T IC ' 065. JAN 3 o 9!I R.J. ROSSI E.I. WAN AFWAL-TR VOLUME VII PRODUCTION OF JET FUELS FROM COAL DERIVED LIQUIDS

Transcription

1 AFWAL-TR VOLUME VII PRODUCTION OF JET FUELS FROM COAL DERIVED LIQUIDS I C S VOLUME VII -- GPGP JET FUELS PRODUCTION PROGRAM -- EVALUATION OF TECHNICAL UNCERTAINTIES FOR PRODUCING JET FUELS FROM LIQUID BY-PRODUCTS OF THE GREAT PLAINS GASIFICATION PLANT M.D. FRASER R.J. ROSSI E.I. WAN S BURNS AND ROE SERVICES CORPORATION SCIENCE APPLICATIONS INTERNATIONAL CORPORATION PITTSBURGH, PA JANUARY 1989 INTERIM REPORT FOR THE PERIOD OCTOBER 2, SEPTEMBER 30, 1988 APPROVED FOR PUBLIC RELEASE; DISTRIBUTION IS UNLIMITED T IC JAN 3 9!I AERO PROPULSION AND POWER LABORATORY JA 09 AIR FORCE WRIGHT AERONAUTICAL LABORATORIES"11 AIR FORCE SYSTEMS COMMAND WRIGHT-PATTERSON AIR FORCE BASE, OHIO ' 065

2 UNCLASSIFIED SECURITY CLASSIFICATION OF THIS PAGE la. REPORT SECURITY CLASSIFICATION Unclassified REPORT DOCUMENTATION PAGE lb RESTRICTIVE MARKINGS Nne 2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION /AVAILABILITY OF REPORT N/A Apprved fr public release; 2b. DECLASSIFICATION/DOWNGRADING SCHEDULE distributin is unlimited. /A distributin sunlimited. 4. PERFORMING ORGANIZATION REPORT NUMBER(S) S. MONITORING ORGANIZATION REPORT NUMBER(S) OM#N. OO"018 N/A AFWAL-TR , Vl VII. ME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL (if appikabl) 7a. AME OF MqOIT I G ORGANIZATION ir Frce wright Aernautical Labratries Burns and Re Services Crp. N/A Aer Prpulsin Labratry (AFWAL/POSF) 6c. ADDRESS (City, State, and ZIP Cde) 7b. ADDRESS (City, State, and ZIP Cde) P. 0. Bx Wright-Pattersn AFB OH Pittsburgh PA S.. NAME OF FUNDING/SPONSORING b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER ORGANIZATION 0(If appika We) Ift. ADDRESS (City, State, and ZIP Cda) FY N SOURCE OF FUNDING NUMBERS PROGRAM ELEMENT NO. PROJECT NO. TASK NO WORK UNIT CCESSION NO F TITLE (Incude Security Classifcatin) Prductin f Jet Fuels frm Cal Derived Liquids, Vl VII - GPGP Jet Fuels Prductin Prgram - Evaluatin f Technical Uncertainties fr Prducing Jet Fuels frm Liquid By-Prducts f the Great Plains Gasificatin Plant 12. PERSONAL AUTHOR(S) M. D. Fraser. R. J. Rssi. and E. I. Wan 13.. TYPE OF REPORT 13b. TIME COVERED 14. DATE OF REPORT (Year, Mnth, Oay) IS. PAGE COUNT Interim FROMIjR7 TO January SUPPLEMENTARY NOTATION 17. COSATI CODES 18. SUBJECT TERMS (Cntinue n revern I necesry and ldentify by blck number) FIELD GROUP SUB-GROUP urbine fuel, JP-4, JP-8, Great Plains Gasificatin Plant, 1 21 Q7 tar ili crude phenls, naphtha, jet fuel, by-prduct na n nr prductin v,1-, l. ABSTRACT (Cntinue n reverse if necessary and Iden by blck number) In September 1986, the Fuels Branch f the Aer Prpulsin Labratry at Wright-Pattersn Air Frce Base, Ohi, cmmenced an investigatin f the ptential f jet fuel prductin frm the liquid by-prduct streams prduced by the gasificatin f lignite at the Great Plains Gasificatin Plant (GPGP) in Beulah, Nrth Dakta. Funding has been prvided by the Department f Energy (DOE) Pittsburgh Energy Technlgy Center (PETC) t administer the experimental prtin f this effrt. This dcument reprts the results. f the effrt by Burns and Re Services Crpratin/Science Applicatins Internatinal Creratin (BRSC/SAIC) t analyze GPGP peratins and develp crrelatins fr the liquid _'by-prducts and plant perating factrs such as cal feed rate and cal characteristics.-'this reprt discusses thse parameters which are predictrs f by-prduct prductin as well as the variability f the by-prduct prductin. 20. DISTRIBUTION IAVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION 0 UNCLASSIFIED/UNLIMITED [ SAME AS RPT. OTIC D] USERS UNCLASSIFIED 22s. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (Include Area Cde) 22c. OFFICE SYMBOL William Harrisn (513) AFWALIPOSF DO Frm 1473, JUN 86 Prevlus editins are bslete. SECURITY CLASSIFICATION OF THIS PAGE UNCLASSIFIED

3 FOREWORD In September 1986, the Fuels Branch f the Aer Prpulsin Labratry at Wright-Pattersn Air Frce Base, Ohi cmmenced an investigatin f the ptential f prductin f jet fuel frm the liquid by-prduct streams prduced by the gasificatin f lignite at the Great Plains Gasificatin Plant (GPGP) in Beulah, Nrth Dakta. Funding was prvided t the Department f Energy (DOE) Pittsburgh Energy Technlgy Center (PETC) t administer the experimental prtin f this effrt. This reprt details the effrt f Burns and Re Services Crpratin/Science Applicatins Internatinal Crpratin (BRSC/SAIC), wh, as a cntractr f DOE (DOE Cntract N DE-AC22-84PC72571), analyzed plant perating data and develped crrelatins f the liquid by-prduct yields frm the GPGP. DOE/PETC was funded thrugh Military Interdepartmental Purchase Request (MIPR) FY N0657. Mr. William E. Harrisn III was the Air Frce Prgram Manager and Mr. Gary Stiegel was the DOE/PETC Prgram Manager. Aceess1n Fr NTIS 3RPA&I ETTZ T6?,, r,.".. < I _.tlity Cde9... Avall. aii-d/r" DI!7t Seci ]. iii/iv

