of the cumulative mass fraceiparticles less than a given particle size at the low and high partloading conditions, respectively. Data presented in Figure 3-1 foparticulate loading tests show a distinct difference becween test5 (low/high) and east Run 4 (low/low). As expected, there is a lafraction of smaller particles in the low/high samples. This obserconfirmed by comparing the MMD of 23.9 microns for the lew/low conversus a MUD of 3.3 microns for the low/high condition shown in Tal
JBS057 3-19
AR3Q3882.
'-
0
I2
3|0W. -
o-30MlW
!*•ws'».
. .JhflWi
.«
IK'
M
'
£ •-i--Ii-13:
• c **J * IJif153
Is.V « 0J --5* 13i-- * it J i11-Jl2
!j• s* u= 1
4*.Ji• fc- W33
^5|!
S- •*n«•s t0 •»
•• * IA AJ .- ' 41 . '_• • • « *»Jed <j •- jt ev•» ,3. 5 % *,; •
2 s . * -*s' " > vM • • -* 0A I n m 7 5* i 4i
3 4 S 335 353 333535
-• • 0 • • 0 • • 0 • • Al • . MA i < w . 0A|* *» — . *• v . »» a .M| «• «A - < 9 w t f t . Mw« AlWiM M?4T
• •
'
« $ 55 33 S3 33.* .* *. .. . .• ' . . . . . • *i A ' * * A , * i tA t t (A * l l A i *
7 * < T 5 5 * " S * *. S T S 35 * 2 * 20 2W S K
t^J •tSJ -:aa ' s a ^31IA * * «* « • « « * . M*. M*.
1 . l4, I - «4 - -». *I - ^111 Jf Jf JI ii1 i s • . . s
' ' - ' ' - ' - ' ' . ' - I
t
*
IV
\ '
R303883
o
3AR303883 '
3-20
0.01
Figure 3-1. PSD Composite Curves Generated from Inlet Testing Data atLow Qrainloading Conditions, John Zink Research Co., Tulsa, OK (April 1989)
3.21 flR30388I*
10 10 44 LQ 10
FTgurt 3-2. PSO Composite Curves Generated from Inlet Testing Data atHigh Grafnloadlng Conditions, John Zink Research Co., Tulsa. OK (April 1969)
3-22 AR303885
As shown in Figure 3-2. particle size results for che high particulateloading tests suggest that there is no apparent difference in the proportionof smaller particles for1 the high/high test condition compared to thehigh/low test condition. However, a comparison of the average HMD of 10microns for the high/high test versus a HMD of 14.7 microns for the high/lowtest. Increased proportion of smaller particles at the high/high condition.The difference between the high fraction of fine particles condition and thelow fraction of fine particles condition is much smaller for the highpartieulate loading case (10 vs. 14.7 microns) than for the low particulaceloading case (23.9 vs. 3.5 micron*).
No data are reported in this section for Runs 1 and 3 because of a highisokinetie sampling rate for Run 1 and a leak in the final particulatefilter during Run 3. . ' '
/ 3.4 HYDROCHLORIC ACID (HC1) RESULTS -
Sampling for HC1 removal across the Hydro-Sonic scrubber system wasi '' * ' - ' . " '
V j performed during the last four test runs. Three runs were performed at.boththe inlet and outlet (10-12), and the fourth was completed at the outlet .only (13). -. ; ' , . . . - . " ,
Test results for the HC1 sampling are shown in Table 3-13. HC1i . ' » • • . .emissions were below the analytical detection limit (2 parts per million byvolume, ppfflv) for all outlet samples except Run 12. The HC1 emission racefor Run 12 was 0.056 Ib/hr. Removal efficiencies for HC1 based oncomparison of inlet and outlet samples' were >99.9 percent, >99.7 percent.and 99.2 percent for Runs 10. 11. and 12, respectively. Removalefficiencies based on carbon tetrachioride feed rates of 13.45 Ib/hr were99.5 percent for Run 12 and >99.9 percent for the remaining three runs.
