Embed Size (px)
Citation preview
GRI -91/0193
WITH COMPLIMENTSGAS RESEARCH INSTITUTE
LIBRARY SERVICES
TREATABILITY AND SCALE-UP PROTOCOLSFOR POLYNUCLEAR AROMATIC HYDROCARBON
BIOREMEDIATION OF MANUFACTUREDGAS PLANT SOILS
FINAL REPORTVOLUME 1
(SEPTEMBER 1987-JULY 1991)
LIBRARYROBERT S. KERR ENVIKONMEN
RESEARCH LABORATORY
Gas Research Institute8600 West Bryn Mawr Avenue
Chicago, Illinois 60631
[
TREATABILITY AND SCALE-UP PROTOCOLSFOR POLYNUCLEAR AROMATIC HYDROCARBON BIOREMEDIATION
OF MANUFACTURED GAS PLANT SOILS
FINAL REPORTVOLUME I
Prepared byJ. W. Blackburn, P. M. DiQrazia, and J. Sanseverino
University of Tennessee423 S. Stadium
Knoxville. TN 37996
ForGAS RESEARCH INSTITUTEContract No. 5087-253-1490
GRI Project ManagerDavid G. Unz
Environment and Safety Research
July 1991
.CPOir MCU-ENTATWH '«7T917oi93
A. THM •** »«•*«••Treatabil i ty and Scale-Up Protocols for Polynuclear AromaticHydrocarbon Bioremediation of M a n u f a c t u r e d Gas Plant SoilsFinal Report , Volume I
7. A*flH*rlt)J . W . Blackburn, P . M . DiGrazia , and L. Sanseverino
«. p*ffo**"f|f Of|»nu»tW" *••*• ••»« M"*«*University of Tennessee, Knoxville423 South Stadium HallKnoxville, TN 37996
12. fto»*i»»nn( Oryoni»ti«n *um« •** **•*«*»
Gas Research Institute8600 West Bryn Mawr AvenueChicago, IL 60631
1. ft»u»M*«-i iiiiii j *•
*• **m* DM._Ju lv R 1991
•V
•V ^tof^m Orv**tf»li*A N«»i He
M. *IIILI/Tofth/**rk Unn M»
. , , . N/A11. Bantu rtfC( •» ftr»Mt(C) Noto 5087-253-1490(0)
IL Trw •< JU«*« L PV«« te««r«eFinal Report9/87-7/91
14.
II.
tOO vw«)This report describes activities to develop a framework to reliably scale-up and
apply challenging bioremediation processes to polynuclear aromatic hydrocarbons inManufactured Gas Plant (MGP) soils. It includes: a discussion of the accuracy neededfor competitive application of bioremediation; a framework and examples for treatabilityand scale-up protocols for selection, design and application of these processes: bothbatch and continuous testing protocols for developing predictive rate data: and specialpredictive relationships that nay be used in process selectioD/scale-up.
This work, coupled with subsequent work (as recommended) to develop an MGP soildesorption/diffusion protocol and nev scale-up methods, and with subsequent scale-uptestinp should lead to the capability for improved selection of MGP sites for bio-remediation and improved performance, success, and reliability of field applications.With this greater predictive reliability, bioremediation will be used more often in thefield on the most favorable applications and its cost advantages over other remediationoptions will be realized.
17.Coal tars, organic wastes, MGP sites, polynuclear aromatic hydrocarbons, naphthalene,phenanthrene, anthracene, bioremediation, desorptioc, kinetics, bioavailability , treat'ability studies, scale-up, dynamic analysis, systems identification
' bioremediation of soils, slurry reactors
COtATi
release unlimited!«. SMwrttr Ctowunclassified
CIMI fTfcw
fl. No. ml »•*•*
(fcoo ANftt-Ut.ll) rout* 272 <*-THTIt-JI)
GRI DISCLAIMER
LEGAL NOTICE: This report was prepared by the University of Tennessee as anaccount of work sponsored by the Gas Research Institute (GRI). Neither GRI,members of GRI, nor any person acting on behalf of either:
a. Makes any warranty of representation, express or implied, with respectto the accuracy, completeness, or usefulness of the information contained inthis report, or that the use of any apparatus, method, or process disclosedin this report may not infringe privately owned rights; or
b. Assumes any liability with respect to the use of, or for damages resultingfrom the use of, any information, apparatus, method, or process disclosedin this report.
Trartabilrty and Scalt-up Protocols for FINAL REPORTPolynuelaar Aromatic Hydrocarbon i VOLUME IBiortmtdiation of MQP Soils July 1991
PROLOGUE
The Gas Research Institute (GR1) has embarked on a multi-year program toaddress the Management of Manufactured Gas Plant (MGP) Sites. The programfocuses on three primary technical areas: site investigation, risk assessment, andsite restoration. Program activities in each of these areas are separated into twophases. Phase I emphasizes the identification, collation, and analysis of thestate-of-the-art applicable to specific MGP site situations. Phase II involves thegeneration of data through targeted research and development projects.
The first phase of the program has produced a four-volume report whichdocuments the state-of-the-art findings. The second phase of the program isproducing a series of topical and final reports, such as this one, from laboratoryand field experimental test programs. These programs are designed to fill datagaps required for the effective management of these sites and to advance thestate-of-the-art for technologies determined during the first phase as potentiallyapplicable to MGP site wastes. Other reports from the second phase willcontinue to be produced through 1991 as various portions of the program arecompleted.
This project, one of the Phase II efforts, has provided two final technicalreports. Volume I deals with the results of this project from an applications engi-neering perspective. Volume II provides the scientific backup and supportunderpinning Volume I as well as specific scientific findings of importance to theultimate application of environmental biotechnology to MGP site remediation.Accordingly, two report styles, engineering and scientific, will be evident in therespective documents.
Final Report. Volume I. Treatability and Scale-up Protocols for PolynuclearAromatic Hydrocarbon Bioremediation of Manufactured Gas Plant Soils providesa discussion of the design accuracy required in order to reliably scale-up MGP
Trotability »nd Sc«lt-up Protocol* lor FINAL REPORTPolynuelttr Aromatic Hydrocarbon iii VOLUME IBiortm«di«tion of MGP Soils July 1W1
soil bioremediation processes; a framework and examples for both treatabilityand scale-up protocols for selection, design and application of these processes;a predictive relationship between PAH biodegradation kinetics and site charac-teristics; a continuous, dynamic protocol for testing process effectiveness, stabil-ity and robustness and determination of biodegradation rate models andconstants in disturbed, field-like test systems; and a statistical analysis of theimportance of analytical and experimental variation and error on reliable scale-up.Detailed experimental descriptions are available in an appendix under separatecover.
Final Report. Volume II. Molecular Bioanalytical Methods for MonitoringPolvnuclear Aromatic Hydrocarbon Biodegradation in Manufactured Gas PlantSoils includes technical reviews of PAH bacterial metabolism, bioanalysis of PAH-degrading bacteria and their activity; an analysis of PAH-degrading activity inMGP soils; a discussion of a PAH-degrader culture collection developed duringthe project; and an analysts of the use of gene probes and other newapproaches for improved bioanalysis of PAH degradation in MGP soil wastes.
FINAL REPORT Tr«l»bilty and Scalt-up Protocols forVOLUME I iv Potynuelaar Aromatic HydrocarbonJuly 1091 Biortmtdiation of MOP Soils
RESEARCH SUMMARY
TITLE Treatability and Scale-up Protocols for Polynuclear AromaticHydrocarbon Bioremediation of Manufactured Gas PlantSoils
CONTRACTOR University of TennesseeGRI Contract #5087-253-1490
PRINCIPALINVESTIGATOR
REPORTINGPERIOD
OBJECTIVE
Dr. James W. Blackburn
September, 1987, through July. 1991Final Report, Volume I
To develop testing and evaluation protocols for polynucleararomatic hydrocarbon (PAH) bioremediation processes thatlead to improved reliability In the field-scale remediation ofmanufactured gas plant (MGP) site materials.
TECHNICALPERSPECTIVE
The management of manufactured gas plant sites may haveto be addressed as part of voluntary property redevelop-ment actions or evolving regulatory requirements. Thenature of many chemicals-of-interest present in the soils atthese sites make them candidates for biological treatment.This management strategy represents a potentially techni-cally viable alternative that may be capable of cost-effectively treating the soils and reducing futureenvironmental liability associated with these residues.
Instability and Scala-up Protocol* lorPolynuclaar Aromatic HydrocarbonBioramtdiation ot MGP Soils
FINAL REPORTVOLUME IJuly 1991
While prior studies have suggested that many of the com-pounds present in these residues are biodegradable, pre-diction of the biokinetic rates needed for field-scale designand performance estimation of engineered bioremediationprocesses may be difficult. The highly concentrated, oftentarry and viscous nature of MQP soils, the variable butoften long-term weathering of MGP soils, and the oftenvariable nature of the fundamental biodegradation pro-cesses lead to a unique and challenging application for bio-remediation.
Exploitation of the potential of MGP site bioremediation atthe commercial scale will require the ability: 1) to select themost appropriate and predictable sites on which to use bio-remediation while rejecting those sites where performanceis either poor or uncertain and 2) to develop an accurateprediction of the field-scale performance of a given design..According to an informal consensus of bioremediation pro-fessionals, application of bioremediation processes willrequire relative accuracy in the prediction of field-scalecosts and performance (treatment times and final removaleffectiveness) of around ±20% if the technology is tocompete successfully with other remediation alternatives.
This requirement establishes the criterion for accuracy andprecision needed for bioremediation testing and evaluation(treatability) and scale-up protocols. An improvedapproach over that currently used is needed for reliablePAH bioremediation technology application to MGP sitematerials.
FINAL REPORT Traatability and Sealt-up Protocols forVOLUME I vi Polynuclttr Aromatic HydrocarbonJuly 1991 Biort mediation of MGP Soils
RESULTS An overall testing and scale-up protocol framework for PAHbioremediation processes on MGP sites is proposed thatincludes: 1) determination of PAH biotransformation anddesorption/diffusion rates in separate experiments, 2} deter-mination of idealized PAH biotransformation rates from bothstandardized batch and continuous test reactor systems, 3)a concept for quantification of desorption/diffusion rates intest systems, and 4) a concept for the development of"scale-up" relationships for future improved field perform-ance prediction.
Naphthalene, phenanthrene, and anthracene batch ideal-ized biotransformation kinetic potential (IBKP) tests wereperformed in small batch vials each containing one of fiveMGP soils. A radiolabeled PAH was used and led toimproved analytical precision as compared to classical PAHchemical analysts. This improved analytical precisionenabled an analysis of the complex nature of the biotrans-formation rate mechanism with the conclusion that batchPAH biotransformation in MGP soils is generallynon-elementary (e.g., not zero-, or first-order) and may varywith time over the experiment (time-variant). These resultsindicate that conventional bioprocess modellingapproaches to design and scale-up of batch bioremediationprocesses may be inherently less certain than those fornon-biological remediation processes.
The improved analytical precision and the use of a pseudo-first-order biotransformation rate equation based only onthe initial and final PAH concentrations ted to a statisticallysignificant relationship (correlation coefficient, r? - 0.978)between phenanthrene biotransformation rate constants in
Titatability and Sctl*-up Protocols lor FINAL REPORTPolynucl«ar Aromatic Hydrocarbon vii VOLUME IBiorcmtdiation of MOP Soils July 1991
these standardized batch IBKP tests based only on theinitial total PAH concentration. This indicates that creationof meaningful and reliable scale-up relationships to improveearly selection of the best (and early rejection of the worst)cases of MGP soils for bioremediation may be possible.
A continuous IBKP test of naphthalene biotransformation ina dynamically-perturbed, continuous soil slurry reactorcharged with an MGP soil was developed to: 1) examinewhether batch and continuous kinetic results can becompared, 2) to determine what impact dynamic field-scaledisturbances might have on PAH biotransformation kinetics,3) to estimate short-term PAH desorption/diffusion behaviorin a continuous system operating on MGP soils, and 4} totest the feasibility of extracting process models ultimately fordesign and scale-up from experimental, time-series input-output data.
While first-order rate behavior varied widely between thebatch and continuous IBKPs, comparison of the sec-ond-order rates between different lab-scale systems wasencouraging. The batch IBKP second-order rate behaviormay offer an estimate of the worst-case second-order rateconstant for the continuous system and the first step in areliable scale-up procedure.
Continuous IBKP test results for the MGP soil testedsuggest a system prone to time-variant state changes (oftendriving the naphthalene concentrations to undetectablelevels, but otherwise a stable and robust naphthalene-degradation process. Rates of naphthalene biotransforma-
FINAL REPORT Trtatability and Scalt-up Protocol* forVOLUME I viii Polynuclaar Aromatic HydrocarbonJuly 1991 Biortmediation of MGP Soils
tion were resistant to temperature and concentrationchanges and were adversely affected only when oxygenavailability was severely restricted.
Two types of sorption/desorption behavior for PAHs in MGPsoils were identified based only on the dynamic behavior."Fast" sorption/desorption processes (less than 24 hours inresponse time) could be quantitatively analyzed in the con-tinuous IBKP experiments. "Fast" naphthalene sorption/de-sorption rates in high organic carbon MGP soil wereindistinguishable from those seen in uncontaminated controlsoils. "Slow" sorption/desorption processes were docu-mented in special experiments and led to the hypothesisthat diffusion within the soil tar phase observed sometimesmay be the rate-limiting step for biotransformation of PAHson MGP soils.
Batch IBKP results using conventional analysis (includingextraction and analysis of the soil phase) likely include theeffect of the "slow" processes, where neither the batch IBKPusing labeled analysis nor the continuous IBKP "see" theslow effects. For these reasons, batch labeled and continu-ous IBKPs tend to measure only biotransformation andminimize "slow" sorption/desorption effects and may lead tovalid comparisons of biotic rates.
The continuous IBKP Rate Equation was found to befirst-order using parameter estimation techniques, asopposed to the non-elementary, time-variant rate behaviorfound for batch processes. No phenomenological processmodel based on frequency response was needed for thisMGP soil slurry system.
Trt«t»biltty and $cal*up Protocols tor FINAL REPORTPofynueliar Aromatic Hydrocarbon be VOLUME 1Biortmtdialion of MOP Soils July
Task 3. Development of a Calibration/Scaling FactorProtocol: To develop a calibration/scaling factor protocol torelate biotransformation in one lab-, pilot-, or field-scale testwith other such tests,
Task 4. Application of Combined Protocols to Town-GasSites: The application of the above tools and protocols toone or more actual GR1 town-gas sites with a processdesign and forecast of the system performance, and
Tasks. Verification of the Predictive Protocols: The verifi-cation of the predictive and scaling capability throughsupport of efforts by other GRI contractors to implementenvironmental biotechnology at pilot and field scales.
Parallel treatability study activity was implemented byother GRI contractors (ReTeC, Inc.) on a variety of MGP sitematerials early in the overall GRI program and difficulty inselecting a site for field-scale application was encounteredleading to a focus on the Task 1 and 2 objectives through-out this project. Task 3 was, with the concurrence of GRIproject management, postponed to later projects. Task 4became a cooperative effort across GR; contractors to testand analyze a number of. MGP wastes and became a variedsource of materials from which to recover organisms anddetermine kinetic activity for both Tasks 1 and 2. Task 5was postponed to a later effort.
PROJECT This work, coupled with subsequent work to develop a MGPIMPLICATIONS soil desorption/diffusion protocol and scale-up protocols,
and subsequent scale-up testing should lead to the capabil-ity for improved selection of MGP sites for bioremedtation
TfMttbilrty *nd Sc«l«-up Protocol* for FINAL REPORTPoiynuclttr Aromatic Hydrocarbon xi VOLUME 1Biortmtdifttion of MOP Soils July 1991
TECHNICALAPPROACH
Continuous bioremediation processes had higher biotrans-formation rates versus batch processes and may haveadvantages in both performance and stability as comparedto batch processes. These features may make continuousprocesses (continuous in terms of liquid feedstream) attrac-tive even though higher capital and operating costs mayoccur expected.
These lab-scale methods are not expected to provide rateinformation directly applicable to field-scale systems.Rather, new "scale-up" relationships should be developedas both standardized lab-scale and field-scale rate data aredeveloped. Because of delays in planned field-scale trials,these "scale-up" relationships could not be developed asinitially planned in this work.
This project was initially structured as a number of Taskstogether accomplishing the overall objective of developmentand verification of an improved bioremediation testing andevaluation protocol.
Task 1. Development of New Molecular Tools to EnumerateDegradative Organisms: To develop molecular tools fromdegradative genes to track microorganisms capable ofdegrading polynuclear aromatic hydrocarbons and othercompounds of interest in town-gas sites,
Task 2. Development of a Dynamic Systems AnalysisProtocol: To develop a protocol for dynamic systemsanalysis of the biological system.
FINAL REPORTVOLUME tJuly 1991
Iff «t*bility and Seal«-up Protocols forPorynucl«ar Aromatic Hydrocarbon
Biortmtdiation of MQP Soil*
and improvement in performance, chances for success, andreliability of cost estimates. With this greater predictive reli-ability, bioremediation will be used more often in the field onthe most favorable applications and its cost advantagesover other remediation options will be realized by the GasIndustry.
Because of delays in field applications planned for othercomponents of the overall GRI program, these methodshave not been tested to establish whether they actuallyimprove scale-up reliability. These and possibly other treat-ability and scale-up protocols should be incorporated intofuture pilot- and field-scale bioremediation tests todetermine the potential for improved reliability for MGP PAHbioremediation applications.
David 3. LinzGRI Project Manager
Environment and Safety Research
FINAL REPORT Tr»Ubility and Scale-up Protocols lorVOLUME I xjj Potynucbar Aromatic HydrocarbonJuly 1991 Biorcmadittion of MOP Soils
TABLE OF CONTENTS
TITLE PAGE
GRI DISCLAIMER................................................................................................................. i
REPORT DOCUMENTATION PAGE................................................................................... ii
PROLOGUE.......................................................................................................................... iii
RESEARCH SUMMARY....................................................................................................... v
TABLE OF CONTENTS....................................................................................................... xiii
TABLE OF TABLES............................................................................................................. xv
TABLE OF FIGURES..,,,.,.,,...,.,,..,,.,.,,,....,,,......,.,..,.,,,..,,.,....,,.....,,,.,., xvii
1.0 TREATABILITY AND SCALE-UP Bl ORE MEDIATION PROTOCOLS,,,,,.,.,,,,,, 11.1 PREDICTIVE RELIABILITY IN BIOREMEDIATION....,,,,.,.,.,,..,,,.,..,,,.... 1
1.1.1 DESIGN EQUATIONS AND RATE EQUATIONS........................................ 41.1.2 PERFORMANCE PARAMETER SELECTION ............................................. B1.1.3 COMPLEXITIES OF CATABOLIC RATE BEHAVIOR,,.,,,,,,,,,,,,,., 131.1.4 SCALE-UP OF PERFORMANCE DATA FROM TEST SYSTEMS.............. , 14
1.2 OVERALL TREATABiLTTY AND SCALE-UP PROTOCOL CONCEPT..,,,,,.. 15
2.0 A BATCH INTRINSIC BI ©TRANSFORMATION KINETIC POTENTIAL TEST .....,...„ 232.1 RESULTS OF LABORATORY-SCALE BATCH IBKP EVALUATIONS.............. 24
2.2.1 MICROBIOLOGICAL ANALYSES,,,,,,.,,,,,...,,.,,,,,,.,,,,.,,,,,, 242.1.2 PHENANTHRENE BlOTRANSFORMATlON AND MINERALIZATION.,,.. 252.1.3 NAPHTHALENE BlOTRANSFORMATlON AND MINERALIZATION,,,,, 392.1.4 ANTHRACENE BlOTRANSFORMATlON AND MINERALIZATION,,,,,, 43
2.2 DISCUSSION OF BATCH IBKP RESULTS,,,,,,,,,,,,,,,,,,,,,,,,,,,., 462.2.1 BlOTRANSFORMATlON VS. MINERALIZATION KINETICS,,,,,,,,,.,, 462.2.2 AXENIC (PURE CULTURE) VS. MGP SOIL KINETICS.,,,,,,,,,,,,,,, 472.2.3 RELATIVE PAH BlOTRANSFORMATlON KINETICS,,,,,....,,,...,..,,., 48
2.3 PREDICTIVE KINETIC RELATIONSHIPS FOR COMPARISONS BETWEENBATCH TESTS ,,„,,,,,,,,,,.,,,.,,„...,.....,.,,...„.,,,...,,...,„.,,„..,....,, 49
2.4 ANALYSIS OF ALLOWABLE ERROR FOR RELIABLE SCALE-UP ,,,,„„„ 522.4.1 IMPACT OF EXPERIMENTAL ERRORS ON THE RATE CONSTANT....... 522.4.2 IMPACT OF CHEMICAL ANALYSIS ERRORS ON RATE CONSTANT
ERROR,,,,,,,,.,,,,,,,,,.,,,,,..,,,,,,,,.,,,,,,,,,,,,,,,,,,,,,,. 552.4.3 IMPACT OF RATE CONSTANT ERROR ON SCALE-UP,,,,,.,.,,,,,,, 56
Trtatability and Se*f*-up Protocols for FINAL REPORTPolynuclaar Aromatic Hydrocarbon xiii VOLUME I8 ion mediation of MOP Soil* Jury 1991
2.5 DESORPTION/DIFFUSION AS A BIOTRANSFORMATION RATELIMITATION............................................................................... 60
2.6 CONCLUSIONS ...................................MM...................................................... 61
3.0 A CONTINUOUS INTRINSIC BIOTRANSFORMATION POTENTIAL TEST............... 653.1 DYNAMIC MICROBIAL SYSTEMS ANALYSIS.................................................. 653.2 DESCRIPTION OF EQUIPMENT....................................................................... 693.3 RESULTS OF PRELIMINARY STUDIES............................................................. 72
3.3.1 MATERIALS................................................................................................... 733.3.2 OPERATING PARAMETERS........................................................................ 743.3.3 PARTITION COEFFICIENTS........................................................................ 753.3.4 KINETIC STABILITY AND ROBUSTNESS................................................... 79
3.4 NAPHTHALENE BIOTRANSFORMATION IN THE CSTR TEST SYSTEM ...... 823.4.1 RESULTS FROM MANUAL SOIL RECYCLE MODE................................... 833.4.2 REACTOR FILTER INSERT......................................................................... 89
3.4.2.1 EARLY SYSTEM OPERABIUTY STUDIES............................................ 903.4.2.2 FREQUENCY-RESPONSE ANALYSIS,,,,,,,..,,,.,,,,.,,,,,,,,, 90
3.5 CONCLUSIONS...,,,,.,,,,,.,,,,.,,,...,.,,,.,.,,,,,,,,,,,,.,,,.,,,,,, 101
4.0 OBSERVATIONS ON NAPHTHALENE BIOREMEDIATION KINETICS,,,,,,,.,,,. 4034.1 COMPARISON OF KINETIC RATE DATA BETWEEN SIMILAR REACTOR
CONFIGURATIONS,,,,,,,,.,,,,,,,,,...,,.,,.,,,,,,.,,.,.,,,,,,,,,,,,, 1034.2 PREDICTION OF BATCH TIME-CONCENTRATION PROFILES.,,,,.,,,,,.. 1034.3 COMPARISON OF BATCH AND CONTINUOUS RESULTS............................ 1034.4 SECOND-ORDER KINETICS............................................................................. 1054.5 MASS TRANSFER ,.,,.,,,.,,,,,,,,.,,...„.,,.,,.,,,,,,,,,,,,,,,,,,,,, 1054.6 NAPHTHALENE BIOTRANSFORMATION IN SOILS VS. WASTEWATER...... 1064.7 FIELD-SCALE PROCESSES.............................................................................. 107
5.0 REFERENCES.,,,.,,,.,,.,,,,,,,,,,,,,...,,,....,,,,,,,,,,,,,,,,,,,,,,,,,, 109
APPENDIX A. SUMMARY OF CALCULATIONS AND MODELS..,.,,.,,,,,,,,.,.,,,... A-1
APPENDIX B. DESCRIPTIONS OF EXPERIMENTS....,,,,,,,,,,,,.,,,,,,,,-,,,,,. B-1
FINAL REPORT Traatabilfty and Scala-up Protocol* forVOLUME I xrv Potynucltar Aromatic HydrocarbonJuly 1991 Bioramtdiation of MGP Soil*
TABLE OF TABLES
TABLE TITLE PAGE
1 Characterization of the Contaminated MGP Soils Tested........ 24
2 Results of MGP Soil Microbiological Analyses........................... 25
3 Effect of Acetone Solvent Carrier on NaphthaleneMineralization................................................................................ 29
4 Summary of Overall Batch Pseudo-First- and Second-OrderRate Constants for PAH Biotransformation andMineralization................................................................................ 46
5 PAH Carbon Mineralization Rate as a Percentage of theBiotransformation Rate................................................................ 47
6 Biotransformation and Mineralization Rates of Naphthalene,Phenanthrene Relative that of Anthracene in MGP Soils.......... 48
7 Regressions for Predictive Kinetic Relationships......................., 51
8 Coefficient of Variation Error Statistic for ConventionalChemical PAH Analysis in Initial MGP Soils............................... 55
9 Coefficient of Variation Error Statistic for Radiolabeted PAHAnalysis in Initial MGP Soils and Average for aBiotransformation Test................................................................. 56
10 Characteristics of Uncontaminated Tennessee Control Soil.... 74
11 Determination of Stripping and Sorption Abiotic Rate Con-stants in Perturbed CSTR Experiments...................................... 78
12 Pseudo-First- and Second-Order Rate Constants from CSTRReactor Studies and Batch Naphthalene Biotransformationand Mineralization Assays........................................................... 87
Tr«t«bilrty and Seal*-up Protocols lor FINAL REPORTPorynuclttr Aromatic Hydrocarbon xv VOLUME IBiortmtdialion of MGP Soil* July 1991
13 Summary of Interval Continuous IBKP Pseudo-First-OrderNaphthalene Biotransformation Rate Constants and ReactorLiquid Concentrations.................................................................. 93
14 Summary of Interval Continuous IBKP Pseudo-Second-Order Naphthalene Biotransformation Rate Constants............ 94
15 Summary of Naphthalene Biotransformation RateConstants...................................................................................... 104
A-1 Types of Error in Three-Parameter Design Equations.............. A-4
FINAL REPORT Traatabilrty and Scala-up Protocols forVOLUME I xvi Porynudaar Aromatic HydrocarbonJuly 1991 Biortmtdiation of MGP Soils
TABLE OF FIGURES
FIGURE TITLE PAGE
1 A Simplified Biochemical Pathway for the MicrobialConversion of Toluene................................................................. 10
2 The Impact of Biotransformation on the Physical Propertiesand Fates of Toluene Metabolites............................................... 12
3 A Concept for Reliable Treatability and Scale-up Protocols orPAH Biotransformation in MGP Soils.......................................... 19
4 Concept for a Laboratory-Based Treatability Protocol forMGP Soils...................................................................................... 20
5 Results from Radiolabeled Phenanthrene Batch MGP SoilFate Tests, MGP Soils A and B................................................... 26
6 Results from Radiolabeled Phenanthrene Batch MGP SoilFate Tests. MGP Soils C and D..................................................." 27
7 Results from Radiolabeled Phenanthrene Batch MGP SoilFate Tests, MGP Soil E................................................................ 28
8 Rates of Radiolabeled Phenanthrene Biotransformation andMineralization in Batch MGP Soil Tests...................................... 33
9 Interval-Based Pseudo-First-Order Rate Constants for Radio-labeled Phenanthrene Biotransformation and Mineralizationin Batch MGP Soil and Axenic Organism Tests......................... 36
10 Actual and Predicted Radiolabeled Phenanthrene Concen-trations in the Hexane Extract from Batch MGP Soil andAxenic Organism Tests................................................................ 37
Trtatibilrty and Scal*-up Protocols tor FINAL REPORTPolynucltar Aromatic Hydrocarbon xvii VOLUME IBiottmtdiation of MOP Soils Jury 1991
11 Overall Pseudo-First- and Second-Order Radiolabeled Phe-nanthrene Rate Constants from Batch MGP Soil and AxenicOrganism Tests............................................................................ 38
12 Interval-Based Pseudo-First-Order Rate Constants for Radio-labeled Naphthalene Biotransformation and Mineralization inBatch MGP Soil and Axenic Organism Tests............................. 41
13 Actual and Predicted Radiolabeled Naphthalene Concentra-tions in the Hexane Extract from Batch MGP Soil and AxenicOrganism Tests............................................................................ 42
14 Overall Pseudo-First- and Second-Order RadiolabeledNaphthalene Rate Constants from Batch MGP Soil andAxenic Organism Tests................................................................ 43
15 Interval-Based Pseudo-First-Order Rate Constants for Radio-labeled Anthracene Biotransformation and Mineralization inBatch MGP Soil and Axenic Organism Tests............................. 44
16 Overall Pseudo-First* and Second-Order RadiolabeledAnthracene Rate Constants from Batch MGP Soil and AxenicOrganism Tests............................................................................ 45
17 Predictive Relationship for Phenanthrene Overall Pseudo-First-Order Biotransformation Rate Constants in MGP Soils.... 51
18 Error Propagation to the Calculated First-Order RateConstant From Measurement Errors in Concentration andResidence Time in Continuous Tests......................................... 53
19 Error Propagation to the Calculated Zero- and First-OrderRate Constants From Measurement Errors in Concentrationand Sampling Time in Batch Tests............................................. 54
20 Relationship Between the Number of Replicate Kinetic Tests.the Coefficient of Variation Between Rate Constants BetweenReplicates and the Relative Error of the Calculated Mean....... 57
FINAL REPORT Ti««t»bilrty and S«l*-up Protocols lorVOLUME t xviii Pofynuelaar Aromatic HydrocarbonJuly 1991 Bior»m«diation of MOP Soils
21 Impact of Error in the Rate Constant on the Probability ofExtended or Reduced Treatment Times Relative to thePrediction...................................................................................... 59
22 Impact of Soil and Non-Aqueous Phase Partitioning onNaphthalene Biotransformation in an MGP Soil SlurryReactor.......................................................................................... 61
23 The Direct and Inverse Problems of Systems Analysis............. 66
24 Schematic Diagram of CSTR Experimental System.................. 71
25 Schematic Diagram of Modified Reactor with Metal FilterInsert.............................................................................................. 72
26 Bode Diagram Comparing CSTR Naphthalene ReactorLiquid Concentration Response to Induced Sinusoidal Per-turbations in Naphthalene Feed Concentration......................... 79
27 CSTR Response to Sinusoidal Perturbation in FeedNaphthalene Concentration......................................................... 85
28 CSTR Response to Sinusoidal Perturbations in Naphthalene •and Non-Naphthalene Organic Carbon Concentrations.......... 86
29 Total Heterotrophic and Gene Probe-Positive Bacteria in theCSTR............................................................................................. 88
30 CSTR Response to Sinusoidal Perturbation in FeedNaphthalene Concentration......................................................... 91
31 Naphthalene Biotransformation Rate vs. NaphthaleneReactor Liquid Concentrations.................................................... 92
32 Bode Frequency Response Diagram Comparing Naphtha-lene Biotransformation Response in CSTR and in ActivatedSludge Systems............................................................................ 97
33 System Time-Variance and Possible State Bifurcation Under8-Hour Period Perturbations........................................................ 98
ility and Sealt-up Protocol* for FINAL REPORTPolynuclaar Aromatic Hydrocarbon not VOLUME lBior*m«di»tion of MOP Soil* July 1991
34 Naphthalene Biotransformation Rate vs. Naphthalene FeedRate................................................................................................ 99
35 Radiolabeled Analysis of Naphthalene Fates in PerturbedCSTR Experiment ...................................................................... 100
FINAL REPORT Trtatability and Scalt-up Protocols forVOLUME I ttc Polynucltlf Aromatic HydrocarbonJuly 1991 Bioramadiation of MGP Soils
1.0 TREATABILITY AND SCALE-UP BIOREMEDIATION PROTOCOLS
1.1 PREDICTIVE RELIABILITY IN BIOREMEDIATIONThroughout this decade as bioremediation has emerged and been tested
as a potentially economical waste treatment alternative, difficulties in predictingor maintaining the predicted field-scale performance of these processes havefrom time-to-time been experienced. Questions relating to the nature of appro-priate biological testing of hazardous wastes to determine field-performanceparameters have been posed:1
Testing the commonly-held view that lab-scale studies are more controlled, thus morereliable for extrapolation to other systems, it is noted that: 1) Compound fates estab-lished in lab studies may be substantially different than in pilot-, full-, or even other tab-scale studies; 2) Positive biotransformation results found in small-scale lab systems areoften not reproduced in different systems or even in the same systems at different times;3) Where a given specific compound biotransformation is achieved at a larger scale, thekinetics are often unrelated to distributed measures of biomass; 4) Instantaneous bio-transformation rates vary widely and in an apparently stochastic manner even In "well-operated, steady-state' systems; 5) Effects of competing fate mechanisms such asvolatilization/stripping, sorption, and chemical reactions are not well understood at anyscale and reported biological removal of a compound often includes removal/conversionrelated to competing abiotic processes. Comparisons of kinetics across systems andwaste streams are confounded; 6) Competing abiotic mechanisms may be quiteconfiguration-specific and will likely scale differently than the biotransformation mecha-nism.