4 TABLE OF CONTENTS Sectin Pg 1.0 EXECUTIVE SUMMARY GOALS AND OBJECTIVES APPROACH PRODUCTION OF LIQUIDS IN GASIFICATION SYSTEMS INITIAL CORRELATION WORK Data Base Develpment Preliminary Identificatin f Crrelating Variables Crrelatins with One Independent Variable Crrelatin with Tw Independent Variables FINAL CORRELATIONS Data Base Develpment Identificatin f Crrelating Variables... 1is 6.3 Crrelatin with One Independent Variable SNG Prductin Tar Oil Prductin Naphtha Prductin Phenl Prductin Crrelatins with Tw Independent Variables Tar Oil Prductin Phenl Prductin DISCUSSION OF RESULTS CONCLUSIONS RECOMMENDATIONS REFERENCES APPENDIX... A-1 V

5 LIST OF FIGURES Figures Page 1 Chemical Reactins and Temperatures in a Mving Bed Gasifier SNG Prductin vs. Maf Cal Feed Rate -- Cmbined Large Data Base SNG Prductin vs. Maf Cal Feed Rate Data Tar Oil Prductin vs. Maf Cal Feed Rate -- Cmbined Large Data Base Tar Oil Prductin vs. Maf Cal Feed Rate Data Naphtha Prductin vs. Maf Cal Feed Rate -- Cmbined Large Data Base Naphtha Prductins Maf Cal Feed Rate Data Phenl Prductin vs. Maf Cal Feed Rate -- Cmbined Large Data Base Phenl Prduct vs. Maf Cal Feed Rate Data Tar Oil Prductin vs. Maf Cal Feed Rate -- Cmbined Small Data Base Tar Oil Per Tn vs. Atmic H/C Rati -- Cmbined Small Data Base Phenl Prductin vs. Maf Cal Feed Rate -- Cmbined Small Data Base Phenl Per Tn vs. Atmic H/C Rati -- Cmbined Small Data Base A-1 SNG Prductin vs. Maf Cal Feed Data... A-17 A-2 Tar Oil Prductin vs. Maf Cal Feed Data... A-18 A-3 Naphtha Prductin vs. Maf Cal Feed Data... A-19 A-4 Phenl Prductin vs. Maf Cal Feed Data... A-20 A-5 Tar Oil Prductin vs. Maf Cal Feed Rate and Parameter H/C Rati Data... A-21 A-6 Tar Oil Prductin vs. Atmic H/C Data... A-22 vi

6 LIST OF FIQRES (cncluded) Figures Page A-7 Phenl Prductin vs. Maf Cal Feed Data. A-23 A-8 Phenl Prductin vs. Atmic H/C Data... A-24 vii

7 Tables LIST OF TABLES 1 Parameters f Final Crrelatins f Plant Prducts vs. the Maf Cal Feed Rate Characteristics f Data Used fr Final Crrelatin (Table 1) Parameters f Final Crrelatins f Plant Prducts with Tw Independent Variables Characteristics f Data Used fr Final Crrelatins (Table 3) A-1 Preliminary Crrelatins f Tar Oil Prductin in GPGP Data A-2 A-2 Characteristics f Data Used fr Preliminary Tar Oil Prductin Crrelatin (Table A-i) Data... A-3 A-3 Preliminary Crrelatin with One Independent Variables t Identify Imprtant Variables Data... A-4 A-4 Characteristics f Data Used fr Preliminary Crrelatins (Table A-3) Data... A-5 A-5 Parameters f Initial Crrelatins f Plant Prducts versus the Maf Cal Feed Rate Data... A-6 A-6 Characteristics f Data Used fr Initial Crrelatins (Table A-5) Data... A-7 A-7 Crrelatins with Tw Independent Variables Data... A-8 A-8 Characteristics f Data Used fr Crrelatin (Table A-7) Data... A-9 Page A-9 Characteristics f the Crrelatin f SNG Prductin vs. Cal (Maf) Feed Rate Data... A-10 A-10 Characteristics f the Crrelatin f Tar Oil Prductin vs. Cal (maf) Feed Rate -- Cmbined Large Data Base... A-11 A-11 Characteristics f the Crrelatin f Naphtha Prductin vs. Cal (maf) Feed Rate -- Cmbined Large Data Base... A-12 A-12 Characteristics f the Crrelatin f Phenl Prductin vs. Cal (maf) Feed Rate -- Cmbined Large Data Base... A-13 viii

8 Tables LIST OF TABLES (cncluded) Pa e A-13 Characteristics f the Crrelatin f Tar Oil Prductin -- Cmbined Small Data Base... A-14 A-14 Characteristics f the Crrelatin f Tar Oil Prductin Per Tn vs. Atmic H/C Rati -- Cmbined Small Data Base... A-15 A-15 Characteristics f the Crrelatin f Phenl Prductin -- Cmbined Small Data Base... A-16 ix

9 1.0 EXECUTIVE SUMMARY The Great Plains Gasificatin Plant (GPGP) represents the first cmmercial Synfuels Plant in the United States. perated fr the Department f Energy (DOE) The plant is currently being by the ANG Cal Gasificatin Cmpany t prduce synthetic natural gas (SNG) frm Nrth Dakta lignite. Althugh the plant was designed t nminally prduce MMSCF f SNG per day, ANG has mre recently perated at an average rate appraching 150 MMSCF SNG per day. In additin t SNG, the GPGP als prduces three liquid hydrcarbn by-prduct streams: Rectisl naphtha, crude phenl, and tar il. While these by-prduct liquids have a ptential market value as transprtatin fuels and/r chemicals, they are currently burned in the plants bilers as fuels. In 1986, the Department f Defense and DOE kicked-ff a prgram aimed at assessing the feasibility f prducing specificatin jet fuels frm the GPGP by-prducts, and in particular, the tar il. Because it is knwn that the naphtha and crude phenl streams cntain ptentially valuable chemicals, the Jet Fuels Prductin Prgram is als investigating the feasibility f recvering these chemicals as c-prducts and selling them t subsidize the prductin f jet fuels. It is imprtant t assess the variability f the by-prduct yield during plant peratins in rder t be able t predict by-prduct yields as a functin f future plant perating cnditins and thus assist the design and ecnmic assessment f the Jet Fuels Prductin Facility. ANG has cnsistently mnitred by-prduct prductin rates as well as key perating parameters since The specific purpse f the wrk discussed in this reprt was t analyze plant perating data and develp crrelatins f the liquid by-prduct yields frm the GPGP. Imprtant variables cntrlling the prductin f these by-prducts were t be identified and the results utilized t estimate the technical uncertainties assciated with the prductin f such feedstcks. This study cncludes that the prductin f SNG, naphtha, crude phenl, and tar il is mst strngly influenced by the misture and ash free 1