I ' Sampling locations are discussed in Section 4.1 and sampling methodsj . and procedures are discussed in Section 4.2.-. • ' - . _ -" . r
I 4.1 SAMPLING LOCATIONS "v . ,] : - ' * . . " . " 'r] The gas sampling locations at the control device inlet and. outlet arei; . , .h . - * .. . discussed in the following two sections.\ : : ' . . • • . " ' . • " ' ' ' . • .r .4.1.1 ' InltfC LocationI ' • • V . . " ' - 'S ~"- • - ' - '
\ J Sampling at the inlet location was performed downstream of the flue gasf w* • . . . . , " _ ' •conditioning cower and upstream of the quench pot as shown in Figure 2-1.
. Sampling was performed In t vertical run of round duct, 12-inches in '| diameter. Two, 4.inch ports situated at right angles were provided for
1 aetals/particulate sample acquisition. The sampling plane was located'. approximately 8.7 duct diameters downstream and 2.7 diameters upstream fromi the nearest flow disturbances. Six points on each port traverse were used• for sampling as shown in Figure 4-1. .
A third 4-inch1 port was located upstream of the inletpartieulatt/aetals sampling plane and used for particle size sampleacquisition; The nearest flow disturbances were 5.8 diameters downstreamand 4.0 diameters upstream. A preliminary six-point traverse was completedat the PSD sample location as shown in Figure 4-2. The measured velocity
JBS057 " \,, ; - - / 4-1
AR303888
GondMoningTower
70 ki
ttft
33 h
1 To Own* TowerendAtMuton
wo**C3L
/ •• . • V Art
t 1a u48
Inldt PSO and Particulate/Metalt Stmpnng Locttlont,John Zink Re arch Co, Tulsa, OK (April 1969)
4-2 AR303889
Figure 4-2. Inlet PSO Sampling Configuration Using the Anderson HCSS HighGrainloading Impactor, John Zink Research Co, Tulsa. OK (April, 1989)
4.3 AR303890
.heads ranged froa 0.18" HjO to .26 HjO. Tho average velocity head of0.23" HjO was bracketed by che two center most points. It was, therefore,.decided to sample for PSD at a single point in che center of the duct.
', i. - . • * ' • '
4.1.2 Outlet Loesclona '
Different outlet sampling locations were established for each of theemission control systems tested as shown in che process flov diagrams inFigures 2*1 and 2*3. The Beltran Wet -ESP outlet gases were sampled In avertical run of 8* inch duet. Two 3 -inch ports placed at 90 degrees vereprovided for saople acquisition approximately 10 diameters downstream and3.8 diameters upstream froa the nearest flov disturbances. The gas flov toche CS? was provided by means of a slip sere am with flov being controlled byche use of a daapener valve as shown in Figure 4*3. Sampling was completedby following tha procedures outlined in EPA Method 2A for small ducts. •
The Hydro-Sonic scrubber outlet gases were sampled in a vertical run of17* inch duct, downstream of che induced draft b lover. Two 4- inch samplingports placed at 90 degrees war* located approximately 18 duct diametersdownstream and approximately 7.3 duct diameters upstream from the nearestflow disturbance. A schematic of this sampling location is shown inFigure 4-4.
4 . 2 SAM?UNG PROCEDURES ' - • _ . ' _ ' .
The sampling procedures used during this test prograa art described inchis section. Where- applicable, standard EPA mechods were employed. If nostandard swthods wu available, then scace-of-che-art methods vas applied.The flue> ga* sampling procedures are discussed in Section 4.2.1 and processsampling procedures are described la Section 4.2.2.
4.2.1
• • ' . i
Flue gas samples vert collected at che inlet and outlet* ,;emission- control systems for metals/particulate analyses. Additional flue
JBS037 ' 4-4
AR30389I
. . ,Figure 4-3. Outlet Particulate/Metals Sampling Location for the Venturi
Scrubber/Wet ESP Performance Trials. John Zink Research Co., Tulsa, OK(ApriU989) AR303892
L.*,
t Is us •« 1S8 tej» , 1t
tall
. . .Rgure 4-4. Outlet Metala/Partlculate$ Sampling Location for the Hydro-Sorto MScrubber Performance Trials, John 2nk Research CO, Tufsa, OK (April 1989)
UR303893 4-$
= • -•!, - ' • , . ' . ,
gas samples were taken at the inlet for PSD analyses. Each type of flue gassample to be collected during this test program is described in detail inche'following sections.