For owners of problematic MGP waste sites to accept bioremediation as atreatment alternative, it is critical for the users of the technology (engineeringfirms, environmental biotechnology vendor companies, etc.) to be able topredict the field-scale performance of bioremediation from the behavior ofvarious site samples with at least the same confidence as other technical treat-ment alternatives can be predicted and scaled. This requirement is in additionto the normal process selection requirements of process-specific effectiveness,process economy, safety, and so on.
Trtaiability and Scal*-up Protocol* tor FINAL REPORTPolynucltar Aromatic Hydrocarbon 1 VOLUME IBiortmtdiatton of MOP Soils July 1991
At a recent symposium on the topic a several professionals in the biore-mediation field were informally questioned about the acceptable error theybelieved sustainable in field-scale cost and performance predictions withoutputting them either at economic risk or impairing their competitiveness. Aninformal consensus was that accuracy in cost estimates of about ±20% wasneeded. For an in-situ or land-farming batch process, cost is roughly propor-tional to treatment time, therefore design treatment time should be accurate toabout ±20% (e.g., 5 months give or take one month).
A paucity of reported information and major confidentiality concerns aboutfield-scale predictive reliability complicates the assessment of whether the±20% target is currently being achieved. However, imprecision in one factorknown to contribute to inaccurate estimates is easy to document-chemicalanalysis in complex matrices. Recent studies with a number of MGP soils2
demonstrate that traditional PAH chemical analysis on replicates of the soilscan often vary 100-200% of the mean of the replicates. Likewise, others havereported coefficients of variation (the standard deviation divided by the mean)for biological soil treatment of coal gasification wastes for the PAH compoundsnaphthalene, acenapthene. phenanthrene, and benz(a)anthracene of between37-69% and 23-157% for the initial and final soil concentrations, respectively.3
While this effect has been attributed to non-homogeneous sampling difficulties,these variations directly affect the confidence intervals on rate parametersderived from these data. This will be discussed further in § 2.4.2.
The importance of the issue of improved predictive reliability for bioreme-diation testing is evidenced by a recent draft treatability protocol proposal cir-culated by the EPA. This proposal recognizes the need for nationalstandardization of a tiered treatability protocol approach beginning with alaboratory screening evaluation:4
• Environmental Biotechnology-Moving From the Flask to the Field, Symposium, University ofTennessee, Knoxville, J. W. Blackburn-Symposium Coordinator, October, 1990.
FINAL REPORT Trwub.lrty and Sc»I*-up Protocol* lorVOLUME I 2 Polynueltar Aromatic HydrocarbonJuly 1991 Biortmadiation of MGP Soils
• Laboratory screening is the first level of testing. It is used to establish the validity of atechnology to treat a waste. These studies are generally low cost (e.g., S10K-50K) andusually require hours to days to complete. They yield data that can be used as indica-tors of a technology's potential to meet performance goals and can identify operatingstandards for investigation during bench- or pilot-scale testing. They generate little, Hany, design or cost data and should not be used as the sole basis for selection of aremedy.
• Bench-seal* testing is the second level of testing. It is used to identify the technolo-gy's performance on a waste-specific basis for an operable unit. These studies gener-ally are of moderate cost (e.g. S50K-250K) and may require days to weeks to complete.They yield data that verify that the technology can meet expected cleanup goals andcan provide information in support of the detailed analysis of the alternative....
• Pilot-scale testing is the third level of testing. R is used to provide quantitative per-formance, cost, and design information for remediating an operable unit. This level oftesting also can produce data required to optimize performance. These studies are ofmoderate to high cost (S250K-1.000K) and may require weeks to months to complete.They yield data that verify performance to a higher degree than the bench-scale andprovide detailed design information. They are most often performed during the remedyimplementation phase of a site cleanup, although this level may be appropriate tosupport the remedy evaluation of innovative technologies.
This kind of treatability scale-up approach is a standard engineering pro-cedure long known and used in many engineering disciplines in order toscate-up physical, chemical and biological processes. Central to theapplication of this type of approach is the premise that changing the size orphysical dimensions of the test produces changes in biological system struc-ture requiring adjustments of equations or parameters in the design model.These might range from different expressions for the kinetic Rate Equations tosimple modifications of, say, one or more mass transfer constants. The needfor scale-up experiments, equations or protocols arises from our inability todescribe and solve the complete set of descriptive differential equations-oftenthe hydrodynamic flow, the equations describing ecological interactions, or theequations describing important cellular metabolic processes.5
A common assumption has been that the system structure remainsessentially the same over time. In other words, the design model applies a
Traatabilrty »nd Scalt-up Protocol* for FINAL REPORTPolynucltar Aromatic Hydrocarbon 3 VOLUME IBiortmtdiation of MOP Soils July
priori to the field-scale case and is time-invariant. Unfortunately, for mixed-culture biological systems of importance in environmental processes, theseassumptions-so well accepted in the application of physical, chemical orthermal environmental processes-may not generally apply to biological treat-ment systems. In an inherently changing system, the populations and pro-cesses are in a state of flux and the system's biodegradation activity may varysignificantly from moment-to-moment in either a well-operated reactor systemor a natural "environmental reactor" (e.g. land treatment, in-situ, etc.).6
1.1.1 DESIGN EQUATIONS AND RATE EQUATIONSOne of the realities for designers of any environmental treatment
process including environmental biotechnical processes is the necessity forpredicting the performance of the process in terms of the level of treatmentover a given time frame. This is needed in order to design the system, but isalso essential in estimating the costs of treatment.
Equations that describe the rate of conversion of a compound to aproduct or intermediate are called Rate Equations and must be known inorder to quantitatively describe and scale-up the conversion process. Gen-erally, these equations are based upon elementary rate models that oftendescribe chemical reactions. Elementary rate models are based onelementary reactions where the Rate Equation is suggested by a stoichio-metric chemical equation which represents the actual mode of action.7
Selection of a rate model and equation in the absence of publishedstudies that report the form of the rate model (in chemical processes,libraries of literature exist reporting the nature of the rate models for thou-sands of chemical processes) must follow the procedure of taking chemicalconversion data in a reactor of a known configuration and then fining theexperimental time-concentration profiles to elementary rate models until agood fit is achieved. If in a batch experiment, this happens to yield astraight-line fit of say:
FINAL REPORT Tr««t»bilrty and Seal*-up Protocol* lorVOLUME I 4 Polynucltar Aromatic HydrocarbonJuly 1991 Bior«m»di«tion of MQP Soils
, ( Concentration*) j Com pound A \ r, ,__-In -———————————-——TT-—————rr \versestlapsedTime\InitialConcentrationof Compound A J
...then the conversion reaction is said to be first-order in concentration. Simi-larly, the reaction can be zero-, second- or even fractional order with onlyminor variations in the data analysis.7 From now on in this report, the term"second-order" Rate Equation means a rate model first-order in bothcompound and biomass concentration. Formally, this type of rate model isknown as "second-order overall."
When there is no direct correspondence between stoichiometric equations and therate expression, we have a non-elementary reaction [and Rate Equation]....Nonele-mentary reactions are explained by assuming that what we observe as a singlereaction is in reality the overall effect of a sequence of elementary reactions. Thereason for observing only a single reaction rather than two or more elementary reac-tions is that the amount of intermediates formed is negligibly small and unmeasura-ble7
So even in chemical reaction analysis, it is often found that elementaryrate models do not adequately describe the conversion of a givencompound. Non-elementary rate behavior may be related to a network ofreaction steps made of elementary models in series and parallel, time lags inthe metabolic processes related to storage or absence of critical reactants,products or regulatory compounds, or to other types of nonlinearities or dis-continuities in the process. Obviously, in these cases, the ability to extractrate models diminishes with the ability to analyze metabolic intermediatesand to decode the complexities of the process mechanism and results inaggregation of mechanisms into more simplified compartments.
In the ultimate case where we haven't the experimental tools to under-stand the mechanism (a common event in chemical catalytic processes, forexample) then a purely empirical equation must be used for the rateequation. The sub-disciplines of systems analysis and process control have
Trot ability and Scalt-up Protocol* lor FINAL REPORTPolynuel«r Aromitic Hydrocarbon 5 VOLUME IBiorcmtdiation of MOP Soils Jury 1991
developed mathematical strategies for fitting a process response to an inputusing only the measured response to a measured input with the conversionprocess mode! remaining a "black-box." This approach may be applied tobioremediation systems and is discussed in § 3.0.
Once designers have a Rate Equation that describes compound con-version rates, it is incorporated into a Design Equation permitting the calcula-tion of time-concentration information based on the given reactortype-batch, continuous or one of many possible variations. DesignEquations are fundamentally derived from material balances specific to theactual system design using the following general approach:
Material Entering the System • Material Removed From the System =Rate*of Accumulation or Depletion of Material in the System
Examples of Rate Equations for zero- and first-order rate models andthe related Design Equations for batch and continuous (backmix) reactorsare presented Equations A.1-A.12 (Appendix A), tt is noted that more com-plicated forms may be required based upon the complexity of the rate mech-anisms as well as more complex and nonideal reactor configurations. Manysuch cases have been analyzed in the chemical reaction engineering designliterature.
The design and scale-up process now seems deceptivelystraightforward: first a Rate Equation and its related constants are selectedfrom the literature or are derived from an experiment and subsequent curvefitting, and then this equation is incorporated into the appropriate DesignEquation, and mathematically manipulated to provide a relationship betweenthe feed concentration (continuous case) or initial concentration (batch case)of the system and the effluent concentration (continuous) or time-concentration profile (batch) of the system.
FINAL REPORT Traatabilrty and Scala-up Protocols torVOLUME I 6 Polynucltar Aromatic HydrocarbonJuly 1991 Biortmadiation of MOP Soil*
A number of critical problems increase the error and reduce the predic-tive reliability of this generally correct design and scale-up approach as it isapplied to the problem of scaling-up environmental biotechnical processes.These include:
• The selection of the performance parameter to be predicted.
• The limited knowledge base on Rate Equations for environmental bio-technical processes.
• The general inadequacy of elementary rate models for describing in-vivometabolic activity.
• The extreme experimental difficulties in establishing mechanistic meta-bolic knowledge in environmentally-operating systems.
• The reality that, unlike most chemical processes, microbial processesadapt in response to environmental changes leading to Rate Equationsthat vary in structure with time (time-variance).
• The wide variability in experimental test kinetic results, even from thereplicate tests, complicating analysis, comparison and design.
• The ideal system Design Equations rarely exactly describe actualbehavior in real engineered or natural biotechnical reactors (includingin-situ reaction zones) and generally empirically-based scale-up predic-tive equations relating controlled system behavior to environmentaloperation must be employed.
• These scale-up predictive equations do not currently exist. Scale-up iscurrently performed based on heuristics and/or past experience.
Ttt»tability and Scalt-up Protocol* lor FINAL REPORTPolynuelaar Aromatic Hydrocarbon 7 VOLUME IBiortmtdialion of MOP Soils July 199:
1.1.2 PERFORMANCE PARAMETER SELECTIONIn the past, the selection of a compound or parameter for determining
system performance was based on the compound or parameter on whichthe system was regulated. Removal of non-specific parameters such as Bio-chemical Oxygen Demand (BOD), Chemical Oxygen Demand (COD), andTotal Organic Carbon (TOC) provided a convenient way to describe systemperformance-large removals of these non-specific parameters were recog-nized to result in an improvement in effluent quality. Further, while there wasgenerally a large scatter in experimental data leading to the selection of aRate Equation, it seemed that a given elementary rate model probablyserved as well as another, and the design process was thought to be insen-sitive to the formulation of the Rate Equation.
However, in today's world of concern about specific toxic compounds,system performance is increasingly specified in terms of removal of a specificcompound. The Rate Equation for the specific compound's conversion canbe estimated by a Rate Equation for a non-specific parameter if and only if -the conversion rate behavior of the specific compound and the non-specificparameter are similar.
The probability of the conversion behavior of specific and non-specificparameters not being similar can be illustrated by looking at a simplified cata-bolic (degradative) biochemical pathway for toluene (Figure 1). Toluene isknown to biotransform either to benzyl alcohol or to methyl catechol,depending on the bacterial organisms and genetics present and active in agiven system. Oxygen is incorporated into the initial conversion and intolater biotransformations at various points in the pathway. It is clear that mea-surements of the overall oxygen uptake of a microbial process with toluenepresent are averages taken over all of the specific oxygen-consumingbiotransformation steps.
FINAL REPORT Tr«*Ubilty snd Scal*-up Protocols forVOLUME I 9 Polynuclw Aromatic HydrocarbonJuly 1991 8<or«m*difttion of MQP Soils
The same is true for the evolution of carbon dioxide from the catabolismof a given compound (biochemical breaking-down of an organic compound).Denoted carbon mineralization, the overall rate of evolution of carbon dioxidefrom an organic parent compound is a weighted average of the individualrates of specific catabolic steps.
Is the measured average oxygen uptake rate or mineralization rateuseful in predicting the system's rate of conversion of toluene and thereforemeaningful in a Rate Equation to predict toluene biotransformation? It istempting to answer in the affirmative since non-specific parameters are mucheasier and less expensive to measure.
In any real system, the overall removal of toluene is described not onlyby biotransformation but by actions of abiotic fate mechanisms such as thestripping of toluene from the aqueous reactor liquid into the air in contactwith the system. As shown in Figure 2, the biotransformation of toluene toeither benzyl alcohol or methyl catechol changes the physical properties*(Henry's Law constant and solubility) of the compound dramatically. Benzylalcohol can be expected to strip at a rate about 10,000 times lower thantoluene in a given system and would exhibit an approximately 100-fold highersolubility (therefore lower octanol-water partition coefficient). Prediction ofthe toluene stripping from an operating system is an essential part of thescale-up process.
TrMtabilrty tnd Sc*lt-up Protocol! for FINAL REPORTPolynucl«r Aromatic Hydrocarbon 9 VOLUME IBiortmtdiation of MOP Soils July 1991
>^_ _ Jl int«rm«diottisttps)
COOHCOOH
—*OCttylCoA •*• SuCCinat t
Co*
CO,
r .^
Figure 1. A Simplified Biochemical Pathway for theMicrobial Conversion of Toluene
Stripping and biodegradation in an operating system are coupled, eachreducing the liquid concentration. In turn, the rate of each is usually relatedto the liquid concentration. The necessary biodegradation rate equationrequired to couple with the stripping rate is the one describing the biotrans-formation of toluene to benzyl alcohol or methyl catechol. Any otherdescribes a different reaction or group of reactions and does not describethe biological removal of toluene.
FINAL REPORTVOLUME IJuly 1991
10TfMltbility and Sc»!*-up Protocol* for
Porynucltir Aromatic HydrocarbonBiorcmadialion of MOP Soils
Some have suggested that the rate of oxygen uptake in the first bio-transformation of aromatic organic compounds may be the slowest biochem-ical oxygen uptake process in a pathway and may be rate-limiting. If so, andif an experiment can be designed to measure the initial oxygen uptakecorresponding only to the first biotransformation step, some suggest thatproportional oxygen uptake rate data may be collected.
Our work indicates that the biotransformation rate is often much fasterthan the mineralization rate-mineralization rates are often of the same orderas the oxygen uptake rates. Even when initial oxygen uptake biochemicalmetabolic transformation is rate-limiting, the overall oxygen uptake will beweighted by the other oxygen uptake biochemical metabolic transformationsunder way in the cells and may quantitatively vary widely from the overall ratemeasured. Designing a suitable experiment may be difficult.
In general, the parameter required for scale-up prediction of a system '$performance is the biotransformation rate of the parent compound to the firstmetabolite. In practice, this is arrived at by measuring the overall removal ofthe parent compound from the system and then correcting this for abioticprocesses either by measuring them or by estimating them with modelsverified for that test system.
A systems analysis/identification strategy sometimes called parameterestimation can be used to calculate biotransformation rates or constantsfrom a series of concentration and flow measurements. Parameter estima-tion will be discussed later (§3.1) in conjunction with dynamic analysis in acontinuous PAH-degrading system.
TrMtability and Seilt-up Protocols for FINAL REPORTPotynucltar Aromatic Hydrocarbon 11 VOLUME IBiortmtdiation of MOP Soils Jury 1991
I2
a •/»S *«
3
!§
-2 -efj-S .3
-4 •
s* s°5|R -6
S -7
•IOT*AN$FCRMATiON'S EFFECT ON SOLUB _TY
HOTRANSrORMAtlON'S EFFECT ON
•ENZVL 6EN2AIDEMTOEALCOHOL
METMTL CATtCMOL
4C1DCATECMO.
Figure 2. The Impact of Biotransformation on the Physical Properties andFates of Toluene Metabolites
Most data in the literature on microbial processes either make use ofnon-specific parameter kinetics or do not correct parent compound conver-sion data for abiotic processes arising in a given test method.8 Unfortu-nately, while it appears that the literature is rich with kinetic data, most of thereports have little utility for system scale-up and may even lead to greatlyerroneous scale-up performance predictions with no warning prior to field-scale failure to meet predicted performance. With current knowledge, treat-ability and scale-up experiments are needed for each application ofenvironmental biotechnology to PAH bioremediation in MGP soils.
FINAL REPORTVOLUME IJuly 1991
12TfMtabiltty and Sc«l«-up Protocol! for
PorynuclMf AromttJC HydrocarbonBiortmcdiation of MOP Soils
1.1.3 COMPLEXITIES OF CATABOLIC RATE BEHAVIORIn general, It is optimistic to expect elementary rate models to describe
catabolic (biodegradation) processes. While a given biotransformation mayfollow elementary behavior, at least in the axenic (pure culture) or cell-freesystems used by the microbiologists and biochemists, this conversion maybe under the influence of a wide variety of biochemical and molecular"control loops" in the environmental system. Also, time lags and dead timebetween enzymatic conversion steps can affect the apparent biotransforma-tion rates observed in the experiment. Networks of biotransformations canlead to dynamic and oscillatory behavior beyond description with simplemodels.5 In this context, the rate model is an approximation of observedbehavior. The prefix of "pseudo-" is often affixed to such an approximation(e.g., pseudo-first-order model) T
A characteristic of mathematical models of non-elementary and nonlin-ear rate processes and their networks is the potential for such processes topossess more than one "equilibrium" operating state. The mathematical termfor this effect is "bifurcation." Allowed to come to equilibrium, the system cansettle into one of several states, and when perturbed by outside influencescan shift to another operating state. It is possible, that a system steadilybiotransforming, say toluene could receive an upset in some environmentalcondition that shifts either the toluene-degrading organisms or the toluene-degrading pathway to another state with different biotransformation kineticbehavior. This a type of system time-variance.
