EPA PROJECT REPORT NO. 74-LIM-6 AIR POLLUTION EMISSION TEST OOW CHEMICAL freeport, Texas UNITED STATES ENVIRONMENTAL PROTECTION AGENCY Office of Air and Waste Management Office of Air Quality Planning and Standards Emission Measurement Branch , Research Triangle Park. North Carolina ------- A STUDY OF GASEOUS AND PARTICULATE EMISSIONS FROM LIME KILNS AT DOW CHEMICAL CORPORATIONS'S PLANT B FREEPORT, TEXAS May, 1974 environmental science and engineering, inc. ,' EPA-RTF LIBRARY ------- A STUDY OF GASEOUS AND PARTICULATE EMISSIONS FROM LIME KILNS AT DOW CHEMICAL CORPORATION PLANT B FREEPORT, TEXAS May, 1974 PREPARED AND SUBMITTED BY: ENVIRONMENTAL SCIENCE AND ENGINEERING, INC. POST OFFICE BOX 13454 GAINESVILLE, FLORIDA 32604 PN 73 Oil 042 PREPARED FOR: U.S. ENVIRONMENTAL PROTECTION AGENCY EMISSIONS MEASUREMENT BRANCH RESEARCH TRIANGLE PARK, NORTH CAROLINA environmental science and engineering, inc. ------- TABLE OF CONTENTS Page Number List of Tables ii List of Figures iii 1.0 Introduction . 1 2.0 Summary and Discussion of Results 3 3.0 Process Description and Operation 9 4.0 Location of Sampling Points 15 5.0 Sampling and Analytical Procedures 15 Appendices A - Calculated Emissions Data B - Field Data C - Analytical Procedures and Sample Calculations D - Lab Report E - Test Log F - Notes on Plant Operation G - Project Participants -i- environmental science and engineering, inc. ------- LIST OF TABLES Page Number Table 2.1 Participate Data Summary - English 4 Table 2.2 Particulate Data Summary - Metric 5 Table 2.3 Gas Data Summary 6 Table 3.1 Summary of Operating Variables 14 -11- environmental science and engineering, inc. ------- LIST OF FIGURES Page Number Figure 3.1 Precipitator Plan and Elevation 11 Figure 3.2 Stack Elevations 12 -iii- environmental science and engineering, inc. ------- 1.0 INTRODUCTION A part of the Environmental Protection Agency's ongoing work in pollu- tion abatement is concerned with setting performance standards for new and substantially modified stationary sources. One source under con- sideration at present is rotary kilns used to calcine lime. Dow Chemical Corporation operates a large chemical complex in Freeport, Texas which uses slaked lime in recovering magnesium from seawater and various other operations. There are three rotary kilns used to calcine dolomite (CaCO^ • MgCO ). The plant is capable of operating one, two or all three kilns simultan- eously. The kilns are serviced by electrostatic precipitators (ESP's) for the control of particulate emissions. The gas stream from the operating kilns i s combined prior to the ESP's and then split and ducted through two ESP's and vented to the atmosphere through two stacks, one for each ESP. During the week of April 29, 1974, Environmental Science and Engineering, Inc. (ESE) conducted three series of tests on the two outlet stacks under contract for the EPA. During the testing period two of the kilns were operating simultaneously. Each stack was monitored for particulate matter, using EPA method 5; opacity, using a modified version of EPA method 9; C02 and Og, using a modified version of EPA method 3; and CO, using EPA method 10. In addition, S02 concentrations were measured on the north stack using EPA method 6 and NOX concentrations were measured on the south stack using EPA method 7. -1- . environmental science and engineering, inc. ------- Another provision of the tests was to conduct comparison tests for participate matter at a single location in the stack using an in-stack filter and an EPA-5 train. No valid comparison data were obtained due to equinment difficulties. -2- . environmental science and engineering, inc. ------- 2.0 SUMMARY AND DISCUSSION OF RESULTS 2.1 Participate matter, Orsat, CO, S0_, NOV and opacity data are summarized £. A in Tables 2.1 - 2.3. The total average particulate emission rate was 4.9kg/hr. (10.8 Ibs/hr.) for the two stacks combined. However, this value is probably somewhat high as the first test on the north stack indicated 5.7 kg/hr. (12.6 Ibs/hr.) alone. The average for runs 2 and 3 was 3.6 kg/hr. (7.9 Ibs/hr.) for the two stacks combined, which is substantially lower. It is not known why the first run on the north stack indicated such a high emission rate. Some error may have been introduced because the person manipulating the probe misunderstood where the probe markings were to be located; instead of locating the marking at the outside of the pipe nipple, he