4,2.1.1 Parriculaee end Ketala Sampling Method
Sampling for total partlculace maccer. toxic metals and non-toxicmetals was conducted using a modified version of the Method ,5 sampling'train. This method, entitled "Methodology for the Determination of TraceMetal Emissions in Exhaust Gases from Stationary Source CombustionProcesses"<l), has been proposed for acceptance by the U.S. EPA. Aschematic of the sampling train is shown in Figure 4-5.
s ' ' , . '
The front half of the sample train is identical to a Method 5 crain.'The nozzle, probe, and filter holder were constructed using borosilicateglass. A Teflon eoaced screen'was used to support che filter. Both probeand filter holder were heated ito 248 + 25°F to prevenc moisturecondensation. A high purity glass fiber filter without organic binder *r.£with a 99.95 percent collection efficiency on 0.3 micron dioccyl phchal*:e(OOP) smoke particles was used. Sampling was conducted for 120 minutes A;che ouclec to provide for an adequate sample volume. The grain loading e:the inlet allowed a smaller volume of gas to be sampled and icill insuredcollection of tan adequate mass of particulate and metals. This resulted tr.a 60 minute test run which allowed for a PSD run to be completed in betweenport changes of partieulate/metals. In this way, all inlet and ouclec •testing wait conducted simultaneously.
• • /• •,The back half of the impinger -train was modified to capture the metal*
iin the vapor phase. The impinger* were arranged as follows:
An ice bath was used to maintain an impinger exit temperacure S 68°F.All connections were made with leak tight, ground glass joints. Theglassware used for the partieulate/toxle metals train was cleaned accordingco the procedures outlined in Table 4-1,
The flue gas was withdrawn from the stack isokinecically. Ac che endof the test run. the train was 'disassembled into three parts: 1) probe. 2}filter and connecting glassware, and 3) impingers. Each part of che sampletrain was sealed to prevent sample loss or contamination. These pares ofthe sample train were then transferred co che field laboratory where cheywere recovered as shown in Table 4-2. A final weight for each impinger vasalso obcained during recovery for moisture determination. -All samplecontainers were tared and final sample weights were determined and liquidlevels marked so that it could determined if any sample was lost duringtransport. Caps with tiny holes drilled in chem were used for sampleconcainer Number 4. The holes (»72 geuge) could allow continual vencing coavoid pressure buildup in the sample bottle. No sample leakage was observed
• i . • " . i ' iduring transport of the samples '.
•4.2.1.2 Parclelg Size Distribution . .
Particle size distribution was determined at the outlet of theconditioning tower and prior to the quench pot. An Andersen HCSS particlesizing stack sampler was used. • The Andersen HCSS is an in-staek,multistage, cascade impaetor that measures cne size distribution of particleemissions and total particulate mass concentration. A cascade impaetor'• .. _ _ — '.«&*•
I. Soak all glassware in hot, soapy water (Alconox*).
2. .Kins* with tap water (X3>., - s ' ' ' .
3. Rirue with distilled, deionized water (X3).
4. Soak in 10 percent HKO. for 8 hours.
5. Kins* with distilled, deionized water (X3).
6. Rinse with acecone to dry.
7. Allow glassware to air dry in a contamination-free environment.' " ' . • " ' ' " . -"8. Cap glassware vith paraffin film, rubber caps, or glass plugs.
X3 - Threev times.
X"/
4.10AR303897
TABLE 4-2. SAMPLE RECOVERY COMPONENTS FOR THE PARTICUUTE/METALS .THAIS
Component Cod* Item
Fronc Half
Back Half
PR-AceC Acetone rinse of probe liner, nozzle endfront half of filter housing using a nylonbristle brush with Teflon* -handle.
PR-HNO, ' 0.1N nitric acid rinses of probe liner,. nozzle, and front half of filter housing.
F Filter.
4 IR-1 Concents of imp ing en 1-3 and rinses withHN03/H202 solution.
S IR-2 50 mL aliquot from IR-1.