Extrapolating rate behavior from systems operated in a typical labora-tory steady-state mode can be dangerous if the system does not happen tobe robust (resisting changes in activity in the face of environmentaldisturbances that the field-scale system may experience at the field scale).Dynamic response testing may be required if the tendency for upsets in bio-transformation rates is important. Dynamic microbial system analysis of bio-logical treatment systems at the laboratory scale was first demonstrated on
Trtatabtlrty and Sc»l*-up Protocol* (or FINAL REPORTPolynudtar Aromatic Hydrocarbon 13 VOLUME IBioramadiation of MOP Soil* Jury 1991
environmental biotechnical processes byBlackburn 9 .10 ..a PAH-degrading activated sludge system. Strongevidence that perturbations of naphthalene fed to this continuous system ledto a significant change in naphthalene biotransformation response wasreported.
1.1.4 SCALE-UP OF PERFORMANCE DATA FROM TEST SYSTEMSEngineering Design Equations for all chemical or biochemical conver-
sion processes (including MQP bioremediation systems) are based on idealphysical configurations that either represent or can be adjusted to representbehavior in engineered vessels. Reactor designers have needed to scale upfrom smaller-scale system results for decades and have relied on the devel-opment of "Scale-up" Equations to provide for estimates of rate behavior atdifferent configurations, sizes or operating conditions than the test. As anexample, transfer rates of oxygen between air and low viscosity liquids inagitated fermentation systems is given by Van't Riet 1 :
For a stirred vessel with water as the liquid and under coalescing agita-tion conditions, (V < 2600 L; 500 < P/V < 10.000 W/m3);
*,a-2.6xlO-2(£l (uOJ)05
For a stirred vessel with water as the liquid and under noncoalescingagitation conditions, (2 < V < 4400 L; 500 < P/V < 10,000 W/m3):
-3/^Yfc ,a-2.0xlO 3[-J 0 2
In the above equations, UgS represents the superficial gas velocity in thereactor in sec'1 and P/V is the power input per unit volume. Many such
FINAL REPORT Trtttabilrty and $eal»-up Protocol* forVOLUME I 14 Pofynuclear Aromatic HydrocarbonJuly 1991 Biortmtdittion of MQP Soils
equations are known for engineered systems and it is the responsibility ofthe designer to carefully select and apply these scale-up predictive equa-tions.
In principle, it should also be possible to develop similar "scale-up" pre-dictors that adjust biotransformation kinetics from a standardized test experi-ment to account for physical and configurational differences in a given fieldapplication. Although the development of scale-up predictors require bothlab- and field-scale results from which the empirical equations can beextracted (and no field-scale results were available to develop these equa-tions in this work), future marriage between treatability protocols and fieldtests should include scale-up predictor development as a priority.
1.2 OVERALL TREATABILITY AND SCALE-UP PROTOCOL CONCEPTTreatability protocols for contaminated soils have been reported and
reviewed in another part of the GRl Program.2 in the same work, a treatabilityprotocol was proposed. This was called the GRl Accelerated TreatabilityProtocol:
Given this picture of contaminant removal from soil systems, a crucial question arises:Does the historical treatability protocol, featuring soil pan or microcosm studies as itscenterpiece, provide the information needed to evaluate the ability of a soil to removecontaminants by biological mechanisms and does it provide this information in a timelyand cost-effective manner? Currently, soil pan or microcosm studies represent the bestavailable laboratory models for simulating the expected performance of land treatmentsystems. However, these studies typically take four to six months to conduct and six tonine months to complete when time for chemical analysis, data interpretation, and reportwriting are included. In addition, soil pan studies provide a gross measure of contam-inant removal, but do not provide information regarding the underlying processesinfluencing bioremediation of soils, such as information on partitioning of contaminantsbetween soil solids and soil water*
Testability and S»l*up Protocol* tor FINAL REPORTPolynuelt«f Aromatic Hydrocarbon 15 VOLUME tBiortmtdiation of MOP Soili July 1»9i
The key parts of the GRI Accelerated Treatability Protocol are:
• Soil Characterization: The soil is characterized physically to determine the particlesize distribution; chemically to determine concentrations of total organic carbon andchemicals-of-interest...; and microbialty to determine the total microbial populations andthe portion of this population capable of degrading PAH compounds. Characterizationprovides essential information on soil texture and contaminant loading.
• Desorotion Testing: Equilibrium desorption of the chemicals-of-interest from the solidto the aqueous phase is determined, resulting in an estimate of the equilibrium partitioncoefficient, Kp. tf the concentrations of chemicals-of-interest in the aqueous phase arehigh enough, then contaminant removal by biodegradation is likely. Estimates of Kp areuseful in assessing the transport and fate of chemicals-of-interest in the soil, as well asassessing the treatability of the soil.
• Slurry Reactor Testing: The potential treatment endpoint for the soil after biologicalprocessing is evaluated in [batch] slurry reactor testing. The [batch] slurry reactorprovides an optimal environment for contaminant removal from soil because thecontents are agitated, fostering transfer of chemicals from the soil to aqueous phase,and the reactor is highly aerated. Although rate information is provided by these tests, *the primary emphasis is comparing concentration of chemicals-of-interest in the soil atthe beginning and end of the study.2
The protocol's limitations were also noted:
While it is important to recognize the advantages of the GRI Accelerated TreatabilityProtocol, it is also important to recognize the limitations of this protocol and where signif-icant knowledge gaps still exist. First, although rate data are provided by the [batch]slurry reactor, these data cannot be translated to field conditions at this time. Second,and more importantly, the assumption that the treatment endpoints obtained in a[batch] slurry reactor will mimic those that can be obtained in the field (or a soil panexperiment) remains a largely unverified supposition....2
The GRI Accelerated Treatability Protocol, developed by ReTeC, Inc., hasas its objectives 1) the inexpensive screening of MGP soils to identify the bestoptions for field-scale testing and 2} to define the "endpoint" concentrationsthat can be achieved by batch bioremediation processes. The limitations
FINAL REPORT Tr«tabil«y tnd Sc«l#-up Protocol* lorVOLUME t 16 Polynudtar Aromatic HydrocarbonJuly 1991 Biortmtdiation of MOP Soils
include the inability to scale the removal kinetics to reliably predict full-scaleperformance and uncertainty whether the experimental endpoint concentra-tions really match field-scale results.
In this report, an overall scheme for integrating the screening-type tests(like the GRI Accelerated Treatability Protocol) with predictive scale-up capabil-ity is proposed. Figure 3 presents a generalized schematic diagram of thenetwork of relationships required for both treatability and scale-up predictions.The environment and process is generalized to represent tests and trials at allscales. To determine both kinetics and the impact of perturbations on thekinetics, the operating test or system is either monitored in time-series, orsamples of "biomass" are taken periodically and assayed offline.
Like the Accelerated Protocol, separate, standardized tests are used toquantify both the desorption/diffusion behavior and the intrinsic biotransforma-tion kinetic potentiaf (1BKP). The IBKP is a measure of the biotransformationrate of the PAH available for catabolism in the aqueous reactor liquid or soilwater. Unlike the Accelerated Protocol, emphasis is placed on the rates ofdesorption/diffusion and biotransformation potential with the hypothesis thatfour cases require evaluation for the selection of PAH bioremediation pro-cesses:
• Both desorption/diffusion and IBKP rates are greater than some minimumcriterion leading to a recommendation for bioremediation.
• Desorption/diffusion rates are below and IBKP rates are above someminimum criterion leading to scale-up predictions based on the desorp-tion/diffusion rates. Recommendation depends upon the performanceprediction.
Trtatabilrty and Scala-up Protocols for FINAL REPORTPorynuelaar Aromatic Hydrocarbon 17 VOLUME IBiort mediation of MOP Soils Jury 1991
• Desorption/diffusion rates are above and IBKP rates are below someminimum criterion leading to scale-up predictions based on the IBKPrates. Recommendation depends upon the performance prediction,
• Both desorption/diffusion rates and IBKP rates are below some minimumcriterion leading to coupled scale-up performance predictions and verypossibly an early recommendation against bioremediation.
Also differing from the Accelerated Protocol, the batch IBKP test devel-oped in this work makes use of radiolabeled PAH analysis for greater precisionin the time-concentration profile and the continuous IBKP test is designed to beable to evaluate the impact of environmental perturbations on the PAH bio-transformation kinetics as well as to enable the extraction of a type of processmodel directly from the perturbation-response data. These issues arediscussed in § 2.4.2.
FINAL REPORT TruUbiltty and Scil*-up Protocols (orVOLUME I IB Porynuclaar Aromatic HydrocarbonJuly 1991 Biortmtdiation of MGP Soili
CMVIftOMUCMI AMD MOCCSSflASK O» REACTo* 1CSV UIC*OC05H
i»1Cf MA t*
SECOND PRIORITY
Figure 3. A Concept for Reliable Treatability and Scale-up Protocolsfor PAH Biotransformation in MGP Soils
Once the treatability protocol relationship is confirmed for use relatingMGP soils, a scale-up predictor relating the rates in the standardized tests to afield-scale application can be developed and used to predict biotransformationperformance in the specific field-scale case. The critical component of thisphase of work is to make certain that the standardized treatability protocol isimplemented at the field demonstration scale as well as at the lab-scale so thatthe empirical scale-up relationship can be extracted from the data.
Two biotransformation test procedures are reported in this work-onebased on small, batch vials and another based on dynamically-perturbed con-
»nd Sealt-up Protocols (orPolynucltar Afomclic HydrocarbonBioftmtdulion of MGP Soils
19FINAL REPORT
VOLUMEiJury 1991
tinuous soil slurry reactor systems. The batch method is discussed in § 2.0while the continuous and dynamic method will be discussed in § 3.0. Bothapproaches are idealized in several ways and are likely to reveal optimal intrin-sic PAH biotransformation kinetic potential (IBKP) where the slow desorp-tion/diffusion process of PAH into or from non-aqueous phases in thecontaminated soil is experimentally minimized because of the way the test isrun (Figure 4). These IBKP biotransformation kinetics may be related fromMGP soil to MGP soil, but are expected to be fast relative to actual field-scaletrials using land-farming, in situ approaches, bioreactor or other field configura-tions. Therefore, further desorptionfdiffusion and scale-up protocols areneeded in order to reliably use the idealized laboratory-based information toestimate field-scale performance.
//
V\
x£^~
H9imom*G
\
t >«•t*tt TIM
— T*1
1
\»
/'"^%•O"-*WJ(OJS »»«H
•*f1IHOWMC
^ x^ ^"
>.\
j"J
J?CJ
-:Mutewi
M «t»
:--*M*«
i-""
\\\
1
/
/
Figure 4. Concept for a Laboratory-Based Treatability Protocolfor MGP Soils
Radiolabeled analytical approaches for the batch tests are superior toconventional compound analysis for MGP soils and possibly for other highlycontaminated materials. This comparative fact will be discussed in § 2.4.2. Thebatch IBKP method is based on the use of a radiolabeled parent compoundand the estimation of its biotransformation by measuring the label in a hexane-extract phase. The continuous IBKP method makes use of highly sensitivededicated chemicat analysis of the PAH in the aqueous liquid and does notrequire a radiolabel for routine analysis.
FINAL REPORTVOLUME IJuly 1991
Traatability and Scalt-up Protocols lorPolynudaar Aromatic Hydrocarbon
Bioramtdiation of M3P Soils
Neither the batch nor the continuous 1BKP method includes effects oflong-dynamic rate-limiting processes, such as the slow nonaqueous phasecompartment PAH desorption/diffusion process. This biotransformation limita-tion can be determined with a lab-scale abiotic mass-transfer type test on agiven type of MGP soil. The GRI Accelerated Treatability Protocol includes anabiotic test to determine the solid-liquid PAH partition coefficient. A variation ofthe GRI Accelerated Protocol to determine the kinetics of the slow componentof the desorption/diffusion process is suggested:
• Select several organic test compounds where biotransformation in MGPsoils is very slow or can be completely inhibited by some experimentalmeans.
• Obtain labeled forms of these test compounds so that the addedcompound can be discriminated from like unlabeled indigenous com-pounds.
• Add these labeled compounds to the inhibited MGP soil of interest in astandardized fashion and allow to set over a standardized time frame.
• Add this preloaded soil sample to a well-mixed test system, possibly withan aqueous phase present and conduct time-series analysts for the testcompound to determine the overall desorption/diffusion rate behavior.
A quite different approach to determine the kinetics of the fast componentof the desorption/diffusion process is discussed later in § 3.2.3. This informa-tion is needed in order to employ parameter estimation of biotransformationrates in the continuous IBKP test.
The batch IBKP tests can be used to predict overall PAH biotransforma-tion (first and last sampling points) but not of the actual time-concentrationprofile. As will be seen in a later section (§ 2.2), the actual batch
Tr«t«bilrty and Sctlt-up Protocol* for FINAL REPORTPolynuclt»r Aromatic Hydrocarbon 21 VOLUME IBiorcmtdiation of MOP Soil* July 1991
biotransformation rate profile data is often not first-order and can deviate eitherpositively or negatively from the overall first-order rate model assumption. Theuse of time-series sampling and analysis techniques (a number of samplestaken at known intervals throughout the test periods) will aid in characterizationof these profiles. With attention to the potential causes of variable non-first-order rate behavior, time-series sampling can be expected to lead to betterpredictive understanding of the nature of a MGP soil's deviation from first-orderand the resulting improvement in selecting time-variant Rate Equations.
FINAL REPORT TrttUbility »nd Sc«l*-up Protocols lorVOLUME I 22 Potynucl«ar Aiomatic HydrocarbonJuly 1991 BiOf«m«di«tion of MOP Soils
2.0 A BATCH INTRINSIC BIOTRANSFORMATION KINETIC POTENTIALTEST
Two basic types of laboratory-scale batch kinetic tests are discussed in thissection-one for MGP soils and another for axenic (pure culture) organisms. Inthe first, 2 g samples of actual MGP site soils were placed into several 25 mlvials along with 1 ml of a buffer solution. To this mixture, a solution of14C-labeled naphthalene, phenanthrene, or anthracene in acetone carrier was addedand the mixture was incubated at 26°C and shaken. After a period of time,triplicates of the vials were harvested for analysis by addition of acid with collec-tion of 14CC>2 A hexane:isopropanol solution was added to the soil to extractthe parent compound along with non-polar metabolites while the polarmetabolites partitioned into the aqueous/alcohol phase. Finally the remaining soilwas air dried and oxidized in a thermal oxidizer to attempt to quantify non-extractable label. All liquid phases were counted for radioactive decay in a liquidscintillation counter. Further details of the experimental procedure can be foundin experimental descriptions T1CS-08A, T1CS-08N, T1CS-08P, T1CS-27A,T1CS-27N, T1CS-27P, T1CS-31A, T1CS-31P, T1CS-31N. T1CS-51A, T1CS-51N.T1CS-51P, T1CS-97N, and T1CS-97P in Appendix B or in relatedreferences.12.13
The organism axenic test is identical to the soil test except that a knownaxenic inoculum was added in place of the MGP soil sample. A characterizationof the contaminated MGP soils for pH, total organic carbon, total PAHs andparticle size distribution is provided in Table 1. These analyses were provided byReTeC, Inc.2
Tr«atabi1rty and Scal*up Protocols lor FINAL REPORTPolyriuclaar Aromatic Hydrocarbon 23 VOLUME IBiortmtdiation of MQP Soil! July 1991
Table 1. Characterization of the Contaminated MGP Soils Tested3
Soil
Soil A
SoilB(8)
SoilC(97)
SoilD(51)
SoilE(31)
DateCollected
4/20/69
3/27/89
10/12/89
unknown
11/16/88
PH
7.1
8.1
9.0
8.0
7.0
TotalOrganicCarbon(mg/kg)
426,016
5,951
32,883
65,118
4,189
TotalPAHs
(yg/kg)
3,729,906
233,193
3.621,013
334,316
454,097
PercentSilt
4
2
38
17
3
PercentSands
75
70
59
70
79
PercentGravel
21
28
3
12
18
8 Ref. 2. Significant figures as reported.& Numerical designations used in other QRI reports.
2.1 RESULTS OF LABORATORY-SCALE BATCH IBKP EVALUATIONS
2.1.1 MICROBIOLOGICAL ANALYSESAn early hypothesis of this work was that biotransformation kinetic pre-
dictions could be improved if the rate model assumed was second-order,that is, first-order in PAH concentration and first-order in genotypeconcentration (the organisms present possessing the genes necessary forbiotransformation of PAHs). Samples of MQP sites were analyzed for geno-types carrying the NAH7 plasmid as well as for other microbial assays. Themethods use are reported elsewhere.12»13 Table 2 presents the results ofthese microbiological assays.
FINAL REPORT Traatability and Scalt-up Protocols torVOLUME I 24 Potynuelaar Aromatic HydrocarbonJuly 1991 Biortmtdiatton of MQP Soils
Table 2. Results of MGP Soil Microbiological Analyses
Soil
Soil A(27)'SoilB
(8)SoilC(97)
SoilD(51)
SoilE(31)
• Numerical
Total Heterotrophic Bacteria nah(cfu/g MGP Soil)
5.6±0.8x106
2.2±0.3x106
1.2±0.2x108
7.8±1.7x106
1.0± 0.3x1 07
designations used in some GRI reports.
Gene Probe Positive Bacteria(cfu/g MGP Soil)
1.3± 0.4x1 06
5.9+0.8x1 05
3.3*0.5x1 06
2.1±0.5x105
<2.0x106
2.1.2 PHENANTHRENE BIOTRANSFORMATION AND MINERALIZATIONFigures 5-7 present the kinetic experimental data from the batch kinetic
tests for phenanthrene biotransformation and mineralization in Soi!s,A-E. Theabscissas on these figures are not always divided into equal time intervalsbetween samples and therefore the overall shape of the time-concentrationprofiles on these figures has no meaning.
The 14C-phenanthrene in the Hexane Extract (confirmed by liquid chro-matography) is taken as the concentration used in all biotransformationkinetic calculations. The Radiolabeled Carbon Dioxide recovered was usedto calculate mineralization kinetics. The Water Extract was the aqueousphase left after the hexane:isopropanol extraction and was thought tocontain water-soluble metabolites of the phenanthrene biotransformation.The soils after extraction were combusted in a thermal oxidizer and theresulting 14CO2 was included in the Soil Immobilized Radiotabel Category.The sum of all radiolabel recovered was included in the Total RadiolabelRecovered Category.
Trtatibility and Sc»i«-up Protocols for FINAL REPORTPolynuc)«r Aromatic Hydrocarbon 25 VOLUME >Biortmtdiation of MOP Soils July 1901
o 1*0
NoU: lin*t ihilttd tligMly le aid vi«» of error bar
96 120 144ELAPSED TIME (HR)
192 240
PANEL 8, SOIL 8
Total *8d,oob»l R«C«»tr«a
H«.on» Ci'rotl (Bh«no"Ihrtn«)
rbor Dioiid* ft*c»v«rtd11,9111,, -e Q.C *>t» ol tfro
Sell
12 t5* 24 28 32 36 48ELAPSED TIME (HR)
a 0 3 6• *... (*•••>•• k.*. ^ •*»» »••"Total Radi«lob«l fi«c»*«r«d
Citroct (^h*nqrthr*i«)
Sod'«iebt'»d Carbon 0*««id« R«co*tf«d"~ Soil immabiiiMd—— W«ltf Cilroci (Polor Utiobaliltl)
Figure 5. Results from Radiolabeled Phenanthrene Batch MGPSoil Fate Tests, MGP Soils A and B
FINAL REPORTVOLUME IJuly 1991
26and Scalt-up Protocol* tor
Polynuclcar Aromatic HydrocarbonBtoramtdiation at MOP Soils
O
140
ICO
ma «5<* 40u.O
Z 20
PANEL A, SOIL C
(*»»». ftMi»l*fe« f•(•*»'•« •• t»*b»" 4*1 <4« ••<« !fl lh« ••1«•tir«cl rong* * **lu» ««*•' 1n« ioil li
48 96 144tLAPSED TIME (HR)
19? 240
Ne*» L>««« f <i«e tnh'h tt e-d *i«» or error ban
t « 12
ELAPSED TIME (HR)48
Htiant If tret I
IMIMMtfM)
Figure 6. Results from Radiolabeled Phenanthrene Batch MGPSoil Fate Tests, MGP Soils C and D
Trtttability and Se»l**up Pretoeols forPolynuelttr Aiomalic Hydr&ctrbonBiortmtdiation el MGP Soils
27FINAL REPORT
VOLUME tJuly 1991
1*0
oO 120
100
eowCD
3 605* 40
20
SOIL E
0' 3" *" 12* 2** 30»..,. <••-,. •*.•.« — ..."•» ELAPSED TIME (HR)
3* 7 2 *
Figure 7. Results from Radiolabeled Phenanthrene Batch MGPSoil Fate Tests, MGP Soil E
All tests demonstrated significant biotransformation and mineralizationexcept the one for Soil C (Figure 6, Panel A). Here, all indications supportedthe observation of no statistically-significant biological conversion of phenan-threne.
One concern regarding this protocol was that the radiolabeledcompound was added to the experimental vials dissolved in small volumes ofacetone. While the volume of acetone was less than 2 M. L (approaching thetower limit of liquid that can be reproducibly manually handled), it was a pureorganic liquid and was added to an experiment containing 2 ml of aqueousbuffer. Thus, the amount of organic carbon available for biological uptakewas a significant fraction of the dissolved organic carbon in the aqueousextract from the MGP soil. This level of added exogenous carbon mightinhibit or somehow interfere with PAH biotransformation.
FINAL REPORTVOLUME IJuly 1991
TfMttbility and Seal*-up Protocols torPotynuelttr Aromatic Hydrocarbon
Biortmtdiation of MOP Soil*
An additional experiment was performed where MGP soils A and B weretested using the batch IBKP protocol with the modification of varyingamounts of acetone added as the sample carrier (0.5 • 10 u. L) for 14C-naphthalene. Soil A was used as an example of high organic MGP soil andSoil B was used as an example of a low organic MGP soil. With threereplicates of each test, the average coefficients of variations of 14CC>2 gener-ated for Soil A and Soil B were 0.250 and 0.236, respectively. The results ofthis experiment, expressed as a percent of the lowest level of acetone addedis presented in Table 3.
Table 3. Effect of Acetone Solvent Carrier on the Cumulative MineralizationRate of 14C-Naphthalene in Batch IBKP Tests.
Volume of Acetone Carrier Added
Elapsed Time (hr) 2.0 ML 5.0 ti L 10.0 n L
MGP Soil A-High Organic Soil
04
B12
0.9831.B6
0.8550.841
1.261.49
0.7390.872
*
0.8541.04
0.6850.687
MGP Soil 5-Low Organic Soil
04812
n.iw1.071.11
0.970
3.870.8340.8320.930
0.7130.7590.9000.963
* n.a.: not available.
The results of this experiment indicate that acetone carriers for theradiolabel might have an effect on the mineralization kinetics for the MGPsoils with higher organic content. For these soils, the mineralization ratesearly in the test may be increased with moderate additions of acetone, and a
TrMUbility «nd Sc«lt-up Protocols lor FINAL REPORTPotynucl«r Aromatic Hydrocarbon 29 VOLUME IBiottmtdittion of MGP Soils Juty 1991
possible inhibitory effect later in the experiment for higher additions ofacetone. Generally negligible effects were seen in Soil B. Nearly all effectsare small in comparison to the coefficients of variation in the analysis itself.Therefore, the acetone carrier addition is not expected to have a major influ-ence on the results and conclusions presented in this report.
General advantages of radiolabel experiments include the ability todetermine the fates of compounds that may already be present in thesamples without analytical interference and the ability to provide a radiolabelmaterial balance on the labeled element. Throughout our effort to performbatch kinetic soil experiments we have observed that in most cases, theability to recover the label from MGP soils diminishes as the batch experi-ments proceed. This effect is demonstrated here in Soils A. B, and D asdiminishing Total Radiolabel Recovered.
Possibilities for the variation in the 14C recovery was incompletedesorption/diffusion and/or combustion in the oxidizer. Additional controls ,were set-up to determine if acidification and/or hexane extraction somehowaffected desorption/diffusion of the 14C- PAH from the soil. Four treatmentswere set-up with seven MGP soils (the original five, A-E, and two additionaldesignated Gand N):
• Samples were acidified, incubated, and oxidized.• Samples were acidified, incubated, hexane-extracted, and oxidized.• Samples were treated with azide, incubated, and oxidized.• Samples were treated with azide, incubated, hexane-extracted,
and oxidized.
Two g of soil were mixed with 1 ml of a minimal salts buffer in a 25 mlscrew-capped poty-TFE-sealed mineralization vial as described previously.14C-Anthracene (56.600 radioactive disintegrations per minute, dpm) wasadded to each soil. Samples were treated with 0.5 ml of 2 N H2SO4 or 0.1 %
FINAL REPORT Traatabilrty and Scala*up Protocols (orVOLUME I 30 Polynuc'**' Aromatic HydrocarbonJuly 1991 Bioramadiation of MOP Soils
sodium azide (final concentration) to inhibit biotransformations. The hexaneextraction procedure has been described previously. Thermal oxidationtimes were either 4 or 8 minutes.
There was little or no difference in 14C-recoveries for hexane extractedtreatments. Black carbon deposits were visible after oxidation on soils A, D,G, and N implying that oxidation was not complete. Inhibiting the samples by0.1% azide or by acidification did not affect 14C recoveries.