positioned it at the inside stack wall on the first run only. This means that each point sampled was displaced from the equal area center by approximately 4 inches. It seems unlikely that this would introduce a very large error, however., it is possible that the probe tip came so close to the far stack wall that it contaminated the sample with particulate off the stack wall. Laboratory analysis of the particulate samples showed a large amount of variability in the weight fraction caught in the various parts of the sampling train (probe, filter, etc.). The average for all of the runs was as follows: Probe - 20.5%, Filter - 32%, Back Half Water - 29.5%, Back Half Acetone - 18%. This means that only 53% of the total catch was caught in the probe and filter (front half), the remaining portion being in the impingers. -3- ' environmental science and engineering, inc. ------- Table 2.1 Particulate Data Summary (English) RUN NUMBER DATE STACK DESIGNATION Volume of Gas Sampled, DSCF^ Percent Moisture By Volume Average Stack Temperature, °F Stack Volumetric Flow Rate, DSCFM Stack Volumetric Flow Rate, ACFM ' Percent Isokinetlc ,_. Process Weight Rate, Ton/Hr ^' 1 1 April 30, 1974 NORTH SOUTH 43.39 18.5 373 49.144 93.559 128 20.3 99.62 17.1 449 47,834 97,870 119 20.3 2 May 2 NORTH 131.33 17.6 361 45.911 85,521 103 20.0 , 1974 SOUTH 131.66 17.9 412 42,945 85,190 110 20.0 May 3 NORTH 133.72 18.2 366 46,318 87,951 104 20.3 3 , 1974 SOUTH 136.15 18.6 406 45,693 91,500 no 20.3 Averages NORTH SOUTH 102.81 18.1 367 47,124 89,010 111 20.2 122.48 17.9 422 45,490 91 ,520 113 20.2 PARTICIPATES - PROBE AND FILTER Gr/DSCF <3) Gr/ACF Lb/Hr Lb/Ton Feed 46.7 0.0166 0.0087 7.00 0.344 39.5 0.0061 0.0029 2.51 0.124 30.6 0.0036 0.0019 1.42 0.071 21.9 0.0025 0.0013 0.95 0.048 51.2 0.0059 0.0031 2.35 0.116 49.5 0.0056 0.0028 2.20 0.108 42.8 0.0087 0.0046 3.59 0.177 37.0 0.0047 0.0023 1.89 0.093 PARTICULATES. TOTAL Mg Gr/DSCF Gr/ACF Lb/Hr Lb/Ton Feed Percent Implnger Catch 84.2 0.0299 0.0157 12.62 0.62 44.5 63.9 0.0099 0.0047 4.06 0.20 38.2 49.4 0.0058 0.0031 2.28 0.114 38.1 126.4 0.0147 0.0074 5.45 0.272 82.7 71.2 0.0082 0.0043 3.26 0.161 28.1 106.7 0.012 0.006 4.74 0.233 53.6 68.3 0.0146 0.0077 6.05 0.298 36.9 99.0 0.0122 0.0060 4.75 0.235 58.2 Dry standard cubic feet at 70°F and 29.92 Inches Hg. These values are averages for the kilns (see Table 3.1) I.e. total feed rate for run #1 was 40.6 ton per hour. Grains per DSCF. ------- Table 2.2 Particulate Data Summary (Metric) RUN NUMBER DATE STACK DESIGNATION Volume of Gas Sampled, DSCM^' Percent Moisture by Volume Average Stack Temperature, °C Stack Volumetric Flow Rate, DSCMM Stack Volumetric Flow Rate. ACMM Percent Isokinetic /?\ Feed Rate, M-Ton/Hrv ; 1 April 30, NORTH 1.23 18.5 189 1392 2650 128 18.4 1975 SOUTH 2.82 17.1 232 1355 2772 119 18.4 2 May 2, NORTH 3.72 17.6 183 1300 2422 103 18.2 1974 SOUTH 3.73 17.9 211 1216 2413 110 18.2 3 May 3. NORTH 3.79 18.2 186 1312 2491 104 18.4 1974 SOUTH 3.86 18.6 208 1294 2591 no 18.4 Averages NORTH SOUTH 2.91 3.47 18.1 17.9 186 217 1334 1288 2521 2592 111 113 18.3 18.3 PARTICULATES - PROBE AND FILTER Mg Mg/DSCM Mg/ACM Kg/Hr Kg/M-Ton Feed 46.7 38.01 19.96 15.4 0.84 39.5 14.00 6.84 5.52 0.30 30.6 8.23 4.42 3.12 0.17 21.9 5.87 2.96 2.09 0.11 51.2 13.52 7.11 5.17 0.28 49.5 12.84 6.41 4.84 0.26 42.8 19.92 10.50 7.90 0.43 37.0 10.90 5.40 4.15 0.23 PARTICULATES - TOTAL Mg Mg/DSCM Mg/ACM Kg/Hr Kg/M-Ton Feed Percent Impinger Catch 84.2 68.53 35.99 27.76 1.51 44.5 63.9 22.65 11.07 8.93 0.48 38.2 49.4 13.28 7.13 5.02 0.28 38.1 126.4 33.90 17.08 11.99 0.66 82.7 71.2 18.80 9.90 7.17 0.39 28.1 106.7 27.67 13.81 10.43 0.57 58.2 68.3 33.54 17.67 13.32 0.73 36.9 99.0 28.07 13.99 10.45 0.57 58.2 (1) Dry standard cubic meter at 21°C and 29.92 Inches Hg (2) Metric tons per hour (2200 Ibs - 1 M-Ton) ------- Table 2.3 Gas Data Summary CT> Run 1 2 3 Location North South North South North South Flow SCMMD 1392 1355 1300 1216 1312 1294 Orsat CO Visible Data PPMD Opacity % C02 % 02 PPMD NOX PPM S02 1234 8.7 12.2 63 0 00 10.0 11.1 20 129 00 9.7 11.3 27 00 <5^' 10.4 9.7 34 96 00 8.1 10.8 250 0- O^2^ <5^ 10.0 9.8 260 96 0(2* <5( Several readings of 5 and 10 Several readings of 5 ------- After the first run, the Orsat was found to be in error when checked against ambient air. Since it rained all day Wednesday, time was spent calibrating a portable gas chromatograph to measure CO and 0 . When the integrated gas samples from the first day's