6 IR-3 - Contents and rinses, of impingtr number -.Rinse with 4 percent HC1.
JBS057 , ' 4"U
flR303898
. • r^classifies particles according to cheir aerodynamic dianecer as opposed to .che stokes or physical diameter. ;^
The components of che sampling train are similar to an EPA Method 5train except for che impaetor which is attached to che probe. Thep articulate -laden flu* gas was withdrawn iiokinecically through che nozzleand pulled through che impaetor. The gas was then drawn through a 1/2 -inchsteel probe to a series of condensers to remove any moisture prior coentering che pump and DCM.
The- HCSS Is structurally composed of two major components. The firstcomponent consists of the nozzle and the first two imp ace ion chambers. Thesecond component is made up of the cyclone (3rd stage) and che filterthimble housing assembly. . Uhen assembled in a straight line the .entire unitis approximately 3 feet long. Because che inlet duct was only 12 inches indiameter, the HCSS was used in a "zig-zag" configuration was shown inFigure 4-2. This would not allow for in-scack placement of both components
•r ' • ' • ' : -so an oven assembly vas constructed to house the cyclone/filcer assemblyoutside of the stack. Therefore, the first two stages received the samplegas at ftack temperatures and in the back two stages, che sample gastemperature was controlled by the oven. Ideally, sampling should have beenaccomplished at a constant temperature throughout all four stages of theimpaetor. Hovever, matching che two temperatures proved to be verydifficult. Therefore, che stack temperature was used to calculate flovthrough stages one and two and che sample gas temperatur* at the impaetorexit was used to determine flov through stages three- and four. These flowswere then used to calculate each stages repeceive cut point.
The) Impaetor outlet gas temperatur* was monitored vith a K-cypethermocouple and controlled using variable power controllers wired to cheoven heaters. Th* required sampling time vas calculated on- sice based onth* particular* loading and th* velocicy profile measured. A preliminaryvelocity measurement and stack temperatur* were taken immediately before che
'JBS057 4-12
> ' SR303899
PSO run at the single, central traverse point, and used to calculateIsokinetie sampling rates. A leak check was performed prior to the tesc onche back HCSS component (Stages 3 and 4). Anderson states that becausethere is no Inherent prenure drop across the first two stages, a leak checkneed noc be completed on these component!. A post-test leak check could no:be completed in order to maintain the integrity of the PSD sample.
After the test was completed the PSD was transferred to a recovery roomwhere the unit was allowed to cool down. Upon dismantling, the particulacecatches from Stages 1 through 3 were brushed onto tared aluminum foilpieces so that a desicated weight could be determined at a lacer dace. Thechimb le filter was also placed in a Cared aluminum foil piece for futureweighing. The first chree stages were chen rinsed separacely wich acecone,dried and weighed co a constant; value. The rinse weight was then added to"the brushed sample weight to determine total, sample weight of eachrespective stage.
4.2.1.3 HC1 Sargpllnr Meehod.. • • _ .. ,;••'. . -. • .Flue gases were sampled for HC1 deeernlnacions using che back half of
che particulace crains for Runs 10-13. The sampling train consisted of enEPA Method 5 sampling system that had been modified to allow collection ofhydrochloric acid gases in the Impinger portion (back half) of che samplingtrain as shown in Figure 4-6. The following modifications were made:
• The first and second implngers were charged with O.iNsulfurie acid Instead of water;
' ' ' ' • - ! . . ' •
e The third impinger WAS left empty for eondensat* collection;
A fourth impinger contained silica gel
JBS057 4-13
flR3039QQ
AR30390I
Sampling proceeded according co EPA Method 5 guidelines. The changes madein this sampling train were based on the draft method published by the EMSLdivision of the U.S. EPA in July. 1988 titled "Draft Method for theDetermination of HC1 Emissions from Municipal and Hazardous WasteIncinerators.^
4.2.1.4" EPA Methods 1.2.3 tnd 4
- \The scandard method* used to determine che volumeeric flow race
concent, molecular weight and exhaust gas moisture are described in thefollowing paragraphs.
The volumetric gas flow rate was determined during this program usingprocedures described in EPA Method 2. Based on this method, flows werecalculated by multiplying the cross-sectional area of the duct times cheaverage velocity of che exhaust gas.