In non-hexane extracted samples, oxidation did not yield 100% recover-ies. Increasing the oxidation time from 4 min to 8 min yielded higher 14C-recoveries in only two cases. These data indicate that the thermal oxidationcombustion process for these samples is inefficient, possibly due to thecarbonization of the organic carbon present during heating and the captureof the radiolabel in this carbonized matrix. Other possible tosses of 14Cinclude loss of 14CC>2 due to a saturated C02 trap, a poor seal or volatiliza-tion of the 14C to the vial headspace. *
Based on the precision determined for the replicate analysis, the datagenerated on radiolabel biotransformation, carbon dioxide production andpolar metabolites in the water extract are considered reliable. Developmentof a more suitable thermal oxidation method that permits the oxidation of thecarbonized phase is needed in order to close the material balance onMGP-type soils.
For easy visual comparison of phenanthrene biodegradation from soil-to-soil, Figure 8 presents the rates of 14C removal from the hexane extractand generation as COg for Soils A, B, D, and E, with the same scales oneach axis. The data for removal from the hexane extract phase is assumedto be the result of biotransformation, while the generation of carbon dioxideis mineralization. The shapes of these profiles allow comparison of whetherthe conversion process is rapid or gradual from soil to soil. Also, the com-
TnftUbilNy and Sealt-up Prole-colt (or FINAL REPORTPotynuclti' Aromatic Hydrocarbon 31 VOLUME IBiortmtditlion of MQP Soils July
parison of the biotransformation rate with the mineralization rate permits theassessment of whether there are metabolic products forming andaccumulating in the process, since the overall rate of mineralization shouldbe equal to the overall rate of biotransformation if no radiolabeled metabo-lites accumulate.
Equation A. 19 can be use to calculate pseudo-first-order biotransforma-tion rate constants from the hexane extract concentration data and EquationA.20 can likewise be used to calculate pseudo-first-order mineralization rateconstants. These are presented in Figure 9 for Soils A-E, as welt as for threenon-soil axenic (pure) cultures of phenanthrene degraders. These are Pseu-domonas putida G7 (Panel F), DFC50. (Panel G), and DFC 49 (Panel H). P.putida G7 is the strain harboring the NAH7 plasmid which is a geneticelement coding for naphthalene degradation and is used as a gene probefor PAH degradation in this work. The other strains were isolated from MGPsoils early in the study and have been found to degrade PAHs. Note thataxes are not scaled for direct visual comparison from panel to panel, rathereach panel is independently scaled to permit close examination of the ratebehavior.
The effect of using the pseudo-first-order rate constant (Figure 9)instead of the rate itself (Figure 8) is to "normalize" the rate behavior to thehexane extract concentration present at the time that rate is measured.Thus, high pseudo-rate constants can exist late in the experiment when thehexane extract concentration is low, and where rates would be necessarily ~"low because of limited reactant (parent compound) present. The pseudo-first-order rate constant is also a test of whether or not the biotransformationor mineralization process is actually first-order. If it is first-order, the rateconstant trace should be a straight horizontal line in Figure 9. Departurefrom this is evidence that the conversion processes are not first-order and inthese cases indeed, vary with time. Phenanthrene biotransformation in thesebatch tests is non-elementary and/or time-variant.
FINAL REPORT Traatabilrty and Scal*-up Protocols lorVOLUME t 32 Porynucltar Aromatic HydrocarbonJuly 1991 Biortmtdiation of MGP Soil*
400 <
300-
230-
too-e-
-100-
-200-
.too--400-
I
400
too200
100
0
-100
-20C
-15C
PANEL A, sou. A
120 iio 200 i4C
PANEL e. SOL I
**200
too0
-100'
-200
-WO
Y'"J PANEL B. SOIL D'
-400- ———— • —— • ———— • ———— • —— . —— . —— - —————0 *C H 120 UC 200 24
JOG
MO'
100 '
-J-200 PANEL D. SOIL c
0 40 1C 120 1*5 20C 240 0 40 10 12C IW 101 2*0
CLASSED TME (HP) ELAPSED TiMt (MR)
HOT»4H$fO*M*TlON -———— MINERALIZATION ——————-
Figure 8. Rates of Radiotabeled Phenanthrene Biotransformationand Mineralization on Batch MGP Soil Tests
Figure 10 presents the time-concentration profile for phenanthrene bio-transformation batch testing of MGP soils and axenic organisms. For sim-plicity of comparison, an overall first-order assumption is made andEquations A. 19 and A.20 are used with the initial and final concentrations, afirst-order Rate Equation results with behavior shown in Figure 10. Departurefrom first-order behavior is again highlighted in this figure.
While the complex rate behavior is not modelled well with first-orderRate Equations, interval pseudo-first-order rate constants can be used in amodel that assumes that first-order behavior exists between each measuredpoint. The time-concentration profile can then be calculated beginning withthe initial concentration then recursively to the last. The use of this sort of arecursive predictor using pseudo-first-order rate constants calculated for .each sampling interval (Equations A.14 and A.15) is shown (Figure 10) as
Testability and Scal*-up Protocol* (orPolynucltar Aromatic HydrocarbonBioromadialion of MOP Soils
FINAL REPORTVOLUME IJuly 1991
long dashes slightly offset from the actual concentration data. While capableof accurately reproducing the observed concentration profile, this predictiveapproach is limited by the need to calculate and store as much rate constantinformation as the original concentration information. In other words, thismodel offers no generalization potential and therefore no advantage over useof the concentration data alone.
Other strategies exist to model non-elementary or time-variant ratebehavior. One approach might be to use a pattern search approach forgroups of data that fit an elementary model over a given interval andsequence these segments together for an overall Rate Equation. The recur-sive approach could then be employed with the advantage of generalization.Another approach might be to handle the rate constants as parameterdistributions, with time being the independent variable. These distributionscould be modelled with a variety of distributions or functions (e.g., ramp,Poisson, Weibull, exponential, normal, etc.). A worthy topic for future work isto formalize an artificial intelligent system to analyze complex rate behaviorand generate the "best fit" for experimental data for use as predictive RateEquations.
Overall pseudo-first-order rate constants8 calculated from Equations
• Nomenclature for kinetic rate constants varies widely in the literature. The sign of the rateconstant depends upon the formulation of the material balance and the definition of production.Classical chemical rate kinetics usually assume rates for chemical production as positive andrelated rate constants as also positive. Removals can be shown as negative rate terms or aspositive rate terms and negative rate constants. The equations presented in Appendix A showremovals as negative rate terms and therefore rate constants, so calculated should be positive.
However, in this work, biotransformation rate constants, /c* are reported as negative values todiscriminate them from mineralization rate constants, *m (reported as positive values), ft ishoped that the loss of consistency is less problematic to the reader than the increase in clarityfrom not confusing the two types of rate constants reported in this work. Further description ofnomenclature used to designate rate constants may be found in Appendix A, I A.6.
FINAL REPORT Tr«tabtlity and Sealt-up Protocols lorVOLUME I 34 Polynuelaar Aromatic HydrocarbonJuly 1991 Biortmtdiation of MGP Soils
A.19 and A.20 are presented in Panel A of Figure 11. Overall pseudo-second-order rate constants based on Total Heterotrophic Bacteria PlateCounts, klh and /c J1", are presented in Panel B. Overallpseudo-second-order rate constants based on nah gene probe-positivebacteria concentration, /c|p and * Jp, are presented in Panel C of Figure 11.
Both second-order rate constants calculated with total heterotrophicbacteria and gene probe technology are shown as points of reference tocurrent research activity by other groups, ft seems clear that similar but notidentical response patterns in second-order rate constants appear between*S" and ky. As noted in a later discussion, § 2.3, the best predictive relation-ships for the batch test systems make use of the rate constant derived fromgene probe technology (k y) as a measure of active PAH-degradingbiomass.
A generalized analysis of error propagation presented in a subsequentsection (§ 2.4.1, Figure 20) may be used to estimate the typical range of errorin the first-order rate constants presented in Figure 11. Error in these con-stants depends upon the error in four other parameters: the overall conver-sion of the parent compound, the error in the analytical measurements of theinitial and final chemical concentrations, and the error in the measurement oftime between the initial and final samples.
As seen in a later section (§ 2.4.2), a typical relative standard error in theradiolabel concentration measurements is around ±10% (although poten-tially much greater for conventional analyses). As noted in experimentaldescription in Appendix B, parent compound conversions exceed 60% andare often greater than 90% in the tests. If relative standard error in the timemeasurement is conservatively estimated as ±10%. then the relativestandard error in the pseudo-first-order rate constants (Figure 11. Panel A)should not exceed ±20% and may generally be closer to ±10%.
Trtatabitity and Scalt-up Protocol* lor FINAL REPORTPolynuclaar Aromatic Hydrocarbon 35 VOLUME IBiorimediation of MQP Soils July 1991
o
iOo
0.1o
-O.i
-0.2
-0.3
-O.*
0.1
0.05
0
-0.05
-0.1
PANEL A, SOIL A
"TOC 150 200 250 500 ~4
\PANEL C. SOIL C
50 ICO 150 200 250 300 50 60
PANE. E, SOIL E
"0 10 2: iO *0 50 6C 70 80
5 10 '5 20 25 ~~30ELAPSED TIME (HR)
INTERVAL PSEUDO-FIRST-ORDERBIOTRANSFORMATION RATE CONSTANT
10 20 30 43 50 60
5 10 15 20 25 30ELAPSED TIME (HR)
INTERVAL PSEUDO-FIRST-ORDERMINERALIZATION RATE CONSTANT
Figure 9. Interval-Based Pseudo-First-Order Rate Constants forRadiolabeled Phenanthrene Biotransformation and Mineralization in Batch
MGP Soil and Axenic Organism Tests
FINAL REPORTVOLUME IJuly 1991
36Trtattbilfty and Se»l«-up Protocol* lor
Polynuctttr Aromatic HydrocarbonBiortmtdiation of MGP Soils
3o.o
uot
zo
uzou
"•V, PANEL A, SOIL A
SO 100 150 200 2SO ICC 10 20 10 40 SO 60
PANEL C. SOiufC PANEL D. SOIL 0
5: 100 150 200 250 3CC 10 20 30 40 SO 6:
PANEL r.P PVT1DA C
)0 20 30 40 SO 60 70 10 20 30 40 SC 6C
c. STRAIN orcso PANEL K STRAIN DFC45
S 10 IS 2C 25 50ELAPSED TIME (MR)
S 10 15 20ELAPSED TIME (HR)
RECURSIVE PREDICTOR MEASURED OVERALL FIRST-ORDER
Figure 10. Actual and Predicted Radiolabeled Phenanthrene Concentrationsin the Hexane Extract from Batch MQP Soil and Axenic Organism Tests
Tr*ttability and Scal**up Ptotoeol* lorPolynuelctr Aromatic HydrocarbonBiortmvdialion of MQP Soils
37FINAL REPORT
VOLUME IJuly 1991
s c
I!3 VI
21
4. OVCIULL «CuOO-ntST-ollDC» »ATC COMSTAMS
0.1
0
-O.I
-0.2
-0.3
-0.4
-o.s-0.6
-0.7
-08
i
PANEL I. OVEHA.L MEU»-SECOND-0«£1 "*TE COKS'ANTHCTCKOTHO'HIC lACTt*'*
Se I ASo I
Soil C Soil £ 0CCSGSei> 0 f> put C? OFC*I
RATt
ngh CCNE MOII-^OSITIVE lACTERlA
So *Scii B
c so. t . „Soil o f put. cr
Figure 11. Overall Pseudo-First- and Second-Order Radiolabeled Phenan-threne Rate Constants from Batch MGP Soil and Axenic Organism Tests
Second-order constants were calculated by dividing the first-orderconstant by the appropriate microbial concentration from Table 2. Withrelative standard errors of from ± 15-30% and with the general rule of errorpropagation, "where multiplication or division is involved, the relative determi-nate errors are transmitted directly to the result."14 standard errors of thepseudo-second-order rate constants should range from ±15-30%. It isinteresting to note that the major source of error in the second-order calcula-
FINAL REPORTVOLUME IJuty 1991
Trt»1«btirty and Scale-up Protocols torPofynucl«ar Aromatic Hydrocarbon
Bior* mediation of MOP Soils
tions are the errors associated with the microbial analyses and these can bereduced in the future with use of more than three replicate samples formicrobial analysis (see § 2.4.2).
2.1.3 NAPHTHALENE BIOTRANSFORMATION AND MINERALIZATIONNaphthalene was tested in a similar manner as the phenanthrene tests
discussed above. Fate profiles as shown for phenanthrene are not shownfor naphthalene, however, generally similar behavior and analytical error wereexperienced in the naphthalene studies.
r
Equations A. 19 and A.20 were used to calculate interval pseudo-firstorder biotransformation and mineralization rate constants. These are pres-ented in Figure 12 for Soils A-E, as well as for five non-soil axentc (pure)cultures of phenanthrene degraders. These are Pseudomonas putida G7(Panel F). DFC50, (Panel G). and DFC 49 (Panel H), HK44 (Panel I). IGT 305(Panel J), and IGT 306 (Panel K). HK44 is a genetically engineered microor-ganism in which a bioluminescent gene cassette has been inserted into thelower pathway NAH7 naphthalene-degrading genes in such a way as tocouple the degradation of naphthalene with the evolution of light.15 The IGTstrains were isolated by investigators at the Institute of Gas Technology8 andwere kindly provided for these studies. A more detailed discussion of thenature of the axenic organisms is presented elsewhere.13 Note that axesare not scaled for direct visual comparison from panel to panel.
As in the case of phenanthrene, an overall first-order assumption ismade and Equations A. 19 and A.20 are used with the initial and final concen-trations and a first-order Rate Equation results with behavior shown in Figure13. Again departure from first-order behavior is highlighted in this figure.
• Shrivastava, V. J., Institute of Qas Technology, Chicago, II.
Testability and Scal*-up Protocol! tor FINAL REPORTPolynuclMf Aromatic Hydrocarbon 39 VOLUME IBior«m«diation of MOP Soils July 1991
Overall pseudo-first-order rate constants calculated from EquationsA. 19 and A.20 using initial and final concentrations are presented in Pane! Aof Figure 14. Pseudo-second-order rate constants based on Total Hetero-trophic Bacterial Plate Counts, /c|*and fc J". are presented in Pane! B.Pseudo-second-order rate constants based on nan gene probe-positivebacteria concentrations. fc|pand *?p. are presented in Panel C of Figure 14.
As in the case of phenanthrene, the naphthalene conversion processesare not generally first-order and are non-elementary and/or time-variant.Standard error of the first- and second-order rate constants range ±10-20%,and ±15-30%, respectively.
FINAL REPORT Tf«t«bilriy *nd Se«l«-up Protocols torVOLUME I 40 FofynuclMr Aromttie HydrocarbonJuly 1991 Biortmtdiition of MOP Soilt
K. KT SO*
2 4 I i 1 6 1 2 » <
ELAPStD TIME (MR)
INTERVAL PSEUDO-flRST-ORDCRBIOTRANSrORMATlON ————————RAT[ CONSTANT
INTERVAL PSEUDO-riRST-ORDERi I f O
TIME (HR)I* MINERALIZATION
•ATE CONSTANT
Figure 12. Interval-Based Pseudo-First-Order Rate Constants forRadiolabeled Naphthalene Biotransformation and Mineralization in Batch
MGP Sol! and Axenic Organism Tests
Tr«»l*bility and Seatt-up ProtocoU forPolynuclttr Aromatic HydrocarbonBiortmtdiation of MOP Soil*
41FINAL REPORT
VOLUME IJuly 1991
I ; 3 * S « ' « » i O " f l 2 4 k I 1 0 1 2 1 4
2 * * a iC 12
ELAPSED TIME (MR) ELAPSED TIME (HR)MEASURED OVERALL FIRST-ORDER
Figure 13. Actual and Predicted Radiolabeled Naphthalene Concentrationsin the Hexane Extract from Batch MGP Soil and Axenic Organism Tests
FINAL REPORTVOLUME tJuly 1991
42Trtittbility and Sc»tt-up Protocols for
Porynucl*v Aromatic HydrocarbonBior«m*diation of MGP Soils
a >o <i a>- x
I Zo <
trt O0. <J
a ft
3u.(J
OttO
1
o >a o
0:ui
0.5
0
-0.5
-1
-1.5
— ^
-2.5
-3
So<i A Soi C Soil E OFC5C HK44 ICT 306Sell • Sen D P put C7 DFC't ICT 305
U
PANEL B. OVERALL PSEUDO-SCCOKD-ORDER RATE CONSTANTHETEROTROPHlC BACTERIA
0
•O.OC2
C.004
0.006
0.006
• •L-l
-
•
-
! u
L-
PANEL C. OVCBALL PSCU&o-SCCOND-ORDCR RATE CONSTANTnan GENE PROK- POSITIVE lACTCRn
Soil A Sei C So-1 E DfCSC HK44 ICT 306S«il I S«ll 0 f- put C? DTC*9 ICT 305
Figure 14. Overall Pseudo-First- and Second-Order Radiolabeled Naphtha-lene Rate Constants from Batch MGP Soil and Axenic Organism Tests
2.1.4 ANTHRACENE BIOTRANSFORMATION AND MINERALIZATIONAnthracene was tested in Soils A. B, D and E with generally similar but
slower results relative to phenanthrene. Figure 15 presents the interval
Trt«t»bilrty and 6c*l*-up Protocol* lorPolynucl«»r Aromttie HydrocarbonBiortmtdiation of MGP Soils
43FINAL REPORT
VOLUME IJuly 1991
pseudo-first-order rate constants calculated for anthracene biotransformationand mineralization, while Figure 16 presents the overall pseudo-first- andsecond-order rate constants for anthracene biotransformation and mineral-ization.
As in the case of phenanthrene, the anthracene conversion processesare not generally first-order and are non-elementary and/or time-variant.Relative standard error of the first- and second-order rate constants range±10-20%, and ±15-30%, respectively.
r tcu. •
IOC IK
CURSED T1DC IK
C (-B)
INTERVAL PSCUDO-rtRST-OtDCB PKUDO-ri«T-ORDCR
RATE RATE CONSTANT
Figure 15. Interval-Based Pseudo-First-Order Rate Constants forRadiolabeled Anthracene Biotransformation and Mineralization in Batch MGP
Soil and Axenic Organism Tests
FINAL REPORTVOLUME IJuly 1991
Trtatabilrty »nd Scalt-up Protocols forPolynuelwr Aromatic Hydrocarbon
BiOftmtdiation of MOP Soils
81 -«•'ii -0.2SSbJ K
O -0.4
5- 0.05b.OI5 0o
-0.05
-0.1
? -0.15
PANEL A, PSCUDO-FIRST-ORDCR BATE CONSTANT.
SoTTA Soil 8 SoiTD Sod E
PANEL B. PSIUDO-SECONO-ORDER RATE CONSTANTSHCTEROTRQPHIC BACTERIA
OVE
RALL
PS
tUD
O-S
CC
ON
O-O
RD
CR O
V
V 0.5
0
-0.5
-1
-1.5
•
^ ™
PANEL C. PSEUDO-SECOND-ORDER RATE CONSTANTSnoh GENE PROBE-POSITIVE BAC'ERU
Soil A So.i B So>i D So<l E
CD BIOTRANSFORMATION CD MINERALIZATION
Figure 16. Overall Pseudo-First- and Second-Order RadiolabeledAnthracene Rate Constants from Batch MQP Soil and Axenic Organism Tests
Trtaiibilfty and Seal*>up Protocol! forPotynucUar Aromatic HydrocarbonBiortmediation of MQP Soil*
45FINAL REPORT
VOLUME IJury 1991
2.2 DISCUSSION OF BATCH IBKP RESULTS
2.2.1 BIOTRANSFORMATION VS. MINERALIZATION KINETICSThe overall pseudo-first- and second-order rate constants for PAH bio-
transformation and mineralization are tabularly presented in Table 4.
Table 4. Summary of Overall Batch Pseudo-First- and Second-Order RateConstants for PAH Biotransformation and Mineralization
Overall Pseudo-Biotransformation Rate Constants
TWIM
TMIM
*: kA ** •** *!' * -
* *" *!'
So.! ASa.il§0.1 CSMOSO.IE
f. 0W*tt
atOfCSCQFC48MKM
CTJ09IQT3M
• 1 SS•95S•0102•1202 12••79
•••7•429-n 12-724429
•0 1U• 1 94
•0002*• 1.14-1 IS
•04S1
•OU1
•COM•0273
1
-OMi•OM1
•29*10^ -TOUO"7
-i9iiO'10 -i SaiO111"• 1 9HO*7 -i taiO"7
-2 mo-' -ifcio-7
•SIHO4 -»tHO"'
•s'tno^ -staicr*•40HO'*• 1 1HO"7
•72H04
•43HO-*
Overall Pseudo-Mineralization
•1.7iiQ*S >1,3iiO^ -1 2>10''•1.2X10*7 -S.faiO^ -2 fclO^
•3.1iiO* -9 7«10*10*^OklO"* -B.7i10^ -94iiO'•3.s»io* -i i«i(r* -ifcno'7
-i.tatO^ -4tncr*
•atvio*1 -«5Mio'*•4 3»ifl-»-1 Illff7
-72«iO*-S3.10-*
Rate Constants
•7 JiiO->•4SJI10-7
•1.9HO-*•i taiO'7
*
S»: A
So.10
f putdt
o.an0.702001902790.9001.31
DFC90 10SOFC4* 0.907HKW V4«
OT30t 0.734QTJOS 1.07
Q.£200070
02190.074
01Noosa
OOOS 4taiQ->0007 1&10-' 32(10* 1 1ilO'c
l.»ilO-'0 9.7HO-14*OO'I 3-9ilO* l.tolO* I Hid'* 10097 90110** HilO*
2U10'7
Staler7
i mo*7
74*10-10
2.0110*•-fclO-W I.SJMQ-IO
73B10*vine*
7.fc10*1-ltlO*
' HM» e«icuUM«w tn puMi d.
FINAL REPORTVOLUME IJuly 1991
46Tr«tability and Se«l*-up Protocols (or
Potynuclwr Arom*tic HydrocarbonBiortmtdiation of MOP Soils
Relative standard errors for the first-order and second-order rate con-stants (except as noted) range ±10-20%, and ±15-30%, respectively.
The radiolabeled carbon being converted to carbon dioxide as apercent of the removal of the parent compound is presented in Table 5. Ahigh percentage indicates that mineralization is a dominant fate of the parentcompound over the duration of the test. A low percentage may indicate thata significant fraction of the radiolabeled carbon in the parent compound hasa fate other than mineralization. These may include conversion to polarmetabolites, immobilization into the soil or biomass, or volatilization losses inthe test. The latter possibility is unlikely since the test was conducted in asealed vial and the relative standard errors between replicates of all sampleswere generally low (±5-15%)
Table 5. PAH Carbon Mineralization Rate as a Percentage of theBiotransformation Rate
Soil Type
Soil ASoilBSoilCSoilDSoilE
P. putida G7OFC50DFC49HK44
IGT30SIOT306
Naphlhtltnt
1713152324151312131013
Phtmmhrtnt
135
<112IB1541
13
Anthrtctnt
113
616
2.2.2 AXENIC (PURE CULTURE) VS. MGP SOIL KINETICSThe comparative kinetic behavior of axenic PAH-degrading organisms
and indigenous PAH-degrading organisms is interesting. Table 4 indicatesthat overall naphthalene and phenanthrene pseudo-first-order biotransforma-tion and mineralization rate constants (k* or k1?) for axenic organisms are ofthe same order as the fastest of the indigenous MGP soil organisms.
Trttlibility «nd Sc«l»-up Protocol* tor FINAL REPORTPotynueltar Aromttic Hydrocarbon 47 VOLUME IBioftmt diation of MOP Soili Juty 1901
However except for MGP Soil C and naphthalene degradation in Soil E,pseudo-second-order rate constants based on gene probes for the MGPindigenous organisms have a greater magnitude than those for the axenicorganisms. This finding indicates that indigenous organisms in MGP soiltend to degrade PAHs with greater efficiency than axenic organisms inculture. This is additional evidence that pure-culture test kinetic resultsshould not be used to project field-scale activity.
2.2.3 RELATIVE PAH BIOTRANSFORMATION KINETICSTable 6 presents the rates of naphthalene and phenanthrene biotrans-
formation and mineralization relative to those of anthracene. Since the micro-bial concentrations are the same for each soil, these ratios should beindependent of rate constant order.
Table 6. Biotransformation and Mineralization Rates of Naphthalene,Phenanthrene Relative that of Anthracene in MGP Soils
Biotransformation
Soil Type
Soil ASoilBSoilDSoilE
Naphthalene:Anthracene
18213.95.9
Phenanthrene:Anthracene
1.65.63.73.3
Naphthalene:Phenanthrene
113.81.11.8
Mineralization
Soil Type
Soil ASoilBSoilDSoilE
Naphthalene:Anthracene
31100148.6
Phenanthrene:Anthracene
2.2107.33.8
Naphthalene:Phenanthrene
1410ZO2.3
FINAL REPORTVOLUME 1July 1991
Trtttabilrty and Scat* up Protocols lor46 PolynuelMT Aromatic Hydrocarbon
Biorvmtdiation of MOP Soili
The PAH biotransformation rates compared to the anthracene biotrans-formation rate ranges approximately 5-fold while the naphthalene:phenan-threne ratio ranges 10-fold for MGP soils. Mineralization ratios to anthracenerange 11-fold and 5-fold for naphthalene and phenanthrene, respectively and7-fold for naphthalene relative to phenanthrene.
Constant relative degradation rates between PAHs are not maintainedfrom soil to soil. Significant error can result if biotransformation rates of onePAH are calculated as a simple proportionality to others in the absence ofsupporting experiments. This approach is not recommended as a predictivemethod.
2.3 PREDICTIVE KINETIC RELATIONSHIPS FOR COMPARISONSBETWEEN BATCH TESTS
Biotransformation rate constants assuming overall first-order rate behaviorwere calculated from the initial concentration (hexane-extract) and the final con-centration in the experiment. As shown earlier in Figures 9.10,12,13 and 15,the first-order assumption does not model the time-series behavior well.However, it may be used to exactly model any initial and final concentrations ofthe experiment (for that matter any rate model will perfectly model two datapoints) with the implied caution that prediction of intermediate data is notreliable.