10 (MAF) cal feed rate. It was als determined that cal characteristics such as vlatile matter, H/C rati, sulfur, ash, and Na20 cntent, and ash fusin temperature did nt cntribute t a better crrelatin fr SNG prductin. With the exceptin f the atmic H/C rati f the feed cal, this was als true fr the tar il, crude phenl, and naphtha. In the case f the tar il, yields decreased with increased cal H/C rati, while phenl increased slightly with an increase in cal H/C rati. This study als cncludes that it is nt currently pssible t explain all the variatin in prductin n the basis f available data. One particular limitatin regards the restricted nature f key gasifier perating parameters such as xygen-t-steam ratis, temperatures, pressures, and slids residence time. While these data may have aided develpment f mre cmprehensive by-prduct crrelatins, they are prprietary and nt generally available. Lastly, independent analysis and cmparisn f the 1987 by-prduct yields t the yields indicates that the 1987 data n prductin d nt appear t be significantly different than the data. As a result f this effrt t analyze liquid by-prduct prductin as a functin f perating parameters, several prpsed. recmmendatins have been These include further study and analysis f basic prcesses gverning liquids prductin in the gasifier; analyses f dwnstream peratins that pssibly influence liquid yield and quality; and detailed investigatin f the relatinship between cal feed rate, gasifier peratin, cal characteristics, by-prduct prductin and byprduct characteristics. Finally, it is als suggested that the crrelatins develped in this study be utilized in the technical and ecnmics feasibility studies develping an ptimum design fr the jet fuel prductin facilities. 2.0 GOALS AND OBJECTIVES As part f the Jet Fuels Prductin Prgram, the Department f Defense (DOD), and Department f Energy (DOE) requested BRSC/SAIC t review and 2

11 analy~e GPGP by-prduct prductin data and cal characterizatin data fr the perid frm the secnd quarter f 1985 thrugh January The bjective f this wrk was t develp useful crrelatins that culd be used t predict the quantities f by-prducts prduced by the plant. Imprtant variables that culd perhaps be used t cntrl the amunts f the by-prducts were t be identified. Frm the statistical analysis f the data, technical uncertainties assciated with the prductin f the feedstcks fr jet fuel manufacture culd be identified, quantified, and cnsidered in the design f the jet fuel facilities. 3.0 APPROACH The scpe f the wrk dne in the data crrelatin area included the develpment f crrelatins fr the tar il prductin and the phenl prductin. Crrelatins were als develped fr SNG prductin and naphtha since they might prvide infrmatin regarding imprtant cntrlling variables. Crrelatins fr these variables were als needed t predict the pssible effects f changing parameter values. The data base used cmprised the data cntained in the GPGP prgress reprts. These reprts include data n prduct yields and feed cal characteristics, but d nt include data n gasifier perating cnditins, which are nt generally available. The crrelatin wrk was dne in tw phases: an initial preliminary phase t identify imprtant variables and develp crrelatins, and a secndary phase t test the crrelatins with the latest, best data. Each phase is described in the reprt, including the develpment f the data base, identificatin f crrelating variables, develpment f the crrelatins, and the results. 4.0 PRODUCTION OF LIQUIDS IN GASIFICATION SYSTEMS Since the main bjective f perating a cal gasifier is usually t prduce gas, little attentin has been paid t the prductin f liquid by-prducts in such systems. The usual bjective in dealing with 3

12 liquid by-prducts is t minimize r eliminate them, In sme gasifier designs essentially n liquids are prduced, and even in the case f the Lurgi system, the literature suggests that tar and il can be recycled t extinctin in the gasifier. Thus, data n liquid by-prduct prductin in cal gasifiers have nt been cllected and analyzed in detail t determine the underlying variables and their effects n liquid prductin. In the case f the GPGP Lurgi gasifiers, the perating data are prprietary and are nt currently available; nly a few islated example runs are reprted in the literature t supprt a general descriptin f the system. Hwever, the literature (1,2,3,4,5,6) des make the pint that in a gasifier cal is heated thrugh the temperature regime where pyrlytic reactins ccur, and a knwledge f the kinetics f cal pyrlysis is essential in predicting the yield and prducts distributin in a cal gasifier. In the case f a mving bed gasifier such as the Lurgi, the reactr is viewed as cmprising fur reactin znes. These znes are illustrated in Figure 1, which als shws typical temperature ranges and assciated reactins within the znes. The reactin gases travel up thrugh the bed cuntercurrent t the cal mving slwly dwnward. and then enter the gasificatin zne. The gases are heated in the cmbustin zne Gases leaving the gasificatin zne, where reductin reactins ccur, are substantially cler because f the endthermicity f these reactins. Gas entering the devlatilizatin zne then gives up sensible heat t heat the cal t pyrlysis temperatures and t prduce the heat required fr pyrlysis reactins. Cled gas frm this exchange, nw carrying the prducts f pyrlysis, rises thrugh the incming fresh cal t preheat and dry it. T sme extent, heavy pyrlysis prducts are believed t cndense ut in the drying zne t aid in heat transfer. The result is a recycle f heavier hydrcarbns t devlatilizatin, there t be revlatilized and partially pyrlyzed. Cled prduct gases leave the bed at the dew pint temperature (highly variable, but nminally abut OF). 4