The average exhaust gas velocity is calculated from the average pitoctube differencial pressure C P). che average flue gas temperature, weemolecular weight, and the absolute scacic pressure. Temperature anddifferencial pressure data were obtained by traversing che duec. The numberof sampling points required to measure che average gas velocity wasdetermined using the procedures outlined in EPA Method 1. The number ofsampling points and their distance from.the duct wall is a function of cheproximicy of the sampling location to ics nearest upstream and downstreamflow disturbance.
' ' i . • . . . ' ' ' • - ' .
Temperature and pressure profile daca were measured at each of chesampling points using an S-cype pitot tube and K-cype thermocouple. An oilmanometer of the proper range was used to measure the pressure differencialacross che S-cype pitot. The inlet location and the Hydro-Sonic scrubberoutlet location used 0-1" H,0 incllned/1-10" HjO vertical manometer. Thefieltran ESP outlet location required che use of an expanded scale, 0-25" ,H20inclined manometer. A bourden absolute pressure gauge was uSed to obcain
' ' ~'•"". .• ' ^ ' '' : ' • • '• ;;JBS057 4-15
AR303902
_ .
b*JTom«tric prassur* r«adlrigs at least one* a day. Th* static gas pressureat th* emission control devic* inlet and outlet vas aeasured bydisconnecting on* side of th* S-cyp* pitot and chen rotating th* pitot sothat It Is perpendicular to th* gas flov.
i , - 4 '
Th* integrated sampling technique described in EPA Method 3 was used coobtain a composite exhaust gas sample for fixed gas (0.. CO,) analysis, Asmall diaphragm pump and a stainless steal prob* vas used to extract asingle point flu* gas sample which vas collected in a Tedlar* bag. Moisturevas removed from th* gas sample by an ice bath condenser so that th* fixedgas analysis vas on a dry basis. Th* composition of th* gas sample vasdece rained using an Orsat analyzer. If more than 6 passes vere required toobtain a constant (< 0.3 percent difference, absolute) reading for eicher 0,i • *or CO., th* appropriate absorbing solution vas replaced.
The mo is cure content of th* flu* gas vas determined using th*methodology described in EPA Method 4. Based on this method, a known volumeof par ticulate- free gas is pulled through a chilled impinger train. Thequantity of condensed water is determined jravimecrically and than relatedto che volume of gas sampled to determine th* mo is Cur* content.
4.2.2 Process
Samples of th* MSt? ES? fly ash injected into th* afterburner verecollected during th* cest periods. Th* ash was scored la 55 gallon drumsprior to being injected into th* afterburner. Two grab samples of ash werecollecced. Th* ash was analyzed for mecals and also was archived.
JBS057 4-16
«R3039Q3
~
, SECTION 5.0
ANALYTICAL PROCEDURES
Laboratory activities for this program included analyses forparticulate, particulate size distribution, metals, and hydrochloric acid .(HCl). Each of the methods is described briefly in the sections below
5.1 , PARTICLE SIZE DISTRIBUTION
Recovery of che PSD sample vas performed as explained in •Section 4.2.1.2. The brushed PSD fractions were immediately placed in adesiccator upon arrival at Radian's Perimeter Park laboratory. The volumeof the acetone rinses vas measured and each rinse vas placed in a caredbeaker and allowed co evaporate in a hood. The dried beakers were alsoplaced in a desiccator so that additional moisture could be removed. So:.-types of samples were then weighed periodically until a conscane weigh: WASachieved. The brushed fraction and rinse fraction were summed andincorporated into che final calculacions as shown in Appendix A.3.i . . . - .'5i2 TOTAL PARTICULATE HATTER
The quantity of particulate matter recovered was determinedgravlmetrieally. Both the filter and che front half acetone rinses weredried and weighed to a constant weight on a five place analytical balanceThe resulting net weights were then incorporated into the grain loading (determination as shown In Appendix A.I. • /
JBS057 , 5-1 ,
AR30390I*
5.3 METALS ANALYSES
Concencracions of cadmium, chromium^ copper, bismuth, strontiua. andmagnasiua in th* toxie macals train samples were determined by Stf-846Method 6010'. which uses an inductively coupled argon plasma (ICAP)technique. Lead, arsenic and mercury concentrations in th* samples weredetermined by Stf-846 Methods 7421, 7060 and 7470, respectively. Th* lead.arsenic and mercury methods involve th* us* of an atomle absorption (AA)spectromecer. Th* digestion and analysis scheme is shown in Figures 5-1 and5-2.