Pseudo-second-order rate constants were calculated by dividing thepseudo-first-order rate constant by a variable representing the biomasspresent. The pseudo-second-order biotransformation rate constant using totalheterotrophic bacterial plate counts as the measure of biomass is denoted byk2bh while the related constant based on nah gene probe-positive bacterialconcentrations is denoted k2bP. The biomass in each case is Bh and BP.respectively.
Trtatibilrty ind Sealf up Protocols lor • FINAL REPORTPotynucltar Aromatic Hydrocarbon 49 VOLUME IBioftmtdittion of MOP Soils July 1 Ml
Exploration for exponential and power relationships between both themeasured initial TOO and total PAHs of five MQP soils (Table 1) and thepseudo-first- and second-order rate constants was then performed.
Logarithmic transformations of the independent variables and linearregression with the dependent variable permitted investigation of exponentialor logarithmic relationships of the following form:
y (x ) -a !0 6 x
Similarly, linear regression of both independent and dependent variableswas used to test for power function relationships.
The regressions also provided an estimate of the quality of fit (correlationcoefficient, r2). Data on both naphthalene and phenanthrene were tested butnaphthalene results were consistently poorer than phenanthrene and are notpresented. Table 7 presents some of the more significant regression results.
FINAL REPORT TrMUbilrty and S»l*-up Protocols forVOLUME I 50 Pofynuclwr Aromatic HydrocarbonJuly 1991 Bior«m»diation of MQP Soil*
Table 7. Regressions for Predictive Kinetic RelationshipsNumber of
Soils
5
5
5
5
5
5
IndependentVariable
togTPAH
togTPAH
togTPAH
togTOC
togTOC
togTOC
DependentVariables
l * t l
|*S"l.Bh
|*!'|.BP
1*11
i*riBhI*S'I.BP
CorrelationCoefficient, r2
0.978
0.500
0.498
0.456
0.291
0.001
An excellent relationship for the prediction of ki as a function of logTPAH was developed with a correlation coefficient, r2, of 0.978 (Figure 17).TPAH has units of MQ/kg and k-jb has units of day1. The standard error fromthe use of this relationship is comparable to the experimental standard errorestimated for pseudo-ftrst-order rate constants (± 10-20%).
Ii 5 -o.«B E?!t,11 .
-1
Figure 17. Predictive Relationship for Phenanthrene OverallPseudo-First-Order Biotransformation Rate Constants in MGP Soils
Trtattbility and Scale-up Protocols lorPoiynuelcar Aromatic HydrocarbonBioramadialton of MOP Soilt
51FINAL REPORT
VOLUME IJuly 1 Wl
2.4 ANALYSIS OF ALLOWABLE ERROR FOR RELIABLE SCALE-UP
2.4.1 IMPACT OF EXPERIMENTAL ERRORS ON THE RATE CONSTANTAny experiment with the purpose of providing rate information for
scale-up includes experimental error in the measurements for concentrationsand time. For a batch-type test, the time is the sampling time for the concen-tration samples. In a continuous-type test, the time is the reactor residencetime (influenced by flow and volume changes). These errors propagate tothe value calculated for the rate constant for any Rate Equation assumed.The impact of the error is, however mitigated by the mathematical rateconstant calculation itself.
A discussion of this error propagation process is given in § A-2. Notethat this source of error is independent of inaccuracies in the selection of theRate Equation and that incorrect or unknown time-variant Rate Equationsmay contribute additional error to the calculated constants.
Figure 18 describes the impact of measurement error propagated to thecalculated rate constant for a first-order Rate Equation in an ideal continuousstirred tank reactor. In this analysis, Equation A.12 was solved for k-j and ±4, and 10% errors were added to concentration and residence time measure-ments. Relative standard error in the rate constant was calculated accordingto Equation A. 13 for the cases listed in Table A.1 (Appendix A) and theaverage (bold solid lines), minimum (short dashed lines), and maximum(long dashed lines) are shown for each error assumption as a function of theparent compound conversion (Equation A.4).
FINAL REPORT Trttttbility and ScaJ*up Protocols forVOLUME I 52 Porynueltar Aromatic HydrocarbonJuly 1991 8ior*m«diation of MGP Soil*
AVttACt ttunvl STANDARD CMC*. 'OK CASE
40 SO tO 70 ID *0 100
t PERCENT CONVERSION Of THE PARENT COMPOUND IN A CONTINUOUS TEST
Figure 18. Error Propagation to the Calculated First-Order Rate ConstantFrom Measurement Errors in Concentration and Residence Time
in Continuous Tests
For comparative purposes, the desired relative accuracy {±20%} in thekinetic rate constant is shown. As noted earlier, this was determined heuristi-cally by discussion with several environmental biotechnology professionals.
Figure 18 shows that with ±10% relative standard measurement error,the relative standard error in the calculated rate constant is less than ±20%only for systems with greater than 75% conversion of the parent compound.Relative standard errors for the lower measurement error case (±4%)tolerate lower parent compound conversions in order to meet the desiredaccuracy in the rate constant calculation. Obviously, when the measurementerrors are higher than ±10%, greater than 75% conversion is needed toachieve constants with the desired accuracy. Alt else being equal, tests withlower conversions result in greater error in the calculated rate constant.
A similar analysis may be made for zero and first-order rate constantscalculated from data in batch experiments using Equations A. 5-9 (Figure 19).
Tf« iiibilrty and Seal*-up Protocols forPolynucltir Aromatic HydrocarbonBioramtdiation of MOP Soils
53FINAL RE PORT
VOLUME IJuly 1991
In this analysis the relative standard errors are shown for first-order (solidline) and zero-order (dashed line) rate constants-the minimum andmaximum values are omitted. The time measurement refers to the timebetween samples in the batch test.
3s «cs »g K 20« y* ££ I 15
< *
« a_ zz <10
CASE
0 10 40 tO 10 100
PERCENT CONVERSION OF THC PARENT COMPOUND IN A BATCH TEST
Figure 19. Error Propagation to the Calculated Zero- and First-Order RateConstants From Measurement Errors in Concentration and Sampling Time in
Batch Tests
Zero- and first-order Rate Equations propagate error in a similarfashion. Here too, relative measurement errors of ±10% require a parentcompound conversion of nearly 60% to achieve the desired accuracy in therate constant.
In summary, continuous systems where the conversion of the parentcompound is low lead to rate constant calculations with inherently moreerror than systems where the conversion is high. Batch systems with relativestandard measurement error greater than or equal to ± 70% require a parentcompound conversion of at least 60% in order to produce a calculated rateconstant with average relative accuracy of ±20% or better. Batch-derived
FINAL REPORTVOLUME IJuly 1991
Trtliability and Scala-up Protocol* lorPofynucl**r Aromatic Hydrocarbon
Bioramadiation of MOP Soils
kinetics with either conversions less than 60% or relative standard measure-ment errors greater than ±10% will result in calculated rate constants that onthe average exceed the desired ±20% accuracy needed for reliablescale-up.
2.4.2 IMPACT OF CHEMICAL ANALYSIS ERRORS ON RATE CONSTANTERROR
The measurement errors for concentration depend greatly on thecompound measured, the method used, and the sample soil matrix itself. Inanalytical work on MGP soils by ReTeC2 (Table 8). coefficients of variation(CV. standard deviation divided by the mean) for the initial concentration ofTotal PAHs ranged from 0.24 to 1.08 for the soils discussed in this report.Naphthalene and phenanthrene CVs ranged from 0.41-1.34 and 0.78-1.48,respectively.
Table 8. Coefficient of Variation Error Statistic for Conventional ChemicalPAH Analysis in Initial MGP Soils
MGP Soil
Soil ASoilBSoilCSoilDSoilE
Number ofReplicates
4710117
NaphthaleneCV
0.410.670.510.651.34
PhenanthreneCV
1.121.341.080.781.48
Total PAHsCV
0.240.640.520.251.08
• CV » coefficient of variation « standard deviation/mean.
An inherent advantage of the radiolabeled method described in thiswork is that the CVs for radiolabeled analysis in the batch tests describedearlier are significantly lower than actual chemical analysis, even in the samesample matrix (Table 9). Of the same magnitude as measurement errors
Trtitibility and Sealt-up Protocol! for - FINAL REPORTPotynuclMf Aromatic Hydrocarbon 59 VOLUME IBiort mediation of MOP Soila Jury 1&91
assumed in Figures 18 and 19, the radiolabel approach appears to offerinherently greater accuracy in the resulting rate constant calculation thandoes conventional chemical analysis for PAHs in MGP soils.
Table 9. Coefficient of Variation Error Statistic for Radiolabeled PAH Analysisin Initial MGP Soils and Average for a Biotransformation Test, Three Repli-cates for Each Sample
Naphthalene
MOP Soil
Soil ASoilBSoilCSoilDSoilE
Initial
0.010.070.040.050.02
cv*Average
0.200.110.170.130.16
Phenanthrent
Initial
0.120.030.060.070.09
CV*
Average
0.200.150.040.160.10
Anthracene
Initial
0.050.04
0.160.10
CV
Average
0.090.12
0.100.08
* CV = coefficient of variation « standard deviation/mean.
With a normal distribution assumed for the rate constant (this will bediscussed further), the accuracy also depends upon the number of kinetictests performed. The relative error of the mean rate constant value can berelated to a function of the coefficient of variation and the number of kinetictests by manipulating the following equation:^
In this equation, the number of replicate tests is rc ;. z0.os'S a statistic
based on the selection of a 95% confidence limit, x is the mean of the repli-cate tests, nis the true population mean, and a2is the variance of the repli-cate samples. Rearranging this equation in terms of relative percent error,
FINAL REPORT Trtatability and Scale-up Protocol* lorVOLUME I 96 Polynueltar Aromatic HydrocarbonJuly 1991 8ior«m«diatJoft of MGP Soilt
€ . Hi x 100, leads to an expression relating the number of replicate tests
required to achieve a given error based on the coefficient of variationbetween replicate tests.
Figure 20 presents this relationship to indicate the importance inincreasing the number of replicate tests as the coefficient of variation ofbetween tests increases.
T O O
QC «at a
I «^ • '
w *
IZ
CV rtprtltits IhtCotffieitnt of Variation of the
100NUMBtR OF RtPLlCATt K I N E T I C TESTS
Figure 20. Relationship Between the Number of Replicate Kinetic Tests,the Coefficient of Variation Between Rate Constants Between Replicates and
the Relative Error of the Calculated Mean (95% Confidence Interval)
When the coefficient of variation between replicates tests is low. then theeffect of increasing the number of replicate tests on the accuracy of the cal-culated rate constant is small. However, when the coefficient of variation
Trtiiability and $c*l*>up Protocol* forPolynucl««r Aromatic HydrocarbonBiortmaditlion of MOP Soils
57FINAL REPORT
VOLUME IJuly 1*91
between replicate tests is high, more replicates are required to assure thesame low relative error in the calculated rate constant. If it can be assumedthat the coefficients of variation between calculated rate constants from repli-cate kinetic tests using conventional PAH chemical analysis will be greaterthan using radiolabeled PAH analysis (Tables 8 and 9), then usingradiolabeled analysis should a/so reduce the number of experimentsneeded for the same accuracy in the calculated rate constant
2.4.3 IMPACT OF RATE CONSTANT ERROR ON SCALE-UPtf a first-order Rate Equation is assumed, if the rate constant has been
determined with known precision (expressed as a coefficient of variation) andif no additional scale-up predictive equation is needed, the time for treatmentto a given level of parent compound conversion in a batch system can befound by solving Equation A.6 or A.8 for the treatment time, t. Normalized tothe conversion half-life:
1 / 2
...the ±95% confidence intervals and ± standard error interval in the resultingdimensionless time for treatment related to coefficients of variation of the rateconstant, k^ is presented in Figure 21 for the case where 90% of the parentcompound is biotransformed.
The impact of error in the kinetic rate constant on the time of treatmentand related economics is significant. As the coefficient of variation of kiincreases say to 0.4. every one-in-twenty trials (95% confidence interval) thetreatment time will be over 4 times longer than that predicted from the meanrate constant. If the coefficient of variation in k^ is 0.77, then every one-in-
FINAL REPORT Trtatafailtty and Sc«l»-up Protocols forVOLUME I 58 Por/nucltar Aromatic HydrocarbonJuly 1991 Bioramtdiation of MGP Soil*
three trials (standard error) the treatment time will be about 4-fold theexpected value. The expected value of t/tya for 90% of the parent compoundconversion is 3.32 (or log[t/tya] - 0.52).
tw -in§
wfc_tIh.
NteJ
3.32
1:20 Ft*rOty*MGC CWVlLO't(tH COwflOfCE LIK7I)
- 1J KWJtMAlCC CNVtLO«
MUM CM
0.2 0.1 t.2 1 i 1.1
or VARIATION IN OVERALL FIRST-ORDER•^TRANSFORMATION RATE CONSTANT
Figure 21. Impact of Error in the Rate Constant on the Probability ofExtended or Reduced Treatment Times Relative to the Prediction (Based on
90% Conversion of Parent Compound)
Errors resulting in behavior lower than the expected value on Figure 22result in faster treatments than that expected. Greater predictive reliabilityhere results in more accurate bids, potentially lower costs, and greater biore-mediation process competitiveness.
This analysis has assumed that the error in the rate constant is a normaldistribution. In reality, the distribution describing the rate constant is a one-sided bounded distribution. This is intuitive if one thinks about the biotrans-formation rate constant (a negative number) as it approaches zerobiotransformation rate. In this type of distribution, the standard deviationbetween the mean value and zero is less than the standard deviation
Testability and Se*l*-up Protocols forPotynucl«»f Aromatic HydrocarbonBioramtdialion of MOP Soils
FINAL REPORTVOLUME IJuly 1991
between the mean value and •«» For this reason, the confidence envelopeshown as positive time deviations on Figure 22 actually represent limitingworst cases. The 1:20 Performance Envelope case at a coefficient of varia-tion of 0.4 would be something less than 4-fold longer than the expectedvalue and would not have as severe an economic impact as that proposedhere. However, it might still likely be an economic catastrophe for the envi-ronmental biotechnology service company or his client.
2.5 DESORPTION/DIFFUSION AS A BIOTRANSFORMATION RATELIMITATION
The biodegradative kinetic testing reported above can be thought of as ameasure of the optimal intrinsic biotransformation kinetic potential (IBKP) inMQP soils. This is because the PAH compound is added to the test alreadydissolved in. a solvent and has maximum availability to the aqueous phase andthe microorganisms. Partitioning of the compound to microorganisms, to soil,to non-aqueous phase liquid (NAPL), to organic solid or plastic phases, andother solids present clearly occurs and competes with the biotransformationmechanism. As indicated in Figure 2 for toluene, biotransformations areexpected to increase PAH solubility and reduce solid partitioning. This type oftest minimizes the loss to sorption processes and maximizes removal by bio-transformation.
Strong evidence that PAH partitioning onto the non-aqueous phases and"slow" diffusion/desorption rates into the aqueous phase system can be rate-controlling and therefore may be very important in the field-scale MGP biore-mediation process was developed in a separate experiment. An MQP soil waspreloaded with 14C-naphthalene and was allowed to stand for a day. The soilwas then added to a continuously-stirred reactor with a filter insert to containthe solids as described in § 3.0 and was run with a continuous feed containingnon-radiolabeled (12C) naphthalene. Results from this batch test are pres-ented in Figure 22.
FINAL REPORT Trutabiliiy and Scal*-up Protocol* forVOLUME I 60 Polynucltar Aromatic HydrocarbonJuly 1991 Bioramadiation of MQP Soils
DESCRIPTION
CP toil wai prtloodtoC— r
Soil wot oddtC to CSTR with
f*td.
BlOTRANSrORMATlONKINETICS
20 40 tOELAPSED TIME (HR)
80
Sofb*d rad'AiobcitdftepMholtn*
Dit»oi»»d wniab«<«dnophthoUnt
Half-lift
1 1 hour*
<6 ice
Figure 22. Impact of Soil and Non-aqueous Phase Partitioning on NaphthaleneBiotransformation in an MGP Soil Slurry Reactor
The radiolabeled naphthalene preloaded on the MGP soil biotransformedwith a half-life of about 11 hours, while the unlabeled naphthalene steadily fedto the reactor in the aqueous feed biotransformed with a half-life of under.6seconds.
Several transport mechanisms can be proposed that offer explanationsfor the nature and physics of a slow desorption/diffusion process. However, itis very difficult (usually beyond our experimental capabilities) to conclusivelyidentify the responsible mechanism in a real MGP soil. In systems analysislingo, this is known as a system model which is not unique. The deterministicmodel describing the slow process is difficult to uniquely identify. For thesereasons, the protocols proposed in this work have the purpose of quantifyingthe slow process phenomenologically-in terms of time response instead of elu-cidating mechanisms deterministically.
2.6 CONCLUSIONS• A small vial-type batch laboratory test protocol can be used to performtreatability studies on PAH degradation in MGP soils. These kinetic results are
Tftalability and Scalt-up Protocol* forPofynueltar Aromatic HydrocarbonBioftmtdiaiion of MGP Soil*
FINAL REPORTVOLUME IJury 1991
idealized intrinsic biotransformation kinetic potentials (IBKP) that may becompared in a relative way, but should not be used directly to scale-up pro-cesses.
• The batch IBKP tests make use of radiolabeled parent PAH compoundsand their conversion products in a solvent-soil extract. Results appear to haveimproved accuracy and precision relative to similar tests based on chemicalanalysis of PAHs. This is likely a result of the minimization of "slow" sorption-/desorption processes caused by rapid and bioavailable addition of the testsubstrate and the ability to discriminate the added substrate from theindigenous substrate.
• The resulting improved batch IBKP precision leads to good correlationsbetween the specific PAH batch biotransformation rates, and the soil total PAHcomposition.
• When coupled with a protocol (still needing development and verification)to measure the slow desorption/diffusion rates of PAHs from MGP soils, theIBKP results could be used to select the most suitable MGP soils for bioremedi-ation. These treatability studies have as a purpose, selection of the best andrejection of the worst MGP soils for bioremediation. They will not alone besuitable for scale-up.
New scale-up protocols are needed to permit the use of standardizedtreatability test rate results to predict actual PAH bioremediation performance ina given field-scale test. These can only be developed when the laboratorymethod results are coupled to actual field-scale remediation results. Thisconcept is presented in this report but development and verification of thescale-up protocols must await future work.
• PAH biotransformation rate processes in batch PAH bioremediations arenon-elementary and/or time-variant. Prediction of batch field-scale time-
FlNAL REPORT Testability and Scala-up Protocol* forVOLUME I 62 Polynuclaar Aromatic HydrocarbonJuly 1991 BiOftmadiation of MOP Soila
concentration profiles will require a new level of sophistication in resolving thiscomplex behavior. This might take the form of a new artificial intelligencesystem that returns the "best" empirical Rate Equation from experimental time-concentration profile data.
• The need for improved accuracy in the rate constants, leading to betterfield-scale predictive reliability of design and performance mandates greateraccuracy and precision in the initial and final PAH concentrations and samplingtime measurements. Also, attention is required to the test design so that aminimum allowable PAH bio-conversion is met.
• The use of added radiolabeled PAH as a test compound minimizes theeffects of diffusion/desorption on the measured removal and tends to exper-imentally isolate biotic conversion processes from the mass transfer kineticeffects. In addition, this approach appears to diminish the coefficient ofvariation of the analytical results and offer the potential of greater statistical sig-nificance with a lower number of replicate samples or tests.
• Second-order rate constants are formulated by dividing the first-orderconstants by a microbiological parameter. The overall precision in sec-ond-order rate constants may be easily improved by increasing the number ofreplicate microbiological analyses to some number greater than three.
Trtattbility and Scalt-up Protocol* for FINAL REPORTPolynuclaar Aromatic Hydrocarbon 63 VOLUME IBiorcmtdiation of MQP Soil* July 1M1
This page is intentionally left blank.
3.0 A CONTINUOUS INTRINSIC BIOTRANSFORMATION POTENTIAL TEST
3.1 DYNAMIC MICROBIAL SYSTEMS ANALYSISEarlier sections of this report have dealt with the complexity of batch PAH
biotransformation processes in MGP soils. These have been found to be non-elementary and/or time-variant and therefore are difficult to reliably scale-upwith classical deterministic modelling approaches. This problem is not uniqueto PAH bioremediation and alternative approaches have been identified in thefield of Systems Engineering.17
Two broad divisions of the analysis of complex systems are the "directproblem" and the "inverse problem" approaches to systems analysis (Figure23). The direct problem approach describes the traditional approach taken byengineers and scientists when a process is to be modelled and its perform-ance or behavior predicted. Any system can be conceptualized as beingcomposed of three components: the process, input to the process, andoutput from the process. The direct problem deals with formulation of a modelfrom prior knowledge to describe behavior of the process, creation of a"known" input into that process, and subsequent calculation of the "unknown"output from that process. Simply put, two of three Variables" are "known" andthe third-the "unknown" output-is calculated.
This game may be played in other ways, too. Where the objective is toformulate the model or the input as the unknown, the "inverse problem"approach is employed. In Figure 23, three such inverse problems are noted:one seeking the best design, one seeking the nature of an input to give a givenoutput, and one seeking the model of the process. In the third case, theinverse problem approach is called system identification, and with knowninputs and outputs from a "black-box" system, the process model itself isextracted.
Traatabilfty and Scala-up Protocol! for FINAL REPORTPorynuclaar Aromatic Hydrocarbon AS VOLUME IBiortmtdiation of MOP Soil* Jury 1991
SYSTEMS ANALYSIS
DIRECT PROBLEM
PREDICTION
KNOWN S UNKNOWNInput OutputModtl
1DESIGN PROBLEM
KNOW ft S UNKNOWNinput Best DesignOutput
1INPUT IDENTIFICATION
KNOWNS UNKNOWN
Output InputModtl
INVERSE PROBLEM
MODEL OR INPUT
SYSTEM IDENTIFICATION
KNOWS S UNKNOWN
Input ModelOutput
Figure 23. The Direct and Inverse Problems of Systems Analysis
System identification methods are numerous and can be divided into twocategories: nonparametric and parametric model methods. In non-parametricmethodology, little or no prior knowledge on the system structure is neededand it truly is handled as a black box. The type of mathematical model gener-ated is a phenomenological model that maps the known input into the knownoutput. Clearly the resolution of this type of model depends on the ability tocreate or monitor the input and output signals without interference from othernoise from process disturbances.
FINAL REPORTVOLUME IJuly 1991
66Tr«atabilrty and Scale-up Protocol* tor
Polynucl* ar Aromatic HydrocarbonBiortmediation of MGP Soil*
The mathematical model developed in this approach has as its indepen-dent variable, time, and as its dependent variable(s), the input and output vari-ables. An example of such a model might be as follows:
A rt—r* >-:-*"> 9" 0dt2 Tdt
This is the deterministic equation of motion for a damped harmonic oscil-lator, such as a real pendulum. In this case, the parameters have physicalmeaning (y is the damping coefficient and GO is related to the pendulum'sweight and length, and 9 is the angle of the pendulum at an instant of time.
In another case, however, observation and mathematical analysis of anunknown system's response to a disturbance (we'll assume that we're observ-ing a biological damped harmonic oscillator system, but we don't know this inadvance) may give an equation of similar form:
dt2 dt
Here u is the measured input variable and y is the measured outputvariable. The system identification method tells us that the descriptive model isa linear sum of y and its first two time derivatives (each proportional to someparameter). If the system is simple and well-understood, perhaps the coeffi-cients can be interpreted with physical meaning. If the system is not wellunderstood, the physical meaning of the coefficients may remain very obscure.
One non-parametric system identification method used widely in engi-neering is the frequency-response method of Bode.18 •1S An input signal iscreated that is sinusoidal in form and is generated at a known frequency. Thisis done a number of times at various frequencies while the output is measured.For linear systems, the complex system phenomenological model can be
TVtalabiltty and Scal*-up Protocols for • FINAL REPORTPorynueliar Aromatic Hydrocarbon 6? VOLUME IBiortmediation of MOP Soils Jury 1W1
graphically taken from a plot of the output response as a function of input fre-quency. For non-linear systems (where the phenomenological model has atleast one term where variables of time or its derivatives are a product orquotient with another such time-based variable), the model extracted repre-sents a linearized approximation of one of the states of the "true system model"and for time-variant systems, the model represents a linearized snapshot of thesystem response at the time of the experiment.
The parametric approach also has numerous possible methodological alt-ernatives. One potent method when a significant level of knowledge about thesystem exists deals with writing the best possible deterministic model about thesystem's behavior, then knowing the system input and output, solving for oneor more parameters that aggregate variation and uncertainty. For example,Equations A.22-A.40 in Appendix A present an unsteady-state material balancemodel of PAH fates in a continuous-stirred tank reactor. All parameters can bemeasured or estimated except for the rate of PAH biotransformation. Byknowing the system input (feed concentration) and output (reactor effluentconcentration), the biotransformation rate can be estimated. This approach is "called parameter estimation and in this case is based on a deterministicmaterial balance on the PAH in the reactor.
Dynamic microbial systems analysis is a very powerful approach to betaken in the analysts of the complexities of PAH biotransformation in MGP soils.In general, it offers the potential for:20
A standardized approach for testing microbial systems to identify criticalor dominant mechanisms for further targeted research,
* A predemonstration testing and evaluation protocol to classify microbialprocesses for reliability and effectiveness under environmental disturbances,
FINAL REPORT Tr••lability and Scala-up Protocols forVOLUME I M Polynueltv Aromatic HydrocarbonJury 1991 Biortmtdiation of MOP Soil*
• A protocol leading to the identification of regimes of stability and robust-ness to disturbances that offer the potential for optimization by managing thenature of the disturbance,
• A protocol to assess system adaptability and its importance to structureand activity, and
• A dynamic experimental protocol that provides an experimental platformfor testing and evaluation of new monitoring approaches and process improve-ment schemes.