13 C, 0 1~ uj IL C0 0wEu CC 0 (3 z 2 55

14 .4 Slid cal passes successively dwnward thrugh the znes, being heated all the way, first fr drying, next devlatilizatin fllwed by char reductin, and then xidatin; finally, the residues are cled by incming steam-xygen befre the ash is discharged via the grate. the devlatilizatin zne gaseus prducts are frmed as weak bnds in the cal structure are ruptured, and these gaseus prducts are remved frm the gasifier with the gases frm the cmbustin and gasificatin znes. The temperature f the exiting mixture is nt high enugh fr the rapid thermal decmpsitin f the vlatile matter. In The vlatile matter by-prducts can be divided int water-sluble cmpunds, such as phenls, ammnia, and fatty acids, and water-insluble cmpnents, such as tars, ils, and naphtha. Frm this mdel f the reactr and the reactin znes, it appears that the tar and tar il are frmed in the devlatilizatin zne as the result f primary pyrlytic reactins and secndary cracking r plymerizatin reactins f the primary vlatiles. Infrmatin n crrelatin f yields with feed material prperties and imprtant perating variables can then perhaps be fund frm the extensive literature n cal pyrlysis. Sme f this literature was searched, and sme imprtant variables were identified. Where data were available n these variables, crrelatins were attempted, as discussed in the next sectins. Identificatin f parameters that may be imprtant in cntrlling the prductin f tar, tar il, and phenl in the Lurgi system wuld als be useful fr identifying types f data that may be useful t cllect frm the GPGP. Hwever, the main bjective at GPGP is t prduce gas s that there may be relatively slight pprtunity t manipulate certain perating parameters t determine their effect n liquid by-prduct prductin. Imprtant parameters fund t influence the yield and distributin f liquid prducts frm gasificatin/pyrlysis reactins include the fllwing: Temperature 6

15 Rate f heating (slw vs. fast) Pressure Vlatile matter cntent f feed cal Hydrgen/carbn rati f feed cal Particle size f feed cal Data are nt available frm the GPGP n the primary prcess-cntrl variables f xygen rate and the steam-t-xygen rati, which wuld cntrl the temperature and rate f heating. Data n pressure are similarly nt available. Tw f the ther parameters shwn were used as crrelating variables. Data shuld perhaps be btained in the future n the particle size f the feed cal n a regular basis. Because the available data n liquid by-prduct prductin are fr the plant utput rather than the reactr utput, a number f ther perating variables shuld be examined, such as the prprtin f the gas ging thrugh the shift cnverter. affects the quality and quantity f the by-prducts. Hydrgenatin within this unit 5.0 INITIAL CORRELATION WORK The initial preliminary wrk which was dne t develp data crrelatins was based n data frm GPGP frm 1985 and 1986, and was aimed at identifying imprtant variables. The fllwing discussin summarizes the results f this phase f the wrk. 5.1 Data Base Develpment The basic GPGP data base that was available fr the initial wrk phase cnsisted f the daily data cntained in the quarterly prgress reprts frm the Secnd Quarter 1985 thrugh the Third Quarter 1986, and additinal daily data by mnth cntained n three PC diskettes frm January 1986 thrugh January Depending n the variables f interest, nt all types f data were available n a daily basis. Fr example, feed cal cmpsitin 7

16 data, including vlatile matter and ultimate analysis, were available fr nly ne ut f every three r fur days. few data pints were either missing r bviusly in errr. In additin, a Thus, depending n the types f crrelatins t be develped and cmpared, the data base was examined and apprpriate data sets were selected. The data were inserted int a spreadsheet frmat (Ltus TM, Release la) n a PC fr analysis, manipulatin, and presentatin. The data that were available n a daily basis, and entered int a data base, termed the large data base, included: Date Cal delivered Percent misture, a.r. (as received) Percent ash, dry Percent sulfur, dry SNG prductin (MIS,Net) Tar il prductin Naphtha prductin Phenl prductin This data base included apprximately 530 data sets. Data n feed cal cmpsitin were nt available n a daily basis as they were btained nly nce every three r fur days. additin, nt all data f this type were available n the same day. Fr the purpses f cmparing crrelatins based n different variables, it was desired t btain a cherent data base with all the variables in each data set that might be f interest. Cnsequently, the data sets were screened t make sure that all variables were represented in each data set. In A data base was thus develped, termed the small data base, which included the fllwing variables in additin t thse listed abve: 8

17 Percent vlatile matter Percent carbn, maf (misture, ash-free) Percent hydrgen, maf (misture, ash-free) Average number f gasifiers perating This data base included apprximately 150 data sets. In additin t using primary data variables directly as crrelating variables, crrelating variables such as atmic H/C rati, maf cal feed rate, and liquid prduced/tn cal were derived frm the primary data variables. 5.2 Preliminary Identificatin f Crrelating Variables The initial effrt in the preliminary phase was aimed at crrelating the tar il prductin with a number f the primary data variables cntained in the large data base. Because basic data n gasifier perating cnditins were nt available, the SNG prductin rate was chsen, als, as crrelating variable fr tar il prductin; it was thught that SNG prductin might be a variable representative f gasificatin reactin severity. were develped, using standard statistical methds (7), Crrelatins fr tar il prductin as the dependent variable and the fllwing sets f independent variables: SNG prductin Maf cal feed rate Maf cal feed rate and SNG prductin Maf cal feed rate and percent sulfur Maf cal feed rate and percent ash The results are summarized in Table A-1 in the appendix. Table A-2 lists sme characteristics f the data. The mst significant crrelatin f the tar il prductin with ne variable was with the maf cal feed rate, rather than SNG 9

18 prductin, as shwn by a slightly higher sum f squares remved by the crrelatin. Including bth f these variables as independent variables imprved the crrelatin slightly, althugh SNG prductin is bviusly highly crrelated with maf cal feed rate. Including highly crrelated independent variables in a crrelatin usually des nt imprve the crrelatin significantly. A very small imprvement in the crrelatin was btained by including either the percent sulfur (maf) r the percent ash (maf) f the feed cal alng with the maf cal feed rate as independent variables. This small imprvement was nt significantly, hwever, s that fr all practical purpses, the best crrelatin f the tar il prductin frm the list abve was with the maf cal feed rate as the ne independent variable. T try t prduce a better crrelatin f tar il prductin and maf cal feed rate, three-day rlling averages fr bth quantities were used t smth ut daily variatins caused by the way tar il prductin is measured. The initial data were smthed cnsiderably, as shwn by the initial sum f the squares f deviatins. The resulting least-squares regressin was cnsiderably better in terms f the fractin f the sum f the squares f deviatins remved by crrelatin, and the cnfidence limits f the regressin wuld be smaller. Hwever, the crrelatin itself was nt greatly different frm the ne develped by wrking with the riginal daily data. It was cncluded that it was nt necessary r useful t first smth the daily data by calculating rlling averages. T prceed further with the identificatin f imprtant crrelating variables, a number f different crrelatins with ne independent variable were calculated frm the small data base, which included data n feed cal characteristics. The dependent variables included SNG prductin, tar il prductin, naphtha prductin, and phenl prductin. The independent variables 10