5.4 HC1 ANALYSIS
Concentrations of HC1* were determined by analyzing che H.SO. impingersolutions for chloride ions (Cl") concentration. . This vas ace -plishedusing ion chromatography (1C). A multipoint calibration vas completed priorto and following injections of th* samples in order to account for any j;instrument drift which may oeeur. ~
JBS057 5-2
'Aft303905
front Mirt Same* Atcovtry frecoom
Container No. 1 Nontt'• P«iw . eVusn/MinM
i ————— —— - ————— 1
"*oo<. Cyicon* eVutn/Aint4
Demean:. Ae«ton«Wetgn . ft*u
WTK
No. 1 Acetone So. 20, IN A«nM HNOiltau
Ivtoofiit;0«uwate: We<ojn
1 'Parneuiau
imgt
Oivid» mo TwoSection**- '
; - • .
0*ttt ••en Semenwrtn HP end HNOi . . •
Microwave ar Htt , . '<lomo in •
Conventionai Oven .; - ' ,
(fraction tlKttwend
^ Aece fflnee , :
IvMOritt:
tmgj
Ot«c R«ou.in HNOi Knee '
AodifvtopHwWiNrtncAod
ft*due* Volumeto SOmiovHeaong
,
Oidevt wiW HP endMferowev of Pwf•omo n Convenum'w
to. 3 AcetoneDiscard
,
\
FrccDon 1 S
Add KM,0», 0<tft
~~^ —— r H* C m • w«»f i«tn. er Convection Oven for 2 hrt.
D*Jt« to Volume
frecnon 1A- . '
Anetyn M nnj«ffWUM •XMOt Hgbv CAP or AAS
'
AddHvdroxvwiHvdrocNono*.ScamoMCMoi
AMiyiefernutfigCoidVei
AAS •- • . '
•ndride
VJOT
—— -1 £5£
Figure 5-1. ..Digestion and Analysis Scheme f., .Mecal Train Cornponencs - Front Half
, ... /JR303906
1ft
^V H
I
1^ B
I
33
«**£ "<M 3
i!X ftl
i_t W (f s
I 2-* Hf00 4*
a £
AR303907
; SECTION 6
' , ' . • QUALITY-ASSURANCE AND QUALITY CONTROL
This seccion discusses,che qualicy assurance and qualicy concrol(QA/QC) program implemented for the project and che result* obtained fromanalyzing che QA/QC samples. The objeccive of the QA/QC prograa is brieflydiscussed in che seccion below. Summaries of QC data are preseneed in
* . . 'Sections 6.2 chrough 6.2.3. <
6.1 QA/QC PROGRAM OBJECTIVES
For any measurement efforc there exiscs a degree of uncercaincy due coChe inherent limitations of the measurement syscem. The purpose of che QAprogram was co assess che magnitude of che degree of uncercaincy in order toaccuracely assess che results obcained froa cheee measurements. To xeetchis objeccive. che cesc plan was implemeneed and EPA reference mechoc's orstaee-of-che-arc sanpllng/analytical procedures for all criticalmeasurements and QC accivicies were followed. The uleimaee goal was coproduce represencative, complete, and comparable data of known qualicy.
The QA/QC program emphasized 'strict adherence to sampling andanaiycical procedures. The custody of the samples was monitored and chedaca were carefully do'cumented. Sample macrices were spiked and spikeduplicates analyzed to assess analytical mechod accuracy and precision. The
! 'QA/QC program provided a mechod for controlling daca quality and co someextent tit* degree of uncertainty in che measurements system. The programalso provided the basis for estimating the uncertainty by providing data fordefining error limits. The QA objectives for each key parameter measuredduring the cesc program are shown in Table 6-1.
JBS057 * 6-1
AR3039Q8
Paraa*c*r (M.chod)
Total Particular,* (Toxie Metals Train) ± lltb * 10%° 90%
Flu* Gas Metals ± 13tc ± 30%d 90%
Vajce Feed M*tals t 151° ± 30%d 90%V*looity/Volua*trie Flov Rat* i «*b ± 10%b(EPA Mechod 1 and 2)
1 e The use of matrix spikes, matrix spike duplicates, reagent'• blanks, and mechod blanks.