This section describes the application of two system identificationmethods for the study of dynamic responses of biodegradation in soil systems.The work is focused on developing techniques for monitoring and analyzingcontinuous flow, PAH degrading soil systems. Parameter estimation tech-niques provide the framework to extract biotransformation rate constants fromtime-series sampled data. This rate information represents idealized intripsicbiotransformation kinetic potential estimates for continuous systems. Inaddition, Bode non-parametric frequency-response techniques are used toestablish the stability and robustness of PAH-degrading systems. Thisapproach has the potential of identification of phenomenological dynamicresponse "Rate Equations" for us* in design and scale-up, however this wasunnecessary in this work since satisfactory first-order rate behavior was exper-imentally determined for most experiments.
3.2 DESCRIPTION OF EQUIPMENTThis section has a focus on the development of a continuous liquid flow,
complete mix reactor for studying PAH degradation in soil slurries. The systemand operating protocols were developed to maintain a homogeneous soilslurry at constant, controlled operating conditions (temperature, oxygen flow,agitation, feed flow, pH. etc.). On-line monitoring of offgas substrate concen-
TfMtabilrty and Scala-up Protocol* for FINAL REPORTPctynucltir Afomalic Hydrocarbon 68 VOLUME IBioiamtdiation of MOP Soilt July 1991
trations under continuous operating conditions was developed previously.®>10The entire system evolved and was optimized over the course of this three-yearproject.
The experimental apparatus was conceptually adapted by DiGrazia in hisPh.D. research for analysis of PAH degradation in soil slurries from the designused by Blackburn.® The apparatus was composed of four main components(shown schematically in Figure 24): the feed supply system, the bioreactorsystem, the offgas analysis system, and the computer control system.
The feed system consisted of refrigerated containment vessels for storageof feed materials and two, computer controlled positive displacement pumps.All feed lines and other surfaces exposed to feed materials were stainless steel,glass, or poly-TFE in order to minimize adsorption of PAH to reactor materials.
The bioreactor system used for dynamic analysis of naphthalene degra-dation in soil slurries consisted of a direct drive, continuous flow bioreactor with-a 0.75 L working volume. The reactor was equipped with temperature control,air flow control, and dissolved oxygen measurement modules. The reactorsystem was modified for experimentation with soil slurries (Figure 25) by usingan inner reaction vessel fabricated with sintered stainless steel. The innervessel was designed to contain reactor solids while allowing effluent to filterthrough the reactor walls.
FINAL REPORT Trt«t»bility and Scal«-up Protocols forVOLUME I 70 Porynuel«r Aromatic HydrocarbonJury 1991 6iortm«diation of MOP Soil*
Computer Control
Corner Cos
To Wast*
PhoiomuHIplier
. t • EffluintRtoctor
Figure 24. Schematic Diagram of CSTR Experimental System
Offgas from the reactor was routed to a computer controlled samplingvalve. Gas samples were analyzed using gas chromatography with packedcolumn and flame ionization detector. All components of the offgas samplingsystem were heated to 140°C to minimize adsorption of PAH to surfaces of theoffgas sampling system.
Trtatability »nd Scele-up Protocols lorPolynuelear Aromatic HydrocarbonB lore mediation of MOP Soil*
71FINAL REPORT
VOLUME IJuly 1991
GlassShell
CSTR withFilterInsert
LevelControl
Figure 25. Schematic Diagram of Modified Reactor with Metal Filter Insert
The pump flow rates, sample valve, gas chromatograph, and data acqui-sition were controlled using a personal computer. The computer program wascapable of controlling up to eight complete reactor systems simultaneouslywith the option of producing constant, square wave, sinusoidal, or sine sweeppatterns of flow control.
Several experiments were performed to establish operating protocols formicrobial systems analysis of the soil slurry reactor system. See ExperimentsT2.04. T2.05. T2.07. T2.09, T2.10. TZ11. T2.12, T2.14C, T2.15, T2.16, T2.17,T2.24, T2.25. T2.29. T2.30, T2.31, and T2.33 in Appendix B.
3.3 RESULTS OF PRELIMINARY STUDIESEarly continuous system trials were performed in order to begin system
development and define operating parameters for the latter experiments with
FINAL REPORTVOLUME IJury t991
TttatabilHy and Scala-up Protocols forPolynucltar Aromatic Hydrocarbon
Biortmvdiation ol MQP Soil*
MGP soil. In general, these early tests were done with PAH-spiked uncontami-nated soil and the later tests (§ 3.4) were done with MGP Soil A. Because ofdifferent and evolving system designs and operating procedures, comparisonof results beyond those reported here should only proceed after the specificsof each experiment are reviewed (see Appendix B).
3.3.1 MATERIALSTwo model soils were used in the reactor studies. An uncontaminated
Etowah silt loam surface soil from a site in Knoxville, Tennessee was used toproduce soil extract feed material for all of the experiments. This soil wasalso used in preliminary reactor and vial experiments. Table 10 describessome physical and chemical characteristics of the Tennessee soil. Thesecond model soil was a sample provided by GRI from a manufactured gasplant (MGP) site. The soil has been characterized by ReTeC (Soil A) and thedetails of the analysis have been presented previously.2
The feed material for the continuous experiments was prepared by hotwater extraction from the Tennessee soil. After extraction, solids wereremoved by continuous centrifugation followed by pressure filtration. Largequantities were prepared and stored at 4°C to eliminate batch-wise variationover the course of the project. The total organic carbon in the feed beforeaddition of naphthalene averaged 50 mg/L. Naphthalene was added to thefeed by saturating the solution at 50°C and allowing the naphthalene to crys-tallize out of solution while cooling to 4°C.
The inoculum used for the mixed culture soil slurry experiments wasprepared from a series of batch enrichments from MGP soil slurries supplem-ented with naphthalene, phenanthrene, and anthracene, and combined withan archived set of samples from PAH-exposed freshwater microcosms.21
The final enrichment cultures were spiked with additional PAH-degradingstrains, frozen and stored at -90°C to provide inocula for these and futurestudies.
Trtatability and Scal«-up Protocol! for • FINAL REPORTPolynuel«r Aromatic Hydrocarbon 73 VOLUME ISiortm»diation of MOP (oils «"UV
Table 10. Characteristics of Uncontaminated Tennessee Control Soil
pH (1:1 soil:water) 5.9Total Carbon, % 0.8Cation Exchange Capacity, 6.5meq/100g(pH7)
Mechanical Analysis, %Sand 24Coarse Silt 18Silt 33Clay 25
Fe, % as free Fe oxide 1.9Clay-sized Minerals* HKMQG
*K * kaolinite, H « hydroxyinterlayer vermiculite, M - mica, Q = quartz,G - gibbsite, in relative order of abundance.
3.3.2 OPERATING PARAMETERSVarious amounts of soil were added to the reactor to determine the
operating range for maintaining a homogeneous soil slurry in the reactor.The agitation rate and air flow rate were varied to determine the effect ofthese parameters on the physical soil distribution in the reactor. The air flowrate had little effect on the soil distribution. Agitator rates above 400 r.p.m.were sufficient to maintain a homogeneous soil slurry, as visually observed.
The oxygen mass transfer rates from the air feed to the reactor liquidwere determined over a range of air flow and agitation rates (ExperimentT2.07). Experiment T2.12 established the oxygen requirements for microbialcultures used in the soil slurry studies. The results indicated that for all agita-tion rates above 200 r.p.m. and air flow rates above 75 ml min~1, the reactorcould supply sufficient oxygen to maintain an aerobic system.
FINAL REPORT Trtatabilily and Scale-up Protocols forVOLUME I 74 Polynucltar Aromatic HydrocarbonJury 1991 Biorcmcdiation of MGP Soils
To ensure adequate oxygen supply and soil distribution, and tominimize stripping of PAH from the system, the operating air flow rate waschosen as 100 ml min'1 with an agitation rate of 500 r.p.m. This representsa specific power input of 12 hp/1000 gal (2400 W/m3).
3.3.3 PARTITION COEFFICIENTSParameter estimation of biotransformation rates in a defined system
requires complete characterization of parent compound fates, such that theonly unknown component in the mass balance is the term representingremoval due to degradative processes. Characterization of the abiotic fate ofthe parent compound is accomplished by operating the reactor system inthe same manner as when conducting biotic experiments, while suitably sup-pressing biological activity. In the continuous stirred tank reactor (CSTR) soilslurry system, biological activity was inhibited by using a 0.05% mercuricchloride in soil extract solution to "poison" the active population. In addition,the air feed stream was replaced with a pure nitrogen stream in order to*eliminate aerobic biotransformations. This procedure, coupling two indepen-dent methods to inhibit biological activity, maximized the chances that bio-transformation was eliminated in the soil slurry reactor.
As presented in Equation A.22, the major fates of naphthalene in the soilslurry reactor are stripping in the off gas, sorption to solids, and removal inthe effluent. Two types of frequency response experiments were used tocharacterize the abiotic fates of naphthalene in the reactor system. First,sinusoidal perturbations were induced in the feed naphthalene concentrationfed to an abiotic reactor with no soil. This data was used to characterize thestripping process. Second, the same protocol was used to study stripping inthe soil slurry system. Previous data (Experiment T2.14C) suggested that soilin the reactor system did not affect the stripping process. Comparison of thedata from the stripping and soil slurry experiments were compared to testthis hypothesis.
Trtatabiltty and Scal*-up Protocols (or FINAL REPORTPolynuelear Aromatic Hydrocarbon 75 VOLUME IBioramadiation of MOP Soils July 1961
Stripping of naphthalene from agitated biological treatment systems canbe modeled as an equilibrium process by Equation A.33,*i 22 > 23 where thegas concentration is proportional to the liquid concentration. Sorption ofnaphthalene to soil can also be modeled as an equilibrium process byEquation A.34. These models can be substituted into Equation A.31 whichcan be solved analytically to give Equation A.36.
Equation A.40 results when biotic rate processes are set to zero andrelates to the abiotic (sterile or biologically inactive) case. Laplace transfor-mation of Equation A.40 results in a transfer function for the abiotic system(Equation A.45). The amplitude ratio and phase angle of the outputwaveform with respect to the input waveform for a transfer function of thisform can be expressed as Equations A.46 and A.47, respectively.24
The values of the unknown stripping and sorption parameters weretherefore determined from abiotic dynamic experiments. Each individualdata set (one for each frequency tested) was fit to Equation A.40 using non- 'linear regression techniques.
I I I-^!- Eqn.AAO.
All parameters are known save KD and Kn. These results with ther 9
Henry's Law Constant calculated from Kg are reported in Table 11.
If the amplitude ratio and phase angle of the response are plottedagainst the input frequency of the feed perturbations (Figure 26), it isobserved that the data clearly fit two experimental curves. The upper curvein Panel A of Figure 26 represents the amplitude ratios for the abiotic exper-imental runs with no soil. The upper curve in Panel B represents the phaseangle for the abiotic-no soil runs. Fining the amplitude ratio data to EquationA.46 (with A.43 and A.44) and fining the phase angle data to Equation A.47
FINAL REPORT Trtatabilrty and Scale-up Protocol! lorVOLUME I 76 Porynuel«ar Aromatic HydrocarbonJuly 1901 Biorcmtdiation of MOP Soils
(also with A.43 and A.44) results in a sorption parameter of zero and strip-ping parameters of 0.0079 and 0.0110 for the 10°C and 20eC experiments,respectively.
Eqn.AAb
-H 00*-tan" -— Eqn.AA?\ a J
P" ~7——^7~^ Eqn.AAlV\ 1 * —V p.
Eqn.AAS
The lower curves represent the experimental runs with both Tennesseeand MGP soils. Similar mathematical analysis as described above, but usinga non-linear, two parameter curve fitting technique, results in a strippingparameter value of 0.0079 and a sorption parameter value of 12.1 for the10°C experiments, and a stripping parameter value of 0.0110 and a sorptionparameter value of 11.7 for the 20eC experiments. This analytical approach isan alternative to direct use of a material balance (Equation A.40) and permitsextraction of abiotic parameters from a series of perturbation/responseexperiments. These alternate parameter values correspond favorably withthe results presented in Table 11.
The literature values also compare favorably with the experimentalvalues for the Tennessee soil. However, the similarly low sorption parametervalues for the MGP soil were not expected since the fraction of organic
and Scal«-up Protocols for FINAL REPORTPolynuelMf Afomttic Hydrocarbon 77 VOLUME IBior*m«diation of MOP Soil* July 1991
carbon in the MGP soil is much higher than that of the Tennessee soil.Equation A.34 predicts that the sorption parameter should increase with anincrease in the soils organic carbon content. This increase in organic carbonmay be related to the non-aqueous phase tar component. In this case, it islikely that an increase in the sorption parameter would only be observed overlong time frames since sorption into this phase may be rate-limited by therelatively long time required for PAHs to diffuse through the tar itself. Theimpact of the increased organic carbon of the MQP soil would therefore notbe observed in these experiments which focus on the faster rate processes.
Table 11. Determination of Stripping and Sorption Abiotic Rate Constants inPerturbed CSTR Experiments (Temperatures: 10 and 206C)
Type of Soil
No Soil Added
UncontaminatedTennessee Soil
MGP Soil
PerturbationCycle Period
(hr)
461224
4
8122464
481224
Henry's Law Constantat Temperature
(unrtless x10'3)*
10°C 20°C
6.1 197.47.27.7 10
7.9 11
8.16.09.6 8.16.5
7.2 9.58.49.17.7 9.9
Sorption Constant. Ks
at Temperature(mug)
10°C
11.9
10.317.411.915.1
11.615.516.817.8
20*C
10.3
13.8
12.5
14.1
* Unrtless Henry's Law Constant is, for example (mg/L in the gas) / (mg/L in the liquid). It isnot the same as a unities* mole fraction ratio.
FINAL REPORTVOLUME IJuty 1991
78Trtatabilrty and Scale-up Protocols for
Polynucltar Aromatic HydrocarbonBioramtdiation of MOP Soil*
tr
£<
O<UJ1/1xa.
0.1
0.01
(64) PANEL A
(24)
(12)
• UNCONTAUINATED
• MCP SOIL
• NO SOU ADDED
0.01 0.1
y? °uiUJce
S -30
-60
-90
PANEL 8
(6*)
10
0.01 0.1 1 10PERTURBATION FREQUENCY (RADIANS/HR)
(NUMBERS IN PARENTHESES INDICATE CYCLE PERIOD IN MR)
Figure 26. Bode Diagram Comparing CSTR Naphthalene Reactor LiquidConcentration Response to Induced Sinusoidal Perturbations in Naphthalene
Feed Concentration (Abiotic Experiments, Manual Recycle, 10°C)
3.3.4 KINETIC STABILITY AND ROBUSTNESSThe initial CSTR protocol called for operation of the reactor system at
20°C, an air flow rate of 100 mL min*1, and a soil extract feed with 14 mg f1
maximum naphthalene concentration. The initial CSTR runs at these condi-tions established that the naphthalene degradation rates were consistentlyhigh enough to reduce the reactor liquid concentration below the detectionlimit of 0.5 j/g L"1 (Experiments T2.15, T2.16). Frequency response systemsidentification techniques could not be used to characterize dynamic systembehavior since the output signal was below detection. Also, parameter esti-mation or resolution of Rate Equations could not proceed without measured
and Se*l«-up Protocols forPorynuclaar Aromatic HydrocarbonBior»m*diation of MOP Soils
79FINAL REPORT
VOLUME IJury 1991
concentration data. Since the naphthalene analysis used was already verysensitive (-0.5 M9 L'1) and further improvements in chemical analysis sensi-tivity would be at the expense of rapid time-series sampling (-10 sam-ples/hr), bioactivity reduction was investigated as a strategy for increasingthe reactor liquid concentration permitting microbial systems analysis.
The first attempt to reduce the rates of naphthalene biotransformationwas to reduce the reactor temperature. The temperature in two soil slurryreactors operating for several weeks at 20°C was reduced to 10°C to reducebiotransformation activity (Experiment T2.18). The reactors were operatedfor a period of one week before the feed naphthalene concentration wasvaried sinusoidally (cycle period 1 hour). The naphthalene level remainedabove the limit of detection throughout this period, but subsequently fellbelow the detection limit after the cycles in naphthalene feed concentrationwere stopped. This is an experimental indication of time-variance in the bio-transformation process and may also be related to a change of microbialsystem state (state bifurcation).
The reactor temperature was further reduced to 4°C. The reactor liquidnaphthalene concentration increased to a level near 100pg L'1 andremained there for a period of one week, when the experiment was ended.Although this represented a significant increase in the reactor naphthaleneconcentration into a range where the analytical scheme would be effective,with a feed concentration of -10 mg L"1 and effluent concentrations of -100jug L"1. over 99% of the naphthalene in the liquid feed stream was biotrans-formed (Experiment T2.18). That these high biotransformation rates contin-ued even at low temperatures was surprising but later confirmed.
In order to verify that the loss of naphthalene even at low temperatureswas, in fact, due to aerobic biotransformation and not some other obscurebiotic or abiotic fate mechanism, the experimental.system was modified such
FINAL REPORT Traatability and Scale-up Protocol! (orVOLUME I 90 Polynuclaar Aromatic HydrocarbonJuly 1991 Bioramadiation of MOP Soils
that the air feed stream could be combined with a pure nitrogen feed stream.The gas flow rate to the reactor was kept constant while the fraction ofoxygen in the feed was gradually reduced.
The biotransformation rate was affected only when oxygen was almosttotally eliminated from the feed gas, thus forcing the dissolved oxygen con-centration in the reactor below 1 mg L'1 (Experiment T2.20). When a purenitrogen gas feed was used, biodegradative activity ceased, as indicated bya rise in the reactor naphthalene concentration to a steady-state level equalto the calculated level corresponding to abiotic stripping as the only fatemechanism for naphthalene removal. While the approach of limiting theterminal electron acceptor in the experiments was experimentally viable andreductions of activity could be so controlled, it was unclear how to comparethe results with the batch IBKP results and to fteld applications.
A different approach was then considered to reduce biotransformationactivity and increase reactor liquid concentration. A hypothesis was possiblethat one cause for the high naphthalene biotransformation rates was thatnaphthalene was essentially the sole carbon source. By supplying alterna-tive carbon sources to the system, naphthalene would not be a requirementfor cell survival and naphthalene biotransformation rates might be reducedby supplying additional carbon sources to the system.
The first experiment involved using MGP soil in place of Tennessee soil.The MGP soil has multiple PAH contaminants which would serve as alterna-tive carbon sources to the system. This change of soil type had no measur-able effect on system performance (Experiment T2.22) as indicated bynaphthalene concentrations remaining below analytical detection limits. Thesecond experiment involved adding phenanthrene and anthracene to thefeed stream to provide additional organic carbon. This also had no effect onsystem behavior (Experiment T2.26). Finally, yeast extract, peptone, glucose
Tnitobllfly and Scale-up Protocols tor FINAL REPORTPolynuclaar Aromatic Hydrocarbon 81 VOLUME IBiortmtdiation of MGP Soils July 1991
(YEPG) media was used in place of buffered soil extract as the feed material.Again, this media, rich in organic carbon, had no measurable effect on naph-thalene biotransformation rates (Experiment T2.28).
While this series of experiments was unable to identify operating condi-tions where the liquid concentration signals were detectable, indirect resultsshowed that naphthalene biotransformation was surprisingly persistent androbust in the face of changes in concentrations of naphthalene and othercarbon compounds, temperature and oxygen levels.
3.4 NAPHTHALENE BIOTRANSFORMATION IN THE CSTR TEST SYSTEMSeveral choices were possible in the design and operation of the test
reactor. Given that a continuous mode is desirable, the feed can contain bothliquid and soil, or liquid alone, tn the second case, one must decide whether touse: a flow-through strategy where the soil is permitted to leave the reactorwith the liquid effluent or an alternative strategy where the soil is contained orseparated from the liquid effluent and recycled back to the reactor.
Continuous feeding of dry and easily-flowing solids to a reactor has pre-cedent. However, MGP soils are neither dry nor well-behaved. It was believedthat semibatch manual additions in a small reactor could not be reproduciblyaccomplished over the extended operation planned. Further, the additionmight cause an important dynamic disturbance interfering with reactorbehavior. Deciding that this approach would itself require a research projectbeyond the time available for this work, continuous MGP soil addition wasabandoned.
Since a focus of the study was to maintain a long-term system operationto study dynamic response, the alternative of permitting the soil to wash out ofthe reactor was impractical. The half-life concentration of the soil in the reactor
FINAL REPORT Traatabilfty and Sc*l**up Protocol* forVOLUME t 62 PolynuclMr Aromatic HydrocarbonJuly 1991 Bioramtdiation of MOP Soilt
would be the same as the residence time-no more than a day or so. Also, adecline in the soil concentration might strongly impact the state of the systemin ways difficult to interpret.
Capturing or containing the soil was the best operational and simulativechoice. A simplistic approach, similar to that used in lab activated sludgestudies, was first selected. This made use of an operational strategy of dailymanual separation of the soil from the collected effluent by centrifugation, thenreturning the solids back to the system. Named "Manual Soil Recycle" in thisreport, many of test results were collected in this mode.
Because of evidence that the manual soil recycle option itself introducedlarge disturbances into the analysis and because the operation was labor-intensive, an improved alternative using a filter reactor insert was developed,tested and used in the final frequency-response experimental campaign.
3.4.1 RESULTS FROM MANUAL SOIL RECYCLE MODEThe first naphthalene frequency response experiment is outlined in
Experiment T2.15 in Appendix B. The reactor was operated for a period ofone week under steady feed conditions in order to acclimate the degradingculture to the system. After the first week, the feed naphthalene concentra-tion was varied sinusoidally (cycle period 24 hours) for an additional week.The reactor liquid response to the naphthalene feed perturbations is shownin Figure 27.
The reactor liquid naphthalene concentration varied periodicallybetween 0 and 14 /jg L~1, three orders-of-magnitude below the feed concen-tration which ranged from 0 to 14 mg L"1. This experiment demonstratedthat frequency-response methodology could be applied to studybiotransformation in soil slurry reactors, however, problems were identified.The increased naphthalene concentration in the reactor liquid coincided withthe daily maintenance of the reactor, as well as with the increase in feed con-
Trtatabiiity and Scala-up Protocol* for FINAL REPORTPolynuclaar Aromatic Hydrocarbon 63 VOLUME tBiortmediation of MOP Soils Jury 1991
centration. During reactor maintenance the solids were collected from theeffluent and returned to the reactor, tt was difficult to determine whether theapparent sinusoidal response of the reactor was due to system response tothe feed perturbations or to daily maintenance disturbances.
After the 24-hour cycle, the reactor feed naphthalene concentration waskept constant, the reactor naphthalene concentration again fell below thedetection limit.
Al! subsequent experiments at 20°C had similar resurts-the reactorliquid naphthalene concentration remained below the detection limit of 0.5 pgL~1. A series of steps were taken to inhibit biodegradation of naphthalene inthe soil slurry system, as outlined in § 3.3.4. After a reduction in reactortemperature to 10°C, the reactor liquid naphthalene concentration increasedabove the detection limit, indicating that the naphthalene biotransformationrate in the system had also been reduced.
After a one-week acclimation period, a one-hour feed cycle was initi-ated. This one-hour cycle is shown as the "baseline" in both panels of Figure28. Two reactors were evaluated. The first reactor (Panel A) was operated insuch a manner that the feed liquid naphthalene concentration was variedsinusoidaily, which also resulted in a variation in total organic carbon (TOG)to the system. The second reactor (Panel B) had a constant naphthaleneconcentration in the feed while only the non-naphthalene organic carbon wasvaried. The response of the two reactors to the feed perturbations is shownin Figure 28.
Three key observations can be made: 1) the reactor naphthalene con-centration in both reactors varied periodically with the same frequency as thefeed perturbation, 2) a significant increase in reactor naphthaleneconcentration was observed following reactor maintenance each 24 hours,and 3) the amplitude of the reactor liquid naphthalene concentration in Panel
FINAL REPORT TrMtability and Scala-up Protocols forVOLUME I 64 PolynucUar Aromatic HydrocarbonJuly 1991 Bioramtdiation of MOP Soili
A is greater than that of Pane! B. This indicates that the periodic response ofthe reactor in the previous experiment (24-hour cycle, 20°C) was due to thecombined effects of daily reactor maintenance and the feed perturbations.
-14
-12-10- 8- 6- 4
- 2r 0
r>
I*I5 5z £
X
c5O
14 -
12 -
10-
6 - !.
A 'f_< (JI Z
4 -
2-n -
IV1 H.' 1'
> -•v" \ • V "l f
»*
24 46 72ELAPSED TIME (HR)
96 120
Figure 27. CSTR Response to Sinusoidal Perturbation in Feed NaphthaleneConcentration (24-Hour Cycle Period, Uncontaminated Tennessee Soil,
Manual Recycle, Experiment T2.15}
After reducing the temperature in the reactors to 4°C. the reactor liquidnaphthalene concentration rose to a level of 100 g L*1 and remained therefor a period of one week. Slurry samples were taken from Reactor 1 at timepoints at 20°C, 10°C. and 4°C and tested in offline batch assays (similar tothe batch IBKP proposed in § 2.0) to verify the biotransformation activityobserved in the CSTR reactors (Experiment T2.21).