19 included maf cal feed rate, percent vlatile matter in the feed cal, atmic H/C rati f the feed cal, vlatile matter feed rate, and number f gasifiers perating. This latter variable was lked at because it is prprtinal t reactr vlume and thus can perhaps represent the effects f residence time n the prgress f the reactins and prduct yields. The bjective f this analysis was t identify and rank crrelating variables fr each dependent variable in terms f the sum f squares f deviatins remved by the crrelatin. The imprtant variables fund frm this analysis culd then be cmbined t develp the best verall crrelatin fr each dependent variable. The numerical results are shwn in Table A-3 in the appendix. The imprtant cnclusins are discussed belw. 1. The maf cal feed rate was the mst imprtant crrelating variable fr each f the fur dependent variables. 2. The vlatile matter feed rate was in sme cases equal in imprtant t the maf cal feed rate, but as these tw variables are highly crrelated themselves, it wuld nt be useful t include bth f them as crrelating variables. 3. Percent vlatile matter did nt help in crrelating any f the dependent variables. 4. Atmic H/C rati appeared t be a useful crrelating variable fr tar il prductin and phenl prductin. 5. Number f gasifiers perating was a significant crrelating variable fr SNG prductin and tar il prductin. Hwever, this variable is highly crrelated with cal feed rate. 11

20 These results cmpsed the basis fr selecting which crrelatins shuld be develped. The crrelatins that were develped in this initial phase are discussed in the sectins belw. 5.3 Crrelatins with One Independent Variable Based n the effect discussed abve t evaluate the relative significance f the pssible independent crrelating variables, it was cncluded that the mst imprtant variable by far was the cal (maf) feed rate. Cnsequently, crrelatins were develped fr the fur plant prducts with this single crrelating variable. These crrelatins were develped frm the large data base described abve cntaining 529 pints. Fr SNG prductin and naphtha prductin, these crrelatins are the nly useful nes which culd be develped frm the -.ailable data. Fr prductin f tar il and phenl, these ne-variable crrelatins may be useful in the absence f data n the feed cal characteristics, althugh including these latter data in the crrelatin results in a mre accurate estimatin, as discussed belw. Table A-5 in the Appendix lists the imprtant parameters f these fur crrelatins f the plant prducts with the maf cal feed rate with the characteristics f the data used fr the crrelatins shwn in Table A-6. The data are pltted alng with the crrelatins in Figures A-1 (SNG prductin), A-2 (tar il prductin), A-3 (naphtha prductin), and A-4 (phenl prductin). 5.4 Crrelatins with Tw Independent Variables As explained in Sectin 5.2, it was fund that crrelatins fr tar il prductin and phenl prductin culd be imprved by including the feed cal atmic H/C rati as an additinal crrelating variable alng with cal (maf) feed rate. Cnsequently, crrelatins were develped frm the smaller data base fr tar il prductin and phenl prductin with these tw variables as the 12

21 independent crrelating variables. In additin, t illustrate the dependence f the prductin f tar il and phenl n atmic H/C rati, crrelatins were develped fr the tar il prductin and phenl prductin per tn f cal (maf) as a functin f atmic H/C rati. Althugh these latter crrelatins may be linear in nly ne independent variable, the atmic H/C rati, the crrelatins are actually a functin f tw independent variables, the dependence with respect t the maf cal feed rate having been fixed by dividing it int the by-prduct prductin rate. Table A-7 lists the imprtant parameters f these crrelatins with the characteristics f the data used fr the crrelatin shwn in Table A-8. The crrelatin f tar il prductin with cal (maf) feed rate and atmic H/C rati is pltted in Figure A-5 alng with the data pints fr the smaller data base. Anther way f lking at the dependence f tar il prductin upn the atmic H/C rati f the feed cal is t plt the tar il prductin per tn f cal (maf) versus the H/C rati. This plt is shwn in Figure A-6, which shws that the tar il prduced per tn f feed cal (maf) decreases as the atmic H/C rati f the feed cal increases. Tw crrelatins were calculated fr these data: ne which is linear in atmic H/C rati and anther which includes a secnd-rder term, (H/C) 2. The secnd-rder crrelatin was smewhat better. Bth crrelatins are shwn in Figure A-6. The crrelatin f phenl prductin with cal (maf) feed rate and atmic H/C rati is pltted in Figure A-7, alng with the data pints fr the smaller data base. In a manner similar t the analysis f the tar il prductin, the phenl prductin per tn f cal (maf) was pltted against the atmic H/C rati. is shwn in Figure A-B, tn f feed cal (maf) This plt which shws that the phenl prduced per increases smewhat as the atmic H/C rati f the feed cal increases. A linear least-squares crrelatin was calculated fr this relatinship and is pltted in Figure A-8. 13

22 6.0 FINAL CORRELATIONS The secnd phase f the crrelatin wrk invlved analyzing GPGP data frm the first fur mnths f 1987, and cmparing them with the data and the results f the first crrelatin effrt. The fllwing discussin summarizes the results f this secnd phase and the final crrelatins develped frm the cmbined data. 6.1 Data Base Develpment GPGP data frm the first fur mnths f 1987 were supplied t ANG in the frm f files f data by mnth in spreadsheet frmat n several PC diskettes. The diskettes were first read t determine their cntents in terms f variables. The mnthly files were then cmbined t develp a cmprehensive data base f 1987 data with cnsistent variables. Fr a cmbined cmparisn and analysis the 1987 data were added t the data. Tw such cmbined data bases were develped. The s-called large data base, which cntained a ttal f 645 data sets, was built upn the large data base described abve In Sectin 5.1. The primary purpse f this data base was t develp the prduct crrelatins with the maf cal feed rate. A small 1987 data base was als develped cntaining the data n feed cal cmpsitin. Again, these types f data were nt available n a daily basis, being btained nly nce every three r fur days. This data base was used t investigate the ptential inclusin f a number f variables besides feed cal cmpsitin, such as ash fusin temperature, and a s-called small data base was put tgether which cmbined the 1987 and the data. This cmbined small data base cntained 163 r 190 data sets depending n which variables were being examined. 14