testing were analyzed Wednesday afternoon, it appeared that the results were in error, i.e., the percent CO seemed low and the percent 0_ high. However, when comparable results were obtained from test 2 on Thursday, it seemed to indicate that there was possibly a leak, either in the integrated sampler or in the duct work. To find out where the leak was, a direct sample was taken from the stacks on Thursday evening; the results were the same as those obtained from the bag sample. On close examination it was discovered that ambient air was drawn into the system prior to the ESP's. The major places where this occurred were at the kiln bearing seals and the openings (approxi- mately 20 x 40 inches) where the feed rock is dropped from conveyor belts into the kiln. These locations are on the negative side of the fan and, o hence,draw ambient air into the system. The concentration of CO as measured by NDIR from injections taken from the integrated bag samples varied more than one would expect since the gases in the ductwork should have been thoroughly mixed. The average concentration for"all six runs was 109 ppmd.2 However, analysis of the gases from the last day's runs indicated 250 and 260 ppmd as compared to 63 and 20, 27 and 34 ppmd from the first two days. ' Non-dispersive infrared Parts per million dry gas volume -7- environmental science and engineering, inc. ------- Concentrations of NO were 129, 96, and 96 ppmd for the three runs. /\ Each run consisted of four grab samples, the results of which are included in Appendix A. At no time was any SC^ present in detectable amounts. This was expected since the kilns were fired with natural gas and the processed dolomite probably contained little, if any, sulfur inclusions. Visible opacity was in general 0% with a few exceptions. One observer (No. 4) indicated quite a few as less than 5% on the second and third tests. 2.2 COMPARATIVE TESTING Difficulties were experienced in getting a good comparative sample. No valid comparative data were obtained due to equipment malfunctions. EPA supplied some of the equipment to avoid potential problems in inter- facing the special probe assembly with other parts of the sampling train. The difficulties were caused primarily by a leak in the EPA's meter box. Once this was fixed it was found that the meter was still mal- functioning. Finally, the meter stopped working during the second comparative test. Since the meter was not operating properly, it would be meaningless to calculate a grain loading based on the indicated gas volume. "However, a copy of the comparative data can "be found in the appendices. -8- environmental science and engineering, inc. ------- 3.0 PROCESS DESCRIPTION AND OPERATION Limestone consists primarily of calcium carbonate or combinations of calcium and magnesium carbonate with varying amounts of impurities. Lime is a calcined or burned form of limestone, commonly divided into two basic products—quicklime and hydrated lime. Calcination expels carbon dioxide from the raw limestone, leaving calcium oxide (quicklime). With the addition of water (slaking), calcium hydroxide (hydrated lime) is forme'd. In standard chemical notation, the reations are: Calcination CaC03 Heat - CaO-+ C021 Hydration CaO + H20 =>• Ca(OH)2 The basic processes in production are: 1) quarrying the limestone raw material, 2) preparing the limestone for kilns by crushing and sizing, 3) calcining the limestone, and 4) optionally processing the quicklime further by additional crushing and sizing and then hydration. The majority of lime is produced in rotary kilns which can be fired by coal, oil, or gas. Rotary kilns have the advantages of high production per man-hour and a uniform product, but require higher capital invest- ment and have higher unit fuel costs than most vertical kilns. The Dow Chemical plant has three rotary lime kilns operating on Texas dolomitic limestone. The product quicklime is slaked to the hydrate (milk of lime) and used to precipitate magnesium hydroxide from seawater. Most of the purified magnesium hydroxide is reacted with hydrogen chlor- ide to make magnesium chloride which is converted to magnesium metal in electrolytic cells. -9- environmental science ahd engineering, inc. ------- There are three straight bore kilns, each 9 feet 6 inches by 265 feet, with a design capacity of 250 tons per day each. They "very seldom" run all three kilns, usually two are on line. The dolomite feed