Testability and Sc»lt-up Protocols forPolynucltar Aromatic HydrocarbonBiortmtdiation of MOP Soils
esFINAL REPORT
VOLUME IJuly 1991
t o.oe
0.06-
X a 0.04 -
G.02-
*, tEACTOB 1
20 40 (0 10TIME (HO)
100
S 0.08El 0. HEACTCff 2
O C 6 -
T >.OC4
x E
2 *i> i«I cc: - Jratf w>
2C •0 100
ELAPSE:) TIME (MR)
Figure 28. CSTR Response to Sinusoidal Perturbation in Feed NaphthaleneConcentration, Panel A, and Non-Naphthalene Organic Carbon, Panel B(1-Hour Cycle Period, Uncontaminated Tennessee Soil, Manual Recycle,
Experiment T2.18)
Biotransformation rate constants in both the CSTR reactor and thebatch assays were calculated based on the assumption that the biotransfor-mation was first-order with respect to reactor liquid naphthalene concentra-tion. Mineralization rates were based upon 14CC>2 accumulation whereasbiotransformation rates were calculated from 14C-naphthalenedisappearance data. The results are presented in Table 12. It is clear thatthe naphthalene transformation rates in the batch assays were
FINAL REPORTVOLUME IJuly 1991
Tr»Bt»bilrty and Sc«l*-up Protocol* forPolynuelcar Aromatic Hydrocarbon
Biortmediation of MGP Soils
orders-of-magnitude lower than those observed in the CSTR reactor. Alsonote that biotransformation and mineralization activity observed at 4°C wasverified in the batch assay, although the rate was much slower.
Table 12. Pseudo-First- and Second-Order Rate Constants from CSTRReactor Studies and Batch Naphthalene Biotransformation and Mineraliza-tion Assays
CSTRTemp
rcj
20
20
10
10
4
,
Oene Probe-Posrtive
Bacteria*(xl06du
mL'1)
10
5
5
2
5
Batch A**ay Overall Paeudo-FlnM-Order
Rail* Constant(day1)
„,*nd6 3.6
-36 23
nd nd
-1.8 1.1
•0.7 0.2
Batch Aatay Overall Paeudo-$*cond-OrderRate Constant
(xlO-7gefu-1 day1)
h2b kg"1
nd 3.6
.7,7 46
nd nd
-91 5.3
•1.5 0.5
CSTR IntervalP*eud>
Fir«t-OrderBiotrans,
Rale Constant(day1)
^< -15.600
< -15.600
<-9,120
.1.200
•96
in a separate experiment, a 30 g sample of soil was loaded with 14C-naphthalene and added to the reactor (Experiment T2.23). The soil slurrywas sampled periodically for 14C-naphthalene and offgas sampling wasused to monitor 14C02 production. The results are presented in Figure 22.The results indicate that the biotransformation (and mineralization) of the14C-naphthalene occurred at a rate three orders-of-magnitude lower than
Trtatability and Scale-up Protocols forPolynuelear Aromatic HydrocarbonBioremtdiation of MOP Soils
87FINAL REPORT
VOLUME IJury 1991
the naphthalene in the liquid feed stream. This is strong evidence supportingthe hypothesis that desorption/diffusion of the naphthalene from the soil is aslow and rate-limiting step in the overall biodegradation process.
10'
10'.
20 40 60 8C
C.APSEC TIM: (CATS)
PANEL B, REACTOR 220 C , 10 C
10
," ngr" 7.
' M\ • 'I. '•'
DETECTION LIMIT
20 80 ICO
Figure 29. Total Heterotrophic Bacteria and nah Gene Probe-PositiveBacteria Concentrations in CSTR Tennessee Soil Experiments (Panel A.Naphthalene Perturbation; Panel B, Non-Naphthalene Organic Carbon
Perturbation)
In a number of sequential experiments on the same soil sample (Experi-ments T2.16, T2.18. T2.19, T2.20, T2.21, T2.23) biological samples weretaken for analysis of total viable and naphthalene degrading cells. Both
FINAL REPORTVOLUME IJuly 1991
UTrtttabiiity and Scab-up Protocols for
Polynucl*»f Aromatic HydrocarbonBioramtdiation of MOP Soili
one-hour perturbation periods and unperturbed intervals were included inthese experiments. The results are presented in Figure 29. No significantdifferences in microbiat population levels were seen between temperatureranges, as indicated by analysis of variance (95% confidence interval). Thechanges of microbial concentrations within a temperature treatment may besignificant and result from environmental disturbances.
3.4.2 REACTOR FILTER INSERTIn light of the fact that the washout and subsequent recycling of the soil
from the reactor may have affected the performance of the system andmanual soil recycle was highly labor intensive, a new reactor was designedto contain the soil in the reactor while allowing liquid effluent to flow throughthe reactor, This was accomplished by fabricating a reactor filter insert fromsintered stainless steel. A soil slurry was contained in the reactor while liquideffluent flowed through the pores (0.4 to 4 pm) of the reactor. The liquid wasthen collected in an outer sleeve in order to control the reactor slurry volume(Figure 25). Experiments were designed and implemented to verify "that thisdesign was appropriate for the study of naphthalene biotransformation in thesoil slurry. Once the apparatus was tested, the frequency response protocolwas implemented in this newly designed system.
There is an obvious difference between the two sets of continuousexperiments discussed-in the early "manual-recycle" experiments, the soilwas allowed to flow out of the system and was subsequently recycled; in thelater experiments with the reactor insert the soil did not leave the reactor.This may have resulted in differences in biomass activity which may compli-cate direct comparison of the data from the two types of reactor experi-ments.
The manual recycle system may result in a different microbial commu-nity within the reactor, which may effect cellular activity. The centrifugationprocess might not have been efficient enough to guarantee the recycle of all
Traatability and Scal**up Protocol* lor FINAL REPORTPolynuclMf Aromatic Hydrocarbon 69 VOLUME IBiortmtdiation of MOP Soils July 1991
bacteria with preferential loss of the unflocculated bacteria. The mean cellresidence time and possibly the population distributions of the two systemswere likely different. It is unknown at this time what effect these events hadon the operating state of the system, however, ft is possible that the microbialsystem states of the two reactor systems were different. Biotic PAH conver-sion data should be compared between these two systems only with caution.
3.4.2.1 EARLY SYSTEM OPERABILITY STUDIESA sample of the sintered metal was obtained from the manufacturer in
the form of a cylinder for testing for our application. A mass transfer coeffi-cient was experimentally determined (Experiment T2.24 and EquationsA.48-56) which verified that the hydrostatic pressure in a reactor fabricatedfrom the same metal was sufficient for operation at the desired flow rates.Visual inspection of the filtrate indicated that the sintered metal would retainthe soil while allowing liquid to pass through the wall.
A reactor filter insert was then fabricated and tested with an activebiomass/soil slurry to test for blockage of the pores with biomass. Noblockage occurred, however, microscopic observation of the effluentestablished that some biomass. not soil particles, was present. It was con-cluded that the reactor could be operated for an extended period using theestablished protocol without blocking with soil or biomass. However thecapture efficiency of microorganisms in the reactor may have been lessthan complete.
3.4.2.2 FREQUENCY-RESPONSE ANALYSISA series of frequency response experiments were performed over a
range of feed perturbation cycle periods from 1 to 24 hours (ExperimentT2.34). Each experiment was initiated by inoculating the reactor with MGPsoil and the standard inoculum, and allowing the system to acclimateunder continuous flow conditions, constant naphthalene feed concentra-
FINAL REPORT Tfcatabilfty and Scal»-up Protocolt forVOLUME i 90 Porynuetoar Aromatic HydrocarbonJuly 1991 Biortmtdiation of MOP Soils
tion, for a period of four days. Following the acclimation period, a sinusoi-dal variation between 0 and 14 mg/L in naphthalene feed concentrationwas initiated.
The reactor offgas samples were processed periodically throughoutthe cycle to determine reactor naphthalene concentrations. During three ofthe experiments, 14C-naphthalene was injected into the reactor to verifythe degradative activity observed by sampling for 14C02 in the offgas, andfor14C-metabolites and ^C-naphthalene in the reactor. Reactor slurrysamples were taken daily for biomass analysis of both total viable cells andnaphthalene-degrading genotype.
PANCL. A, i HOUR PERTURBATION
12 16 20 24 28 32
^ i PANEL B, 12 HOUR PERTURBATIONso -
20 40 40 BO 100 120 140
CLAPSED TIME (HR)
Figure 30. CSTR Response to Sinusoidal Perturbation in FeedNaphthalene Concentration (Panel A, 1-Hour Cycle Period; Panel B,
12-Hour Cycle Period; MGP Soil, Filter Insert, Experiment T2.34)
Trtatability and Scala-up Protocols tor FINAL REPORTPorynuclaar Aromatic Hydrocarbon 91 VOLUME I6ior«m«diation of MOP Soils July 1991
An example of the typical reactor response, for a short (1 hour, PanelA) and long (12 hours, Panel B) cycle, is shown in Figure 30. The reactorconcentrations were typically three orders of magnitude lower than thefeed concentrations, indicating that the conversion of naphthalene wasconsistently greater than 99%. The rate of biodegradation at each samplepoint was calculated using Equation A.39 and ploned versus reactor liquidconcentration in Figure 31.
700 i——————————————————————! PANEL A, i HOUR PERTURBATION600 r _ _ -AXC500 ' ' " "•• ^ ' ' •" *
* 400? 300 r . ^ •>* • - :~ ' .- *• "* ^ •
K 100 -
I °r< -100 -
20 40 60OEO
Zft-- 700 ———————————————————————————————————————S • PANE. B. 24 HOUR PERTURBATlOtf-r.'tt 600 (• . %V" Z 500 -•-1 !^ 400 -| 300 -^ 200 -Z 100 '-
o ————- J O O :—————————————————————————————————————•
0 20 40 60
NAPHTHALENE REACTOR LIQUID CONCENTRATION (ug/L)
Figure 31. Naphthalene Biotransformation Rate vs. Naphthalene ReactorLiquid Concentrations (Panel A, 1-Hour Cycle Period; Panel B, 12-Hour
Cycle Period; MGP Soil, Filter Insert, Experiment T2.34)
While there is considerable scatter in the data, a linear relationshipexists between naphthalene reactor liquid concentration and the naphtha-lene biotransformation rate. Data for the 1-, 4-, and 24-hour cycles aresimilar to Figure 31, Panel A, all can be fit with an elementary first-order
FINAL REPORT TrMttbilfty and Sc«l*-up Protocols forVOLUME I 92 Potynuclt»r Aromttic HydrocarbonJuly 1991 Biortm«di»tion of MOP Soils
kinetic model (with a zero y-intercept). Data for the 2-, 8-, and 12-hourcycles are similar to Figure 31, Panel B. which can also be fit with afirst-order model. However, there is an apparent minimum liquid naphtha-lene concentration below which the rate of naphthalene biotransformationis zero (i.e., a model of form: R*k(C-Cmin))- The rate constants andminimum concentrations are presented in Tables 13 and 14.
Table 13. Summary of Interval Continuous IBKP Pseudo-First-OrderNaphthalene Biotransformation Rate Constants and Reactor LiquidConcentrations
Cycl*Ptnod
(hr)
l
2a
2b
4
6a
6c
12
24
Naphthaleneinterval PMudc-First-Order Reader Liquid
Biotranstormation Rate Constant, k-|b* Concentrations(day1) KL'1)
Mm Maan
•324 .948
•199 -218
•259 -286
-168 .190
-595 -737
•1620 -2000
-456 -»eo
-416 -475
Max. Mm Mean Max
-372 0
•236 0
-317 0
•214 0
-678 0
4390 0.1 4.6 98
-504 124 14.0 156
-533 0
Trtatability and Scale-up Protocol* forPotynudear Aromatic HydrocarbonBioremediation of MOP Soils
93FINAL REPORT
VOLUME IJuly 1991
Table 14. Summary of Interval Continuous IBKP Pseudo-Second-OrderNaphthalene Biotransformation Rate Constants
Cyclt Interval Pttudo-Steond-Ordtr Biotrantformation Ratt ConstantPariod (xl 0"^ o c*u"1 day*1 )
(hr) Buad on Htttrotrophic Bacteria, k2Dh<
Min. M«an Max.
1 -2.35 -3.17 -4.51
2a -1.58 -2.18 -3.17
2b -2.32 -3.24 -473
4 -1.90 -2.66 -4.08
8* -6.82 -10.5 -16.7
Be -8.06 -12.5 -19.8
12 -11.1 -14.5 -20.4
24 -585 -833 -125
' Baatd on nah Q*n« Prob* PotitivtBacteria. *2bp*
Min. Mtan
-35.0 07.0
•21.9 -30.0
•334 066
-194 -27.6
-145 -225
03.4 47.0
•81.1 -107
008 -581
Max.
•675
03.4
-682
01.3
-355
-107
-150
-866
Interestingly, the first-order rate results (Table 13) show very highkinetic activity at the 8-hour perturbation period. This tends to increase theCV for all of the perturbation experiments (CV»1.0). Both second-orderrate constants (Table 14) tend to even out these 8-hour peaks with theresulting improvement in statistical dispersion (CV for kgbh* • 0.69 andCV for k2bP* = 0.86). The second-order constant based on nah geneprobe enumerations may have a higher dispersion as related to that basedon heterotrophic bacteria because of experimental difficulty in countingcolonies on gene probe plates from these continuous experiments.Colonies were irregularly shaped and mucoid leading to difficulty in colonydiscrimination. In this case, colonies from heterotrophic plate counts weremuch more distinct.
FINAL REPORT Trtatatility and Scal*-up Protocol* forVOLUME I 94 Porynucltar Aromatic HydrocarbonJuly 1991 Biorimtdiation of MOP Soils
The appearance of an apparent threshold concentration in thefirst-order rate constant formulation (Table 13) below which no degradationseems to occur for one of the 8-hour and the 12-hour cycles was unex-pected. Such a limit in substrate concentration has been proposed bycertain investigators. However, since this concentration limit was notobserved in numerous other experiments, experimental artifacts leading tothis result must also be considered.- The fact that these concentrationlimits occurred in successive experiments in only one reactor support thepossibility that these are experimental artifacts.
Fourier spectral analysis was used to determine the amplitude andphase angle of the sinusoidal reactor response with respect to the inputsignal. The results are presented along with comparative results from anactivated sludge petrochemical wastewater treatment system in Figure 32.
The CSTR MGP soil slurry system's amplitude ratio is relativelyconstant with the exception of the response at the 8-hour cycle. Herethree experiments were run because the first 8-hour experiment producedanomalous results and because interesting behavior was seen in priorwork in activated sludge near this frequency. All three experimentsproduced interesting but different responses than those at other frequen-cies.
^A duplicate run was performed at the 2-hour period to see if non-
reproducible results as seen in the 8-hour test replicate frequencyresponse tests were typical. Results of the 2-hour duplicate runs weresimilar to one another. A possible conjecture is that near the 8-hour periodthe system is inherently less robust and microbial system state bifurcationsresulting from some difference in the start-up procedure leads to a differentpopulation distribution with greatly different (and in this case improved)naphthalene biotransformation capability.
Titaiability and Sctlt-up Protocol* for FINAL REPORTPolynuelaar Aromatic Hydrocarbon 95 VOLUME IBiortmtdiation of MOP Soils July 1991
The phase angle frequency response behavior may be a clearer indi-cator of the critical frequencies for system response. In the activatedsludge work, the phase angle began to decrease without apparent boundas the frequency increased. Two observations may be made: a criticalsystem time constant appears at around 12-hour periods and the systemis a "non-minimal-phase" dynamical system. Systems that have dead timesin the naphthalene carbon processing metabolic pathways would exhibitthis non-minimum phase behavior.
The soil slurry phase angle response is very different than the acti-vated sludge response. No 'turning frequency" is evident down to theexperimental limit of 1-hour perturbation period. Therefore, in the soilslurry system, the critical frequency appears to be much less than the acti-vated sludge system and therefore, the response stability is greater thanthe activated sludge system. No statement is possible regarding thepresence or absence of naphthalene metabolic dead times in the soil slurrysystem. In terms of the response to naphthalene concentration perturba-tions, the soil system appears to be more stable than the activated sludgesystem. This is true even though the soil system appears to undergotime-variant state changes, at least at certain perturbation frequencies.
FINAL REPORT Traatability and Scala-up Protocols torVOLUME I 96 Porynueltar Aromatic HydrocarbonJuly 1991 Biortmadiation of MGP Soil*
O.3
0.1
0.01
0.001
O.ODOi
ACTIVATED SLUDGE
PANEL A
'' ^ \S/x \V/\ / MGP SOIL SLURRYV
-20!.01 0.1
FREQUENCY (CYCLES/HR)
Figure 32. Bode Frequency Response Diagram Comparing NaphthaleneBiotransformation Response in CSTR (Dashed Line; MGP Soil, Fitter Insert,
Experiment T2.34) and in Activated Sludge (Solid Line; Reference 9,10)
The concept of system time-variance and state changes is furtherenforced when the actual concentration response data for one of the8-hour replicates is seen (Figure 33). In this figure, the reactor liquid con-
Traalabilrty and Scil*-up Protocolt lorPo)ynucl«r AromMic HydrocarbonBiottmtdiation of MOP Soil*
97FINAL REPORT
VOLUME IJuly 1991
centration in the relaxed condition early in the experiment where the naph-thalene feed concentration is constant also is relatively constant. When the8-hour period perturbation begins, the reactor liquid concentration firstresponds, then drops precipitously to the detection limit. This behaviorwas seen in two of the three 8-hour replicates and in a number of earlymanual soil recycle mode experiments.
2000
Figure 33. System Time-Variance and Possible State Bifurcation Under8-Hour Period Perturbations (MGP Soil, Filter Insert). Reactor Liquid
Concentration-Open Squares, Feed Concentration-Solid Line
The phase angle shift can also be observed by plotting the rate ofbiotransformation versus the naphthalene feed rate. The two rates arenearly equal for all of the data, however, when the output signal is out ofphase with the feed cycle, an apparent hysteresis is observed (Figure 34,Panel A). This type of plot magnifies even small phase angle changes.
FINAL REPORTVOLUME IJuly 1991
Trtatabilrty and Sc»lt>up Protocols lorPolynuclaar Aromatic Hydrocarbon
Bioftmtdiation of MGP Soils
£.\
zo3ecOu.enz<at
200 400 600
700600 PANEL B, 12 HOUR PERTURBATION PERIOD
X0.
200 400 600
NAPHTHALENE FEED RATE (ug/hr)
Figure 34. Naphthalene Biotransformation Rate vs. Naphthalene FeedRate (Panel A, 1-Hour Cycle Period; Panel B, 12-Hour Cycle Period; MGP
Soil. Fitter Insert. Experiment T2.34)
Figure 35 shows the results of the naphthalene radiolabel fateanalysis. In contrast to the batch reactor experiment described in § 2.5,the radiolabel naphthalene was added as a pulse into the otherwise contin-uous liquid feed rather than pre-sorbed on soil. The results of this continu-ous experiment indicate that the naphthalene is oxidized immediatelyfollowing addition to the reactor, as indicated by the insignificant amount ofnaphthalene present at the first sampling point. The radiolabel shows up inthe aqueous phase as polar metabolites (30%) and 14CO2 (20%). Theremaining fraction (50%) is assumed to be resident in the solids containedin the reactor insert.
Treat ability and Sc«l**up Protocols forPorynucltar Aromatic HydrocarbonBior»m«diation of MOP Soils
FINAL REPORTVOLUME IJuly 1991
These results verify that the rates of biotransformation are very rapidwhen the naphthalene is available in the liquid phase. It is also apparentthat complete degradation of the metabolic intermediates was much slowerthan the initial biotransformation. The hydraulic residence time of thisreactor was 18 hours, sufficient to mineralize all but 30% of the naphthale-ne's polar metabolites. Longer residence times may lower the distributionof polar metabolites due to lessening of the reactor washout, but it is clearthat these compounds are more resistant to further biotransformation ascompared to biotransformation of naphthalene itself. Further characteriza-tion of these polar compounds and their toxicity remains a priority forfuture work.
ao
oh-z
110-100-90 ^80 r
70 r
60-50-40-30:20 r
10*-0
TOTAL RADIO-LABEL RECOVERED
WATER EXTRACT (POLAR METABOLITES)
CUMULATIVE CARBON DIOXIDE
NAPHTHALENE IN REACTOR LIQUID
24 28 320 4 8 12 16 20
ELAPSED TIME (HR)
Figure 35. Radiolabeled Analysis of Naphthalene Fates in Perturbed CSTRExperiment (1-Hour Cycle Period, MGP Soil, Fitter Insert, Experiment T2.34)
FINAL REPORTVOLUME IJuly 1991
100TrMttbilrty «nd Sc»l»-up Protocol* lor
Polynucliif Aromttic Hydrocarbon8 ior•mediation o* MOP Soils
3.5 CONCLUSIONS
• Kinetic evaluation of naphthalene biotransformation in the continuous Intrin-sic Biotransformation Kinetic Potential test (IBKP) established that the rateof biotransformation in this reactor system with the MGP soil studied canbe modeled with a first-order rate equation. Use of frequency-responsesystem identification approaches to generate "black box" phenomenologi-cal response models were, in this case, unnecessary.
• Naphthalene biotransformation in the continuous IBKP for the soils testedwas a robust process, resistant to perturbations in substrate concentra-tions, other organic concentrations, temperature, and dissolved oxygenconcentrations.
• The continuous IBKP, combined with frequency-response analysis, offers apractical method for establishing the stability and robustness of biodegra-dation in a specific process. Even though time-variant, state-chan.ge bifur-cations seemed to be possible, naphthalene biotransformation in the soilslurry system studied here was more stable to naphthalene concentrationchanges than an activated sludge system degrading naphthalene in acomplex petrochemical wastewater.
• The continuous IBKP test as developed in this work preferentially character-izes 'last" sorption/desorption mass transfer processes in specific wastes.This is seen in the similar comparative abiotic responses between MGPsoil and uncontaminated PAH-spiked soil. The method is insensitive to"slow" mass transfer processes since addition of MGP soil in a frequency-response mode is experimentally challenging. "Slow" sorption/desorptionkinetics may be more easily characterized in batch abiotic experimentssuggested in § 3.0.
Trtatabilrty and Scilt-up Protocol* (or FINAL REPORTPolynudtAr Aromatic Hydrocarbon 10i VOLUME IBior«m*di»tton of MGP Soils July 1*91
This page is intentionally ten blank.
4.0 OBSERVATIONS ON NAPHTHALENE BIOREMED1ATION KINETICS
4.1 COMPARISON OF KINETIC RATE DATA BETWEEN SIMILARREACTOR CONFIGURATIONS
When overall pseudo-first-order kinetic rate constants based on only thefirst and final concentrations are compared between similar reactor configura-tions (in this case between batch IBKP tests) very significant correlationsrelating the rate constants may be possible. As discussed in § 2.3, acorrelation between five MGP soils using the total PAHs analytically determinedled to a good relationship predicting the pseudo-first-order biotransformationrate constant for phenanthrene. It is also significant that correlations based ontotal carbon and/or microbial counts (pseudo-second-order) for these testswere poor. Here, the best relationships led to first-order predictions based ontotal PAH measurements.
4.2 PREDICTION OF BATCH TIME-CONCENTRATION PROFILESBatch biotransformation of PAHs in MGP soils are generally not first-order
or even elementary in terms of rate expressions. Further, they may be time-variant. Prediction of time-concentration profiles other than the endpoint willbe prone to significant error and is not recommended. Empirical rateexpressions must be developed and verified if reliable time-concentration pre-dictions are desired.
4.3 COMPARISON OF BATCH AND CONTINUOUS RESULTSTable 15 presents a summary of the naphthalene biotransformation kinetic
results from this study, and for comparison, from a prior study in an activatedsludge system operating on a non-MGP petrochemical wastewater.9
Traatabilriy and Scale-up Protocol* lor FINAL REPORTPorynucltar Aromatic Hydrocarbon 103 VOLUME IBioramadiation of MOP Soil» July 1991
In general, all kinetic rate constants (interval) for continuous tests exceedthose of the batch tests (overall). On a pseudo-first-order basis, the differ-ences are very significant with the batch test 1-3 orders of magnitude slowerthan the continuous test. The pseudo-second-order kinetics based on totalheterotrophic bacterial counts compare more favorably with the rate constantsfor the batch and the continuous IBKP (1-4 hour perturbation period) being thesame. The continuous IBKP second-order kinetic constants based on geneprobe assays are presented but should not be considered too critically sincethe gene probe counts in the continuous system were experimentally difficult toenumerate-colony morphology was mucoid, irregular and difficult to discrimi-nate.
Table 15. Summary of Naphthalene Biotransformation Rate ConstantsNaphthalene Biotransfonnation Rate Constants
Ttst System Ref. k^ or kib* k2bn or k2bn* *2bP or k2bP*(day1) (x10-7mlcfu'1 day1) (xlO^mLcfu'1 day1)
Batch IBKP This .1.66 -3.0 -13MGP Soil A work
Continuous IBKP This -592 -7.1 -76MGP Soil A* work
Continuous IBKP This .261 -2.6 -30MGP Soil A* work
Continuous 9 .144Activated Sludge1
• Average of all perturbed intervals (does not include date from relaxed intervals).6 Average of fast perturbation interval*-M hour period*.
For comparison between different system configurations and applica-tions, the batch IBKP overall rate constant results may give a reasonable sec-ond-order estimate of the average minimum continuous IBKP interval rateconstant results (those experiments perturbed with periods less than 4 hours).First-order comparisons between different system configurations and applica-
FINAL REPORT Treatability and Scale-up Protocols forVOLUME I 104 Potynuclear Aromatic HydrocarbonJuly 1991 Bioremediation of MOP Soils
lions are disparate and are not recommended for design and scale-up. Thepotential exists for using second-order lab-scale batch kinetic results toestimate the "worst-case" performance of lab-scale continuous systems.Coupled with new approaches to relate the tab-scale results with field-scaleresults, reliable "worst-case" field performance estimates may be possible.
Because of enhanced robustness and stability, it is possible that continu-ous processes may be more reliably scaled-up than batch processes. Thetendency toward time-variant state changes under perturbed conditions maynegate this possibility.