23 6.2 Identificatin f Crrelating Variables The identificatin f crrelating variables in the cmbined data bases was, f curse, guided by the results frm the first phase. The small and the large cmbined data bases were used t generate graphs f each prduct's prductin versus ptential crrelating variables. The graphs were generated with different symbls fr the tw types f data and in rder t cmpare the tw sets f data. One bjective f examining these graphs was t determine if the newer data really were different frm the earlier data. In general, the 1987 data did nt appear t be significantly different frm the data, and as a result final crrelatins were develped frm the ttal cmbined data base instead f nly the 1987 data. In additin, the 1987 data generally cvered a smaller range f variatin in the independent variable, making it mre difficult t develp crrelatins with reasnable cnfidence limits n the basis f the newer data alne. In an effrt t find additinal crrelating variables besides maf cal feed rate and H/C atmic rati, graphs were generated with an ash fusin temperature (IT r FT), weight percent Na20 in cal (mf), r lignite carbn number as the independent variable. The ash fusin temperature was tried as a variable because it is thugh that variatins in this variable wuld cause and therefre reflect a variatin in gasifier steam rate. Weight percent Na 2 0 was tried because it was thught that this ash cmpnent might have a catalytic effect upn the gasificatin reactins. Graphs were generated frm the data bases fr a number f dependent variables using ash fusin temperature as the independent variables: SNG prductin, SNG per tn f feed cal, tar il prductin, tar il per tn, phenl per tn, and naphtha 15

24 per tn. Nne f these graphs shwed that any f these quantities had any crrelatin with ash fusin temperature. Similarly, graphs with weight percent Na20 r the lignite carbn number as the independent variable shwed n crrelatin. On the ther hand, graphs generated with the independent variables fund frm the first phase t be useful crrelating variables lked very much the same as the graphs with the data. These independent crrelating variables included maf cal feed rate and atmic H/C rati. 6.3 Crrelatins with One Independent Variable Based n the effrt described in Sectin 5.2 and abve t identify crrelating variables, the mst irprtant variable is the maf cal feed rate. Crrelatins were develped fr each f the fur main plant prducts with this crrelating variable. Fr SNG prductin and naphtha prductin, these crrelatins were the nly nes which culd be develped frm the available data. Table 1 lists the imprtant parameters f these fur crrelatins f the plant prducts with the maf cal feed rate with the characteristics f the data used fr the crrelatins shwn in Table 2. The crrelatins in Table 1, which include the 1987 data, can be cmpared with the crrelatins in Table A-5 develped frm the earlier data SNG Prductin. The tw sets f SNG data are pltted in Figure 2, and it can be seen that the tw sets verlap. Althugh the tw sets f data verlap as if there were n significant difference between them, the data in the tw sets are different types f numbers. The data are MIS Net figures taken frm the published prgress reprts whereas the 1987 data in the data base are the sum 16

25 00% ONm.l% t- 0% N 00 G~ a 0% N % 4) 0 0 '0 a 6 I.IM S -0 t- IA z zip a* C. C) 0 40 '40 a t IL. a,, I-. 06' fx v 06 0 t t t lc C

26 . 00c ",4 CY mc * 0 0 CM 4 en N~ N 0 in 00a~ NN 'a G 0 t- LC t.0 '42 a' * a 00 L.- 4a - 'a% U tl A C'. C6 Y U) 06.~.46 0t 01 C. t Nv N -d Uef. :. MRNt vz t CL 4) C SO A. U, U. 4)18

27 00 LL dd < C3 0 N3 0 C3 D0 00 z t 0 19

28 f the quantity f SNG billed and that which was cnsumed internally. The latter figures are usually a few percent higher than the crrespnding MIS Net numbers. Cnsequently, the 1987 SNG data, shwn in Figure 3, were crrelated by themselves, and the crrelatin line is als shwn in Figure 3. The parameters f the calculated crrelatin line are shwn in Table 1 and the detailed characteristics are shwn in Table A-9 in the Appendix. The crrelatin cefficient r is 0.90, which is highly significant statistically (meaning that there is a crrelatin with little chance f being wrng) with 114 degrees f freedm Tar Oil Prductin. The tw sets f tar il prductin data are pltted in Figure 4, and it can be seen that the tw sets verlap with n apparent significant difference between them. Cnsequently, all f the data in bth sets were used t calculate a crrelatin f tar il prductin as a functin f maf cal feed rate. The parameters f the calculated crrelatin are shwn in Table 1, and the - detailed characteristics are shwn in Table A-1O in the Appendix. The crrelatin cefficient r is 0.51, which is highly significant statistically with 643 degrees f freedm. The crrelatin explains nly 26.3 percent f the sum f squares f deviatins, hwever, meaning that ther variables as yet unidentified (as well as a great deal f randm variatin caused by the way in which tar il prductin is measured) play a rle in determining tar il prductin. The 1987 data are pltted in Figure 5, which als shws the crrelatin line Naphtha Prductin. The tw sets f naphtha prductin data are pltted in Figure 6, and it can be seen that the tw sets verlap with n apparent significant difference between them. Cnsequently, all f the data in bth sets 20

29 Li L c al Li In6 U.4 CL tu 0 In 0 - -C S- c C z (.1)1\ in nn a a a 0a 0-21.

30 0 I+ 0b 0 00 cii *Ljj 00 LnL 130 V4 V- - V V V- - AVO)/SWDMY 'N0LLnlQ0d"1 WWI~VJ 22

31 0i A - + -E + c z w d U~ c TC ' N

32 I AL LL. 0L 0i F , c a_ AV/91fl'~t~lQ~ V±.0V

33 were used t calculate a crrelatin f naphtha prductin as a functin f maf cal feed rate. The parameters f the calculated crrelatin are shwn in Table 1, and the detailed characteristics are shwn in Table A-1i appendix. in the The crrelatin cefficient r is 0.60, which is highly significant statistically with 643 degrees f freedm. The crrelatin explains 36.0 percent f the sum f squares f deviatins. The 1987 data are pltted in Figure 7, which als shws the crrelatin line Phenl Prductin. The tw sets f phenl prductin data are pltted in Figure 8, and it can be seen that the tw sets verlap t a cnsiderable extent althugh the 1987 data may be smewhat lwer n the average than the earlier data. The pssible difference between the tw sets was nt judged t be significant enugh t warrant the develpment f tw separate crrelatins. Cnsequently, all f the data in bth sets were used t calculate a crrelatin f phenl prductin as a functin f maf cal feed rate. parameters f the calculated crrelatin are shwn in Table 1, and the detailed characteristics are shwn in A-12 in the appendix. The crrelatin cefficient r is 0.53, which is slightly lwer than that fr the naphtha prductin crrelatin but still highly significant statistically. The 1987 data are pltted in Figure 9, which als shws the crrelatin line. The 6.4 Crrelatins with Tw Independent Variables The cmbined small data base was used t develp imprved crrelatins fr tar il prductin and phenl prductin by including the feed cal atmic H/C rati as an additinal crrelating variable alng with the maf cal feed rate. This effrt was an extensin f the earlier wrk described abve in