stone is brought into the plant sized one half to three quarters inch or three quartersto one and one half inches top size. There is no stone preheater. The kilns are fired with natural gas and the product quicklime is cooled to 200-300°F with satellite coolers. There is no quicklime storage, as all the product is fed into three rotating drum slakers where the milk of lime is produced. The three kilns are provided with an electrostatic precipitator manu- factured by the Western Precipitation Division of the Joy Manufacturing Company. The exit gas from the three kilns is cooled to 500°F by water sprays and enters a common plenum. From this plenum the gas is dis- tributed to the two chambers of the precipitator by manually operated guillotine dampers. Each chamber has three fields, thirty-five gas passages, and a plate area of 35,280 square feet. From data which Dow supplied, it can be calculated (if only two kilns are in operation) that the design velocity is 2.0 feet per second and the design residence time is 10.5 seconds. Following the precipitatbrs, the stack gases are vented to the atmosphere through 80 foot high stacks, one servicing each precipator. The complete layout -is illustrated in Figures 3.1 and 3.2. The dust collected from the precipitators is presently wasted. In the future this dust may be granulated and returned to the kiln. -10- environmental science and engineering, inc. ------- North Chambers Sample Pt. "A" ELEVATION Figure 3.1. Precipitator Plan and Elevation. -11-, ------- 8' -*! 80' 10' E 26' Ground ////// Sample Ports 90° apart •o ELECTROSTATIC PRECIPITATOR Figure 3.2 Stack Elevations -12- ------- Each chamber of the precipitator has nineteen rappers, and there is one for each distribution plate. The rappers operate in sequence, one complete cycle requiring about twenty minutes. The operation of the kilns and the electrostatic precipitator was moni- tored during the tests; process data are summarized in Table 3.1. It appears that the kilns operated normally throughout the test. On May 1 there was heavy rain all day and no sampling was possible. During the 3 to 11 shift on that day, the A field in the south precipitator chamber'began arcing badly and was therefore removed from service and grounded. Plant personnel thought that rain may have leaked into the insulator on the top of the precipitators. On the morning of May 2, the south A field was put back on line and, although there was still some arcing, it was not serious enough to significantly affect the efficiency of the precipitator. On May 3, the south A field was arcing and the voltage on the field was down from 275 V (on May 2) to 250-255 V. The south A field was there- fore removed from service and grounded before testing began. The south chamber, during the last day of testing, was therefore operating with 57% of its normal plate area. Even with this reduced collecting surface, there were no significant visible emissions. -13- environmental science and engineering, inc. ------- Table 3.1 Summary of Operating Variables. Test No. and (Date) Kiln No. 1 (4/30) 2 3 2 (5/2) 2 3 3 (5/3) Stone Feed Rate (tons/hr) Ray 0 Tube Temp. (°F) Slip Ring Temp. (°F) Feed End Temp. (°F) Fan Temp. ("F) Electrostatic Precipitator A Field Primary Current (amps) Primary Voltage (volts) B Field Primary Current (amps) Primary. Voltage (volts) C Field Primary Current (amps) Primary Voltage (volts) 19.4 21.2 2174-2192 2192-2246 1472-1490 1418-1436 1156-1166 1256 500 536 Data '136-140 260-275 185 230 152-155 205-210 19.0 21.0 2156-2210 2156-2174 1436-1460 1400-1436 1184-1202 1220-1229 500-518 527-536 135-137 250-283 183-184 232-235 151-153 200-204 18.9 21.7 2174-2210 2156-2210 1454-1490 1382-1400 1184-1202 1238-1256 518-536 536 135-137* 290-300* 182-185 235-240 151-153 202-205 *The south A field was grounded and not operating during the test. ------- 4.0 LOCATION OF SAMPLING POINTS As the sampling ports were located only 26 feet downstream from the pre- cipitator exit and 10 feet from the top of the stack, it was necessary to sample 22 points on each of two diameters through two ports 90° apart in each stack (total of 44 points in each). These points were selected according to method 1 of the Federal Register which locates each point at the center of equal area zones. The exact location of these points is included in Appendix B. As mentioned previously, for the first run on the north stack, each point was inadvertently shifted 4 inches too far into the stack. Due to the length of probe in the stack, the strain on the union of probe and heated box prevented several points near the far wall of both stacks from being sampled. 