4.4 SECOND-ORDER KINETICSDifferences between experiments in the number of microbial PAH degrad-
ers is likely to have an effect on the overall degradation kinetics. The overallpseudo-first-order biotransformation rate constant from the batch experimentswas reported as -1.66 day1 which is significantly slower than the intervalresults from the CSTR study, -592 day*1 or the activated sludge study,*-144day"1.
Apparently, the continuous process leads to both higher total and PAH-degradative bacterial concentrations in the CSTR reactor than does the batchIBKP system. This supports the hypothesis that active bacterial cellconcentration is a critical parameter needed to predict PAH degradation rateconstants in MGP soil bioremediation systems.
The pseudo-second-order rate constants presented in Table 15 are likelyto be better activity parameters for system comparison and scale-up than thepseudo-first-order rate constants presented in the same table. However, theiruse in scale-up requires an a priori capability for predicting either the total het-erotrophic or genotypic bacterial concentrations in given field-scale systems.This is beyond current the state-of-the-art, but should be given a high R&Dpriority in microbial ecology for further work.
TrMttbility »nd Scallop Protocol for FINAL REPORTPor/nuclwr Aromatic Hydrocarbon 105 VOLUME 1Bior«m«di*lion of MOP Soil* Jury 1991
4.5 MASS TRANSFERMass transfer of PAH from the so/7 to the aqueous phase where it would
be most available to the microorganisms can have a large, possibly rate-controlling effect on the rates of degradation. A clear indication of this isevident when reviewing the work done by ReTeC. Both methods used forstudying PAH degradation were batch methods which resulted in comparabletreatment endpoints. However, the two test systems, a completely-mixedbatch slurry reactor and a batch soil pan reactor, gave vastly different results interms of rates of degradation. The rates in the completely-mixed reactor weremuch faster than in the soil pans, indicating that the PAH was more readilyavailable in the completely-mixed system. Experiments in both batch and con-tinuous systems in this work corroborate that mass transfer limitations can berate-limiting. However, negligible PAH biotransformation rates were alsoestablished suggesting that biotic activity could also be rate-limiting.
It is difficult to directly compare ReTeC's data with data from this study,largely because the ReTeC work reported conventionally-analyzed PAHsaggregated by ring number, and did not report comparable microbiat concen-trations. It is interesting to speculate that because ReTeC studied batchsystems, the differences in their kinetics may have arisen throughdesorption/diffusion limitations while the differences in the kinetics (batch vs.continuous IBKP) in this work may have resulted from both desorption/diffusionlimitations as well as enhanced microbial concentrations arising from the con-tinuous flow-through system.
4.6 NAPHTHALENE BIOTRANSFORMATION IN SOILS VS. WASTEWATERNaphthalene biotransformation (interval pseudo-first-order rate constants,
Table 15) in perturbed, continuous MGP soil slurry systems is faster than thosein perturbed, continuous activated sludge systems (-592 vs. -144 day*1.respectively). However, the activated sludge system's response to feed pertur-bations of less than or equal to 12-hour periods was significantly less than thatof the soil system.
FINAL REPORT Traatabilty and Scale-up Protocols forVOLUME I 106 PolynuclMr Aromatic HydrocarbonJuly 1991 Biortmodiatton of MOP Soils
The MGP soil systems here were able to respond to perturbations withoutsignificant changes in amplitude ratio or phase angle down to 1-hour perturba-tion periods. While being cautious not to disregard differences in the feedcomposition (the activated sludge feed was a concentrated petrochemicalwastewater), and not withstanding the apparent time-variant state bifurcationsevident in the soil slurry testing, one may speculate that there exists a signifi-cantly greater naphthalene biotransformation stability in the soil system thanthat found in the activated sludge system** 1° Frequency responsemethodology appears to be useful in classifying and comparing very differentnaphthalene-degrading systems.
4.7 FIELD-SCALE PROCESSESThe very fast rate behavior of all of these tab-based experiments as
compared to field-scale rates reported in the literature lead to the conclusionthat currently lab-scale testing may be useful for performance comparisonacross systems and applications, but that direct scale-up of rate information
b
may be dangerous. The impact of scale on kinetics is well-established inchemistry and chemical engineering and in these disciplines, separate"scale-up" relationships have been developed over the years to aid in thescale-up process. Analogous relationships for scale-up of bioremediation pro-cesses for environmental applications do not exist. Their development shouldbe a priority.
High pseudo-first-order rates of continuous processes relative to batchprocesses and the apparent enhanced stability and robustness of continuoussoil slurry bioremediation processes suggests that continuous processes('open processes" in terms of the reactor liquid) should be considered forfield-scale development. This effect may be related to the dilution andwash-out of soluble inhibiting metabolites and/or organisms in a continuoussystem vs. their accumulation in a batch system. Enhanced predictability andperformance of continuous processes over batch processes may make up forthe added capital and operating costs of treatment of a continuous process.
Tr«t«bilrty and Scalt»up Protocols for FINAL REPORTPolynucltar Aromatic Hydrocarbon 107 VOLUME IBioram*diation of MGP Soils July 1991
5.0 REFERENCES
1. Blackburn, J. W., 1988, Problems in and Potential for Biological Treatment ofHazardous Wastes, Proceedings of the 81st Annual APCA (now AWMA)Meeting, AWMA, Pittsburgh, PA.
2. Cushey, M. A. and D. J. Morgan, 1990, Biological Treatment of Soils Contain-ing Manufactured Gas Plant Residues. Topical Report, Gas Research Insti-tute, GRI-90/0117, Chicago.
3. Sims, R. C., J. L Sims, and R. R. Dupont, 1986, Human Health EffectsAssays, J. Water Pollution Control Fed.. 58:703-717.
4. EPA Interim Draft Report, 1990, Guide for Conducting Treatability StudiesUnder CERCLA: Aerobic Biodegradation Laboratory Screening, InterimDraft No. 2, RREL. ORD and OEER, U.S. Environmental Protection Agency.
5. Blackburn, J. W., 1989, Is There an "Uncertainty Principle" in Microbial WasteTreatment? In Biotreatment of Agricultural Wastewater. M. Huntley," Ed.,CRC Press. Boca Raton, pp. 149-161.
6. Blackburn, J. W., 1987, Prediction of Organic Chemical Fates in BiologicalTreatment Systems, Environ. Progr.. 6(4) :217-223.
7. Levenspiel. 0., 1962, Chemical Reaction Engineering. John Wiley. New York.
8. Blackburn, J. W.. W. L. Troxler, K. N. Truong, R. P. Zink, S. C. Meckstroth, J.R. Florance, A. Groen, G. S. Sayler, R. A. Minear, O. Yagi, and A. Breen,1985, Organic Chemical Fate Prediction in Activated Sludge Treatment Pro-cesses. Report EPA 600/S2-85/102. U.S. Environmental Protection Agency,(PB 85-247647. NTIS, Springfield, VA).
9. Blackburn, J. W., 1989, Improved Understanding and Application of Hazard-ous Waste Biological Treatment Processes Using Microbial SystemsAnalysis Techniques, Hazardous Wastes & Hazardous Materials.6(2):173-193.
Tnaubilty and Scal*-up Protocol* for FINAL REPORTPotynuclMr Aromatic Hydrocarbon 109 VOLUME IBior«m«diation of MOP Soil* July 1991
10. Blackburn, J. W.. 1990, Frequency Response Analysis of Naphthalene Bio-transformation Activity, Biochemical Engineering VI, Annals of the New YorkAcademy of Sciences. 589:580-592.
11 . K. Van't Riet, 1983, Mass Transfer in Fermentation, Trends in Biotechnology.
12. Sanseverino, J., C, Werner, J. Fleming, B. M. Applegate, J. M. H. King, G. S.Sayler, and J. W. Blackburn, 1990. Molecular Analysis of Manufactured GasPlant Soils for Naphthalene Mineralization, Proceedings of the Symposiumon Oil. Gas, and Goal Biotechnology. Institute of Gas Technology. Chicago,in press.
13. King, J. M. H., J. Sanseverino, P. M. DiGrazia, B. M. Applegate, G. S. Sayler,and J. W. Blackburn, 1991 . Molecular Bioanalytical Methods for MonitoringPolynuclear Aromatic Hydrocarbon Biodegradation in Manufactured GasPlant Soils. Final Report. Vol. H, Gas Research Institute.
14. Day, R. A. and A. L Underwood, 1967, Quantitative Analysis. Prentice- Hall,Englewood Cliffs, p. 64.
15. King, J. M. H., P. M. DiGrazia, B. Applegate, R. Burlage. J. Sanseverino, P.Dunbar, F. Larimer, and G. S. Sayler, 1990, Rapid, Sensitive BioluminescentReporter Technology for Naphthalene Exposure and Degradation, Science.249:778-781.
16. Sachs, L, 1982. Applied Statistics. Spinger-Verlag. p. 249.
17. Natke, H. G. and N. Cottin, 1988, Introduction to System Identification: Fun-damentals and Survey, in Application of System Identification in Engineer-ing, H. G. Natke, Ed.. Springer-Verlag, pp. 4-107.
18. Bode. H. W.. 1945. Network Analysis and Feedback Amplifier Design. D. VanNostrand, Princeton.
19. Murrill, P. W., 1967, Automatic Control of Processes. International Textbook,Scranton. Chap. 13.
FINAL REPORT Tr««abil<ty and Sc«lt-up Protocol* lorVOLUME I 110 Polynucltv Aromatic HydrocarbonJuly 1991 B*r»m«di«ion of MOP Soil*
20. Sayler, G. S. and J. W. Blackburn, 1990, Biodegradative Processes and Bio-logical Waste Treatment: Analysis and Control, In Agricultural and SyntheticPolymers. Biodepradability and Utilization. J. E. Glass and G. Swift, Eds.,Chap. 2.
21. Sayler, G. S. and T. W. Sherrill. 1983, Polyaromatic Hydrocarbon Degrada-tion and Fate in Aquatic Sediments, in G. Babolini and G. Girbino, Eds.,Environment and Lung Diseases. La Grafica Editorials, Messina, Italy, p.184.
22. Blackburn, J. W., W. L Troxler, K. N. Truong, R. P. Zink, S. C. Meckstroth, J.R. Florance, A. Groen. G. S. Sayler, R. A. Minear, 0. Yagi, and A. Breen.1985, Organic Chemical Fate Prediction in Activated Sludge Treatment Pro-cesses. Report EPA 600/S2-85/102, U.S. Environmental Protection Agency,(PB 85-247647, NTIS. Springfield. VA).
23. Blackburn. J. W., W. L. Troxler. and G. S. Sayler. 1984. Prediction of theFates of Organic Chemicals in a Biological Treatment Process«an Overview,Environ. Progress. 3{3):163-176.
24. Stephanopoulos, G., 1984, Frequency Response Analysis of Linear Pro-cesses, in Chemical Process Control. Prentice-Hall, Englewood Cliffs, Chap.17.
Testability and Scal*-up Protocol* for FINAL REPORTPolynuctoar Aromatic Hydrocarbon 111 VOLUME IBiof»mtdi«ion of MOP Soils July
APPENDIX A. SUMMARY OF CALCULATIONS AND MODELS
A-1. BATCH AND CONTINUOUS REACTOR DESIGN MODELSDerivations for the equations in this section are presented in LevenspielT
The zero- and first-order Rate Equations are, respectively:
dCA-r,-- ——-*, Eqn.A.l
dCA~-7- - * i C x Eqn.A.2at
In any constant volume batch reactor, the overall design equation is:
rd.x,t = CAQ ——- Eqn.A.3J ~rA
where the fractional conversion of compound A is
Eqn.AAAQ
The integrated Design Equations for zero- and first-order Rate Equations,therefore respectively become:
Batch, zero-order form-
CAO-C* - C A O X A - kzt Eqn.A.S
FINAL REPORT Tr««t«bilrty *nd Sc«l*-up Protocols forVOLUME I A-1 Polynuetov Aromttie HydrocarbonJuly 1991 Bio ft mediation of MOP Soil*
Batch, first-order forms-
CA-In-— - A : , f £qn./!.6C AQ
. - • • ' Eqn.A.7AQ
- ln(l -*„)- *i* Eqn.A.S
XA- \ -c"*1 ' Eqn.A.9
In a perfectly-mixed, constant volume, continuously stirred tank reactor(CSTR), the design equation is found by using the form for a backmix reactor;
r v'•* AQA A _. , . -T - ——— Eqn.A.lOr A
CSTR Zero-order form-
CAO-CA-C^XA-k'2t Eqn.A.ll
CSTR First-order form-
Tft•ttbility and Sc«l»-up Protocol* for FINAL REPORTPolynuclttr Aromatic Hydrocarbon A-2 VOLUME!Biortmtdiation of MOP Soils Juty 1991
A-2. PROPAGATION OF ERROR TO THE KINETIC RATE CONSTANTSThe parameters that vary as a function of experimental errors in the above
Design Equations are CAO (tne initia' concentration of compound A for a batchreactor or the feed concentration of A for a continuous reactor), CA (the con-centration of A at time, t, in a batch reactor or the effluent or concentration in acontinuous reactor), t (the treatment time in a batch reactor), and T (thehydraulic residence time in a continuous reactor).
Assuming that the chosen Rate Equations apply and that each rateconstant is indeed a constant, the estimation of errors in these parameters onthe rate constants, kz and k-j, can be estimated by assuming values forpositive and negative error in the parameters and calculating the resulting rateconstants. The magnitude of the relative error in the rate constant can then becalculated as follows:
*-*.,Magnitude of Rate Constant Relative Percent Error - ———AT
Eqn.A.\3
Assuming the simplest possible case where all removal from this batchexperiment is biotic in nature (no important abiotic processes are at play),where the reactor is homogeneous (no error is incurred due to sampling fromdifferent locations at one time), and where the biotransformation rate equationis either zero- or first-order; the estimation of experimental errors on the result-ing rate constant is straightforward.
Eight types of error propagation can occur when considering negativeand positive error in three variables. These cases are shown in Table A-1.
FINAL REPORT Traatability and Seal*-up Protocols forVOLUME I A-3 Porynucltar Aromatic HydrocarbonJuly 1991 Biortm»di*tion of MOP Soils
F
For the abiotic experiments, X=0 and R=0, resulting in further simplifica-tion to the form:
V p. ) dt V A0 \V °
With a sinusoidal feed input of naphthalene of the form:
- — -(1 + cos(uuO) Eqn.AAl
Equation A.40 can then be solved analytically to give:
) - a c o s ( u ) f ) + e " '° (acos(ooO~
where
r- , ,-,Eqn.A.43V
and
Laplace transformation of Equation A.40 yields:
Twubility tnd Sc»lt-up Protocol! lor FINAL REPORTPolynuclwr Aromatic Hydrocarbon A-10 VOLUME I8'0(»m«di«lion of MOP Soils Jury
Eqn,AA4I/I 1
Eqn.AAS
The amplitude ratio and phase angle for a transfer function of this formare:
AR - Eqn.AA6
and
*- tan ' 1 -— ] Eqn.AA?V a j ^
A-5. FLOW EQUATIONS FOR THE SINTERED METAL REACTOR INSERTThe liquid mass flow through the reactor insert is proportional to the
surface area and the pressure gradient:
Mass Flux- A'.-^Ap Eqn.AAB
fd-A' 2nrApd2 Eqn.AA9
Jo
-2nrK[ \ p 0dz+ f (p0-p,)dz] Eqn.A.SOJo J t
Pom Pu,^ Eqn.A.Sl
P . -pJz -O Eqn.A.52.
FINAL REPORT Trvctability and Se«l*-up Protocol* torVOLUME I A-11 Polynuclttr Aromatic HydrocarbonJuly 1991 Biortmadiation of MOP Soil*
Term(G)
(EffluQnt\Flow
I r>\ Kate J
f .( „Conccnlra t iony
/ Compound \EffluentCell --FA'C
\Concent ration/Eqn.A.29
/" Compound \^acZor°/^as --
V Concenfraaon y
Substitution of Equations A.23 through A.30 into Equation A.22 yields:
The reactor liquid volume is directly related to soil concentration andslurry volume:
—P.The compound reactor offgas concentration is proportional to the
compound reactor liquid concentration:
Eqn.A.33
The compound reactor soil concentration is proportional to thecompound reactor liquid concentration:
TftitabiliTy and Scal*-up Protocol* torPolynuclwr Aromatic HydrocarbonBiort mtdiation of MOP Soils
A-8FINAL REPORT
VOLUME IJuly 1M1
CSA - 0.62/^A^C, - KPCA Eqn.A.34
The compound reactor biomass concentration is proportional to thecompound reactor liquid concentration:
/ i P & K o————— C,, - KbCA Eqn.A.35
Substitution of Equations A.32 through A.35 into Equation A.31 yields:
dC < d\J F F Rf \ + is I'* A" Y" ^ -•- •*• ( fC r* \ » — (* — — f* f ] •+• k' \ f + if V + A ' \ _ _
Eqn.A.36.
where
A' - 1A' , -— -—— fqn..1.37
P.
and
QH9 p
Equation A.36 is valid for all experiments performed in the CSTR.
For experiments with the sintered metal reactor insert, terms F and G ofEquation A.22 are zero. Also, the liquid fraction in the effluent is one, and thesoil concentration is constant. This results in simplification of Equation A.36 tothe form:
FINAL REPORT TfMtibiltty »nd Scalt-up Protocols forVOLUME I A-9 Potynuel«tf Aromatic HydrocarbonJuly 1991 B>ortmtdi»tion of MGP Soili
In these calculations, C } I ,_0 was taken as the radiolabet concen-
tration in the "time*0" hexane extract samples and not the expected concentra-tion based on what was added.
A-4. MODEL FOR DYNAMIC SYSTEMS ANALYSIS OF A CSTR SOILSLURRY REACTORf Compound Mass
AccumulationintheLiquid
(Compound}I MassFlou. \\ intheFeed (V (£) /
CompoundMassF louinthe
B i o m a s s E f f l u e n t
( CompoundMassAccumulation
intheSoil
Compound \MassFlouinthe \LiquidEf f luent (
(CompoundMassI Accumulation+ <\ intheBiomass\ (C)
/ CompoundI MassFlouinthe\ SoilEf fluent
/ Compound \I Mass Flow V\ i n t h e O f f g a s [
Compound MassConversion &»
Biotransformation
Eqn.A.22
Term(A)-
f Time \Katoof
\ChangeJf Liquid \ ( Compound \1 . ReactorLiquid\VolumeJ[ ^ - „• /V Concentration J
dl ' tat£qn.A.23
TrttUbilrty and ftcftlt^up Preloeelt forPolynuelMr Aromttie HydrocarbonBiortmtdiation of MOP Soils
A-6FINAL REPORT
VOLUME IJuly 1991
Term(B) -/ Tim« \
Kateof\ChangoJ
f Slurry \\Volume)
( Soil \f Compound \Volume KeactorSoil
\ Factor J \Concentration J_
d_dt
(VV\ -
L VP,
Term(C)-/ Time \
KCltQOf
\ChangeJ
/ *, \f ». x- .. \(Slurry \( ReactorCell \,, . „ .\VolumeJ\ConcentrattonJf Compound \l B f _ M )ReactorCQll\ „ /\Conccn(ra(ion/_
**
Eqn.A.24
Eqn.A.25
Term(D)fFeed\f Compound \
FloiL Feed - FCAQ\RcLtQ J \Concentration J
Eqn.A.26
[ E f fluent\ / E f f luent \ f CompoundFlou LiquidVotume LiquidE f f luentKate J\ Fraction J\ Concentration
Eqn.A.27
Term(F)
f E f f l u e n t \ f Effluent \/ Compound \- Flou Soil\'olume SotlEf fluent\ Kate J\ Fraction J\ConcentrationJ
F — CSn
Eqn.A.2B
FINAL REPORTVOLUME IJuly 1991
A-7Tfaatafaility and Scalt-up Protocols tor
Porynuclaar Aromatic HydrocarbonBiortmtdiation ol MOP Soils
Table A-1. Types of Error in Three-Parameter Design Equations
Parameter A B C D E F G
CAO + . + . + .+CA + + - - + +
H
-
:A-3. CALCULATION OF PSEUDO-FIRST-ORDER BIOTRANSFORMATIONAND MINERALIZATION RATE CONSTANTS FROM LABORATORY-SCALEEXPERIMENTS
For biotransformation rates, the average value of the three hexane extractscintillation counts was recorded as a time series. A pseudo-first order bio-transformation rate constant can the calculated for each time increment byapplication of Eqn. A.6 to each time interval. This "interval-based" rate constantcarries a superscript * and is defined as follows:
k*.m_ v ^ . - n - . , Eqn.A.14.
If the above equation is applied to intervals of time series data, the initialconcentration is the concentration at the beginning of the sampling interval andthe rate constant is calculated accordingly. This approach lends itself to arecursive formula where the conversion is calculated for each interval after thepreceding concentration is established.
,.„., Egn./ . lS
An alternative approach is to use a formula to calculate the rate constantwhere the initial concentration is the starting concentration for the experiment,
Testability and Scatt-up Protocols for FINAL REPORTPorynucltar Aromatic Hydrocarbon A-4 VOLUME IB'0f*m«diation of MGP Soili July 1991
but the conversion is the conversion between each successive interval. Inpractice, these rate constants apply to rate equations where concentrations arecalculated from the starting concentration and do not require a recursiveapproach.
XA-XA\\:l0+XA\\:2
l+...+ XA\\:mH.l Eqn.A.lt.
\"nA I (-0
r - r r — rA t(-
AO AO
Eqn.A.17
,, ,.„-,
.tn-tn.
The pseudo-first-order mineralization rate constants were calculated byapplication of Eqn. A. 19 on the concentration of labeled carbon atoms.
i ( 1 - Y I1"1 ^ln[ A -I '*co2 I r - n - i Jt — t1 n * n- 1
Eqn.A.2Q
where
VA u ,.B.|C(ft«Man«»xtrocf)
FINAL REPORTVOLUME IJuly 1991
A-5Trtattbility and Se«lt-up Protocols for
Po*ynucl«»r A(om«ic HydrocarbonBiortmtdiation of MOP Soils
Substitution of Equations A.51 and A.52 into A.50 yields:
Mass Flux* 2nrpu/K"' -o
Idz Eqn.A.53
Solving:
MassFlux-2nrpwK[ Id-- I Eqn.A.S4
Eqn.A.SS
Maximum flow rate occurs when I = d:
| m , x -n rA 'd 2 Eqn.A.56
A-6. NOMENCLATUREAR is the ratio of the output signal amplitude and the input signal amplitude.As is the surface area of the filter insert.B is the concentration of biomass present.CA. CAS, CA^, or CAV are the concentrations, of compound A in a given phase
or compartment. Superscripts s, b, and v represent the concentration ofcompound A in the soil, biomass and offgas phases or compartments,respectfully, while no subscript indicates the concentration in the aqueousliquid phase.
CAO is the starting concentration in the case of a batch reactor system or thefeed concentration in the case of a continuous reactor system.
CQ is the maximum feed concentration.d is the diameter of the fitter insert.F is the volumetric flowrate of the feed stream to a continuous reactor.
TfttlabilfTy and Scal*up Protocol tor FINAL REPORTPorynuctov Aromatic Hydrocarbon A-12 VOLUME IBiortmtdiation of MOP Soils July 1991
fj_ is the weight fraction of lipids in biomass.foe 's tne weight fraction of organic carbon in soil.H is the Henry's Law constant in units of weight concentration in the gas over
weight concentration in the liquid.K is a mass flow constant.Kb is the overall partition coefficient of compound A into biomass.Kg is the overall partition coefficient of compound A into the reactor offgas.KQW is the octanol-water partition coefficient of compound A.Kp is the overall partition coefficient of compound A in soil.k-|, kib, k-|m; k2, k2b, or k2m represent the first-order or second-order overall
rate constants with subscripts signifying first-order, or second-orderRate Equations and superscripts b or m signifying biotransformation ormineralization, respectively. These may be evaluated over many timeintervals or a single interval depending on the nature of the calculation.An asterisk "*" superscript signifies that the constant is being evaluatedover a single time interval and is based upon the compound concentra-tion at the beginning of that interval for a batch reactor configuration
kz is the rate constant for a zero-order rate equation.k., is the rate constant evaluated when its parameters are assigned certain
errors that propagate through to the constant.I is the height of the filter insert.Q is the volumetric flowrate of gas leaving a continuous reactor.R is the mass removal rate of compound A arising from biotic mechanisms.TA is an expression or rate equation for the change in compound A over time.s is the independent Laplace variable.t is either elapsed time or specified times in a time series.V is the volume of the slurry (liquid and solids) in a soil slurry reactor.V|_ is the liquid volume in a soil slurry reactor.W is the weight of the soil present.XA is the conversion of compound A. This may be evaluated over various time
intervals as noted by X A | US for example.Ap is the pressure drop across the filter insert.
FINAL REPORT Trtatabillty and Sca(*-up Protocol* forVOLUME 1 A'13 Potynuclaar Aromatic HydrocarbonJuly 1991 Biortmodiation of MOP Soils
* is the phase angle between the input signal and the output signal.
p, is the density of the soils present.
p 6, p L are the densities of the biomass and the reactor liquid, respectively.
T is the reactor residence time in a continuous reactor.
to is the frequency of oscillation.
Tr««tability and Scal*-up Protocol* lor FINAL REPORTPotynuclaar Aromatic Hydrocarbon A-14 VOLUME tBioramtdiation of MQP Soil*
APPENDIX B. DESCRIPTIONS OF EXPERIMENTS
[This section includes descriptions of the nature and design of each majorexperiment cited in this report. These are intended to enable the comparison ofthe experimental methods from case-to-case. This Appendix is extensive and is
available on request under separate cover from the GRI Project Manager.]
FINAL REPORT TfMttbilrty and Sc«l*up Protocol* lotVOLUME I B-1 Potynueltv Aromatic HydrocarbonJuly 1991 Bior«m«difttion of MOP Soils