' ' ' ' . ' . ' . ' -i Quality control procedures and results are presented in this seccion.
I J Procedures generally applicable to all sampling methods are discussed inf Seccion 6.2.1. Results from metal analyses are presented in Sections 6.2.:f co 6.2.3. • , ' ,", v - '. . ' 'I ' ' •• ' . ' • • ' . . ' .j 6.2.1 General Quality Assurance and Quality General Proceduresi.; • i . • ' - * ' L
• Estimated precision, accuracy and daca capture objectives for this tes:? program are summarized in Table 6*1. These objectives are based EPAI collaborative cest results, as veil as on Radian's recenc experience with' the specific analytical procedures.i - . . • • - . '
• Th* custody of the samples vas monitored and considered crucial inassessing sample integrity sine* samples vere collected and analysesconducted In both th* lab and che field. Samples were assigned unique
! > ' • • • • .idencificacion numbers and integrity seals vere affixed to samplecontainers. A master file of all samples collected vas maintained and an
' _ informal log of daily activities vas used co record other in/ormaclon•i ;. concerning sampling activities. '
: JBS057 . 6 - 3
AR.30.39-10
. . .
Fi*ld equipment, including pitot tubes, noz*l*s, ceoperacure measuring \Jdevices, and dry gas mecers, vere calibrated before and after field samplingas required by th* sampling procedures used. Analytical balances ver*calibrated over th* expected rang* of us* vith standard NBS vcights. Dailycalculations vere mad* vith QC veighcs previously checked vich NBS traceableweights. All calibrations vere documented on vorksheets. Instruments usedfor atonic absorption (AA) and inductively coupled argon plasma (ICAP)analyses v*r* calibraced according to th* procedures recommended in chespecific mechods. . . .
' j
All sample train glassvare, containers and tools were cleaned cominimize contamination. Cleaning procedures ver* specified in the projectplan. Sample train glassware vas capped afcer cleaning and immediatelyafter each test. A clean environment vas maintained for train assembly andrecovery.
performed according to a modified version of Method 5, entitled "Methodologyfor che D*t*raination of Trace Metal Emissions In Exhaust Gases froaStationary Source Combustion Processes." Th* results of isokinetic samplingrace calculations for th* partieulace/mecals testing are presented inTable 6-2* Isokinetic sampling rates were clearly vichin th* QC objectiveof 100.* 10 percent for all of th* inlet test runs and for th* ESP outlettests.
For th* Hydro-Sonic outlet, isokinetic sampling rates ar* reported in.Table 6-2 a* * rang* du* to uncertainty in th* actual molstur* content ofch* flu* gas at this sampl* point. The Hydro-Sonic outlet sampling tookplac* approximately 12' dovnstreaa of th* system ID fan, which vasimmediately dovnstreaa of th* Hydroponics scrubber as shown in Figure 2-1.Flu* gas entering th* fan vas saturated vith water vapor at approximately
JBSOS7 <-4
AR3039II
siM
0M
IA
S
S__ ^ .Ti ii -= X.' -^ • . .» T -^ • | • Cl
•
M *J3 !0
o-2=22
A* » 9•A — 5 S
9- -
5 = 22
«» 33
S** « 3 8^ . -i
v> u " »-. » u~ IA 5 —
•s «j a
|N2S
s s•
m 5 *>u —I * I*
?II
O ftj t • «•» <W
c IA m o — e f«Sl — -• •. »rt «n •" «A lA « »AIM ift m o
S S •: * 3 5• - * * *
« )M«A O" »"• *
0* • t•* Ift <«
«" v% ^. *i * *•
M * iA • < O «•*o m e*.
0.
« •-ft* _ _
S
fi fe ff *• * ttm * M > < * M
*SM * •* *» W W iAM O « • « M .
w -- • «•.*»••S < « w
W tw Mw 3 w« » «« » « 4 4 «M w -•• a o **« o w 5 > *-
IM9 J A « P M ^ « o^ fti »*» i . J> AJ
.