5.0 SAMPLING AND ANALYTICAL PROCEDURES 5.1 PARTICULATE SAMPLING Particulate samples were obtained using the standard EPA method 5 train without a cyclone, following the procedures specified in the December 21, 1971 Federal Register. In addition, .the impinger contents were obtained and analyzed according to the appropriate procedures specified in the proposed Method 5 in the August 17, 1971 Federal Register; except that,the organic extraction was not performed. Prior to the first run, a preliminary moisture run was made in order to obtain a moisture fraction for setting the nomograph. Subsequently, for purposes of setting the nomograph, the moisture fraction was assumed to be 18-19 percent as found from the first particulate runs. -15- environmental science and engineering, inc. ------- Gas temperatures were measured before and during each test with bimetallic dial thermometers accurate to ± 5°F. These measurements were made at a single point in the stack instead of at each sampling point. There appeared to be very little temperature gradient across the stack,making a one point temperature sample sufficient. In addition, before each run except the third run, a preliminary velocity traverse was made on one diameter of each stack for the pur- pose of balancing the flow rates. If necessary, damper, adjustments were made under the direction of the EPA project engineer. At the end of each run, the particulate sampling trains were moved to a nearby room for clean-up. The samples obtained were transferred to D acid-washed glass storage containers with Teflon seals for subsequent laboratory analysis. A description of the analysis procedures is included in Appendix C. 5.2 S02 SAMPLING METHODS Since it was requested by the EPA project officer that SO sampling extend the full period during which particulate samples were obtained, it was necessary to add a dry impinger between the isopropanol bubbler and first peroxide impinger to trap any diluted isopropanol carryover to prevent sample contamination. This was the only deviation from method 6, as specified in the December 21, 1971 Federal Register. 5.3 N0v SAMPLING METHODS A Four NOX grab samples were taken,foil owing Method 7 as specified in the December 21, 1971 Federal Register during each particulate run on the -16- environmental science and engineering, inc. ------- south stack; the arithmetic mean of these four samples is reported as the result for that run. These samples were approximately equally spaced in time during the run in order to obtain a representative average. 5.4 INTEGRATED BAG SAMPLES An integrated bag sample was obtained from each stack during the period of the particul ate runs,foil owing Method 3 as specified in the December 21, 1971 Federal Register. At the end of each run the bag's contents were analyzed by NDIR for CO, and for COg and 0~ by gas chromatography utilizing a thermal detector. A copy of the analytical procedures is included in Appendix C along with the calibration procedure for the GC and sample calculations show- ing corrections for CO? in the CO concentrations, according to Method 10 as specified in the March 8, 1974 Federal Register. 5.5 VISUAL SMOKE OBSERVATIONS During the first particulate tests, two visual observers were assigned to read each stack. Due to manpower requirements, this number was re- duced to three observers on the second particulate test and two observers on the third. Since the opacity did not differ much from 0 at any time, it was possible to have the observers double up and read both stacks simultaneously for the last two tests. The observers followed the guidelines set forth in Method 9}as specified in the December 21, 1971 Federal Register. 5.6 COMPARATIVE TESTING Two probes were attached to a common pi tot tube which allowed isokinetic sampling at approximately the same point in the stack. The in-stack -17- environmental science and engineering, inc. ------- filter train differed from the method 5 train in that an additional in-stack filter holder was placed directly behind the nozzle. The re- mainder of the sampling equipment remained unchanged. Equipment diffi- culties prevented obtaining valid results. -18- environmental science and engineering, inc. ------- |