34 LiJ L6 m 44. C..)+ +u +L C z 0 00 u %4. 0. Or ~~ii~ E.- I L0 in 0 i n EU 0.kVOS'IVDv4 VH~EVNO13naOd 26w

35 0 0 ca I0 <L 0 0 C,) + ) LCf c a.a z 0 C 13t I3t i I-~ 0 0 Av/s7VVI 'NLLncud 10tNGwd 27

36 +Z Lj.I + ++.q. -h.- 0A * cc am 4 IA q D.. + ' 0 zz. In- +28

37 Table 3 lists the imprtant parameters f the by-prduct crrelatins which include the atmic H/C rati as a crrelating variable. The characteristics f the data used fr the crrelatins are shwn in Table Tar Oil- Prductin. The parameters f the calculated crrelatin f tar il prductin with cal (maf) feed rate and atmic H/C rati are shwn in Table 3, and the detailed characteristics are shwn in Table A-13 in the appendix. This crrelatin remves 32.1 percent f the sum f squares f deviatins, and the crrelatin cefficient is The extent f the imprvement in the crrelatin by including H/C rati as a secnd crrelating variable can be seen by cmparing these measures with thse fr a crrelatin with nly cal (maf) feed rate (calculated frm the same data base): 21.7 percent and 0.47, respectively. This crrelatin f tar il prductin with cal (maf) feed rate and atmic H/C rati is pltted in Figure 10 alng with the data pints fr the cmbined small data base. A plt f the tar il prductin per tn f cal is shwn in Figure 11, which shws that the tar il prduced per tn f feed cal decreases as the atmic H/C rati f the feed cal increases. The parameters f this crrelatin are shwn in Tables 2 and 3, and its characteristics in Table A-14 in the appendix. Fr estimating the tar il prductin fr a given cmbinatin f maf cal feed rate and atmic H/C rati, either crrelatin described abve culd be used. Hwever, the crrelatin based n maf cal feed rate and H/C rati, as ppsed t estimating the tar il prductin per tn, gives a slightly better estimate. 29

38 044 - cc4 c t (W. cc 04) 0 0r - 0% enc m4)t- t 0 'U i 0a0 4' U) (Ud.xId '.4 V c' 0 V. 0-0) 4) " V. V0. 4 (U >S~ n cn c ~~30 c c- tv

39 u0 0 - '30 - "N% c ONN I 0% 0% 0 It (U N C m l 0 CY Al L. C~i (U N G ci 0 OD % 0 a9 44~ 0-3 C.0. '31 0

40 Cle 13 %A -I a ~ 0 0) IIn s IM- IOM 32

41 J + a.0 134C3 00 C) r N LlL nw w0 A0 a t0 40 S V V r - 1 NO./6WD 'Cdf) NW.L &id -110 WVI 33

42 ... I II- II i I I l I I I I _ Phenl Prductin. The parameters f the calculated crrelatin f phenl prductin with cal (maf) feed rate and atmic H/C rati are shwn in Tables 2 and 3, and the detailed characteristics are shwn in Table A-15 in the appendix. This crrelatin remves 28.0 percent f the sum f squares f deviatins, and the crrelatin cefficient is The extent f the imprvement in the crrelatin by including the H/C rati as a secnd crrelating variable can be seen by cmparing these measures with thse fr a crrelatin with nly cal (maf) feed rate (calculated frm the same data base): 26.5 percent and 0.51 respectively. This crrelatin f phenl prductin with cal (maf) feed rate and atmic H/C rati is pltted in Figure 12, alng with the data pints fr the cmbined small data base. 7.0 DISCUSSION OF RESULTS A plt f the phenl prductin per tn f cal is shwn in Figure 13, which shws that the phenl prduced per tn f feed cal may increase smewhat as the atmic H/C rati f the feed cal increases. The parameters f the crrelatin are shwn in Table 3, and the crrelatin is pltted in Figure 13. In additin f the 1987 phenl data, which tend t be lw cmpared t the data, t the data base had the effect f making this crrelatin less certain-than the crrelatin calculated frm nly the data. The crrelatin fr pnenl prductin as a functin f cal (maf) feed rate and atmic H/C rati gives a better estimate than the crrelatin fr phenl prductin per tn. Althugh the prduct crrelatins resulting frm this effrt d nt include the effects f gasifier perating cnditins, the results d prvide sme insight int the frmatin f the by-prducts. The 34

43 'U 441 UL +c W 0 a 1 i + L- 0 ~ %A. > 00 LLL 4 0 6J zvv"~ - N - N II'-cudIO314 35U

44 a 0 4A 0+ + C3 < -i I C3 + "E31O + Eu C)I/ -1 00[ + E 00 > 0 0 di rp a Of - I LI.. -( r.0 CD 0C 00 C NO/IO'CM 0NH NI G 36-

45 results illustrate the variability f the plant prductin f the varius prducts, althugh it is nt pssible t explain all f the variability in prductin n the basis f the available data n ptential influential variables. The crrelatins fr the varius prducts may be used t estimate the average amunt f prductin fr any given set f cnditins. The variability f the estimate is als described, in terms f a standard deviatin and cnfidence limits. The fllwing sectins summarize the discussin f crrelatin results with respect t key variables. The amunt f SNG prductin is largely determined by the maf cal feed rate. Over 80 percent f the sum f squares f deviatins was explained by this ne variable. SNG prductin did nt seem t be affected by any f a number f feed cal characteristics including percent vlatile matter, ash, Na20 cntent, ash fusin temperature, sulfur, r H/C rati. SNG prductin perhaps shuld depend n residence tie f the cal within the gasifiers, but this variable tends t be held fairly cnstant at the GPGP and thus shuld nt be a significant factr in the present set f data. The lack f influence f feed cal characteristics upn SNG prductin means that these parameters may be varied t influence the liquid by-prduct prductin withut interfering with SNG. The tar il prductin is mst strngly cntrlled by the maf cal feed rate, althugh this ne variable alne explains little mre than 26 percent f the sum f squares f deviatin. Tar il prductin als appears t be a functin f the atmic H/C rati f the feed cal, decreasing with an increase in value f this cal characteristic. Tar il prductin is als influenced by varius perating parameters dwnstream f the gasifier fr which there are n available data at this time. 0 Prductin f phenl is influenced by the maf cal feed rate, and t sme degree, the atmic H/C rati. As this latter variable 37