S31 8 w *: » 5 a•*«
e* e »* •* « n <*iA ft*
— M « • « —
* Mtn ft*
» e N. •• ft* M
» o • *> o e 0i * M * * ^ * M *... •• *% 0 A* *• ft m «• ft* • o>
S S SS5 g 5 ;.--•2228 »««« •* o e fr1 »
» M, *
6-5
M IM MW 5 W • • U« l» «• V * •— « «• 8 6 MM e «* > » £
AR3039I2
"
?•s =0si
sS v
fe b
•* -.• A
181°f *ind -35 inches of wacer static pressure. Calculations, using sceaatables, shov saturation moisture at cheje conditions to b* approximately59 percent wacer vapor.
• ''•
As th* flu* gas passed through th* fan. th* static pressure increasedto +6 inches of wacer (measured at sampling location) and th* gastemperature ncreascd to approximately 225°F. This increase in pressurecaused some moiscur* Co condense and occasionally vas drained through adrain plug in eh* fan housing. Although th* temperatur* increase acrossth* fan vould appear co be great enough to allow all mo is cure to revaporixe.this did not happen as on-sic* observation shoved the stack gas to b*laden vith encrained vacer droplets.
If none of the wacer vere revaporlzed, flu* gas ac eh* fan ouclec vouldhave been saturated at 181°F and +6 inches of vacer static pressure.Calculations froa sceaa tables shov saturation content at this condition t*b* approximately 50 percent water vapor by volum*. Thus, eh* .degree of jsaturation actually lies somewhere beeveen 50 and 59 percent. \—^
Using che 50 percent value, Isokinecics sampling rates for che ,Hydro-Sonic outlet rang* froa 96 to 107 percent, which is vlehln che QCobjective. Using th* 59 percent moisture content (based on saturation acche fan inlet)*, th* isokinetie sampling ratas for th* Hydro-Sonic ouclecrang* froa 114 to 124 percent, which exceeds th* QC objective.
A moIscur* content of 50 percene vas used on-sic* to decermine samplingraces because of ch* obvious presence of encrained moisture at th* stack.However, after reviewing ch* data. It vas decided to present isoklnecieresult* as * rang*, vith eh* upper value calculated froa th* total mass efmo is cur* collected in th* impinger train. Sine* only sub •micron particlesar* expected to escape control by eh* Hydro-Sonic scrubber, •parelculac*/m*cals results vould not be significantly biased by samplingnpn-isokinecically because eh* sub-micron partial** vould b* hav* as a gas. "S.
/ '"
JBS057 6-6
flR3039l3
Leak .checks vere performed before and afcer each pore change. All chePartieulaee/mecals sampling Grains met che QA criceria of a leak race lesschan or equal co 0.02 acfm. The leak race daca can be found on che fielddaca sheets in Appendix C.I.
Analysis for metals vas performed by etcher inductively coupled argonplasma (ICAP) or by Atomic absorption (M) . Matrix spike, matrix spikeduplicate, mechod blank and reagenc blank resulcs are presented inTables .6*3 Through 6*6. Previously analyzed field samples were spiked witha known concentration of Che target metals co yield che macrix spikesamples. The spikes were duplicated co provide analycical recovery andprecision information,
The QC objeccive for Che metals analysis was + 30 percene recovery.Review of -che data In Table 6-3 shows excellent matrix spike recoveries.ranging from 80 to 113 percene for all of the target metals. Agreementbetween duplicate macrix spikes also shows excellent analycical precisionfor che metals analysis.
The mechod and reagenc blanks for all samples were clean. All resultswere reported as less chan che detection limits; which were dependent on :••metal and sample volumes.
6.2.3 HC1 Samplin and
During HC1 sampling, leak checks vere also performed before and afcereach pore change. All leak checks met the leak check QC criceria. A aatrtxspike, matrix spike duplicate, reagenc blank, and macrix blank vere analysedalong vith the HCl field samples. Chloride recoveries for che cwo macrUspikes war* 103 and 99 percene, which is acceptable. The reagent blank andche laboratory blank Were both clean, with chloride concents belov chedeeeceable level of 2 mg/L.
JBS057 6-7
"AR3039U
TABLE 6-3. MATRIX SPIKE AND MATRIX SPIKE DUPLICATE RESULTSFOR METAL ANALYSES