46 increases, s may phenl prductin. pssible feedstck fr jet fuel prductin. The phenl stream is als a The fact that the tar il and the phenl prductins are functins f feed cal H/C rati perhaps is an indicatin that the H/C rati represents the types f vlatiles cntained in the cal. A higher H/C rati indicates that the vlatiles are smewhat lighter in cmpsitin and have mre single-ring cmpunds, and vice-versa fr a lwer H/C rati - the vlatiles have mre multi-ring cmpunds likely t wind up as tar il. The percent vlatile matter f the feed cal is an imprtant parameter which influences the yield f liquid prducts resulting frm cal pyrlysis. Hwever, this parameter des nt appear t play any rle in determining the yields f liquid by-prducts in cal gasificatin. This is prbably s because the pyrlytic reactins are nt allwed t g t cmpletin. Rather the devlatilizatin zne is cmparatively small, and the cal travels very quickly int the gasificatin regin. The yields f tar il and phenls culd prbably be increased by any mechanism r pretreatment f the cal which wuld allw the vlatiles t escape mre easily r quickly, when the cal is fed int the gasifier. Particle size wuld be ne such variable. The smaller the particle size, the mre easily the vlatiles can escape befre the particle travels int the gasificatin zne. curse, there are limits n the apprpriate cal particle size which can be accmmdated by the gasifier. Hwever, it wuld be f interest t cllect the data n particle size distributin and t try t relate this variable t tar il and phenl prductin. Of The crrelatins fr tar il and phenl indicates that there is a tradeff between the tw by-prducts which can be made as a functin f the atmic H/C rati f the feed cal. A lwer H/C rati - fr example, ging frm 0.8 t wuld increase tar il by 1 38

47 gal/tn maf cal and lwer phenl by 0.1 gal/tn. Since phenl is a mre difficult cmpund t hydrtreat t get rid f the xygen, the net result wuld be t increase jet fuel prductin and t decrease perating csts. On the ther hand, the suggestin has been made t market the phenls. By-prduct credits culd, therefre, be increased by using a feed cal with a high H/C rati. The 1987 data n prductin d nt appear t be significantly different frm the data. 8.0 CONCLUSIONS The fllwing cnclusins summarize the findings develped frm this initial effrt t crrelate the liquid by-prduct prductin data frm the GPGP. Fr all fur prducts analyzed - SNG, tar il, naphtha, and phenl - the maf cal feed rate was the mst influential independent variable. Fr SNG prductin the maf cal feed rate was the nly influential independent variable. With this variable alne mre than 80 percent f the sum f squares f deviatins was remved - a very gd crrelatin with 116 data pints (1987 data nly). N feed cal characteristic examined, including percent vlatile matter, H/C rati, percent sulfur, percent ash, percent Na 2 0, r ash fusin temperature cntributed very much t a better crrelatin fr SNG prductin. Tar il prductin was influenced mst strngly by maf cal feed rate. Althugh the crrelatin remved nly slightly mre than 26 percent f the sum f squares f deviatins, the crrelatin was still fund t be significant because f the relatively large number f data pints (645). 39

48 Tar il prductin was fund t be influenced by the atmic H/C rati f the feed cal, and a crrelatin was develped using bth the maf cal feed rate and the atmic H/C rati as crrelating variables. T!_, il prduced per tn f maf cal was fund t decrease with increased H/C rati. Tar il prductin was nt fund t be influenced by percent vlatile matter, percent sulfur, percent ash, percent Na 2 0, r ash fusin temperature In the feed cal. T smth ut variatins in tar il prductin caused by the measuring methd, three-day rlling averages were used t develp a crrelatin with maf cal feed rate, but the crrelatin was nt very different frm ne calculated frm the riginal daily data. A crrelatin was develped fr naphtha prductin with maf cal feed rate as the crrelating variable. A crrelatin was calculated fr phenl prductin with bth maf cal feed rate and atmic H/C rati as crrelating variables. Phenl prductin per tn was fund t increase slightly with increased H/C rati f the feed cal. This analysis and the crrelatins that were develped describe the variability f the plant prductin f the varius prducts, althugh it is nt currently pssible t explain all f the variability in prductin n the basis f the available data. Theries f cal devlatilizatin and pyrlysis can help explain the frmatin f the liquid by-prducts in gasificatin, althugh mre analysis alng this line f investigatin is required. 40

49 0 The 1987 data n prductin d nt appear t be significantly different frm the data. 9.0 REC0WNEATIONS As the result f this initial effrt t analyze the data n liquid byprduct prductin at the GPGP, the fllwing recmmendatins are made t develp the data crrelatins further and t make use f the crrelatins in the jet fuel prductin prgram. Further study and analysis f the basic prcesses ccurring in the gasifier t prduce the liquid by-prducts shuld be dne. With this infrmatin a mdel fr their frmatin can be develped, and the imprtant parameters cntrlling their prductin can be identified and incrprated int the data crrelatins. An analysis shuld be made f the peratins ccurring after the gasifier that affect the quality and quantity f the liquid byprducts. Imprtant parameters affecting by-prduct prductin can be identified fr cllecting data fr later incrpratin int the crrelatins. Cmpsitin data n the tar il shuld be analyzed and cmpared t feed cal cmpsitin data and perating data t learn mre abut tar il frmatin in the gasifier and feedstck characteristics. An effrt shuld be made t btain gasifier perating data, n a restricted basis if necessary, t incrprate such parameters as the steam rate, temperature, and pressure int the crrelatins. The crrelatins that have been develped shuld be used in technical and ecnmic tradeff studies t estimate tar il and phenl prductin versus feed cal characteristics t develp the ptimum design fr the.iet fuel prductin facilities. 41