|
'I
<br /> 308 HYDROCARBON CONTAMINATED SOILS AND GROUNDWATER GROUNDWATER REMEDIATION LIMITS 309
<br /> more efficient in identifying an asymptote, condition with less data The weak- program When the asymptotic average is above a target cleanup concentration,-
<br /> ness of the exponential regression technique appears to be related to its assunmp- a technologically lirmiting condition may have been reached that would signal a
<br /> tion of a zero asymptote, while most of(lie cliarac.tcristic data sets clearly exhibit reevaluation of the ongoing remedial response
<br /> a nonzero asymptote t Decisions about the further course of action should be based on a variety of
<br /> In Phase 2 of this study, a nonlinear regression procedure has been developed factors cost-eftectivene5s of continuing mplace technology, technical feasibility
<br /> to address the weakness of the exponential procedure PC software has been de- and cost-effectiveness of alternative remedial technologies, and the health and
<br /> veloped to facilitate broader application of the procedures This software, to be - environmental risks posed by the site In addressing the latter issue, it would be
<br /> released by API during 1991, has been applied to two additional data sets (1) important to determine whether the primary oblcctive in maintaining the systcin
<br /> a hypothetical data set included in a guidance document by the Slate of Florida, was to contain further migration rather than restore the aquifer
<br /> and (2) an additional data set provided by API
<br /> F
<br /> Florida Rule 17-770, F A C provides for conclusion of groundwater pump-
<br /> and-treat reimediatton based on a finding that an asymptotic limit has been reached ACKNOWLEDGMENTS
<br /> The PC software is designed to meet the requirements of the Florida rule while
<br /> also providing options for alternative analyses A typical application of the soft- Financial support for this research was provided by time American Petroleum
<br /> ware would include Institute (API)
<br /> 1 Subjective determination that concentrations in a recovery well have
<br /> leveled off and remained steady with random fluctuations for more than REFERENCES
<br /> five sampling events,
<br /> 2 application of the nonlinear regression procedure to verify that the data 1 Tucker, W A , M W Kemblowski, and D E Draney "Technological Limits of
<br /> generally tit an equation of the form C = Coe-k' f Cr, (this conclu- Groundwater Remediation A Statistical Evaluation Method,"in Proceedings of Petro-
<br /> sion can be based on rl or on subjective interpretation), leum Hydrocarbons and Organic Chemicals in Groundwater (National Water Well
<br /> 3 application of the linear asymptotic procedure to objectively identify the 4 Association, Dublin, OH, 1989)
<br /> asynmptotmc region, 2 Environmental Science & Engineering, Inc "Tethnological Limits of Groundwater
<br /> 4 remove outliers and repeat steps 2 and 3 (optional), No ,, 1991 Remediation A Statistical Evaluation Method ' American Petroleum Institute, Publi-
<br /> cation5 use the linear asymptotic results to predict the range expected in A up-
<br /> 3 Tucker W A , C T Huang,J M Bral, and R E Dickinson "Development and
<br /> coming sample results, and Validation cif the Underground Leak Transport Assessment Model (UL'i RA)," in
<br /> 6 verify the model by checking that a minimum of one calendar quarter's Proceedings of Petroleum Hydrocarbons and Organic Chemicals in Groundwater(Na-
<br /> sampling results (not less than two subsequent data points)confirm the tional Water Well Assoc.tation, Dublin, OH, 1986)
<br /> model's prediction 4 Beers, Y hinoductiort to the 7liewy of Eiror(Reading, MA Addison-Wesley Pub-
<br /> lishing Company, Inc , 1957)
<br /> A variety of additional applications are supported by time PC software The linear 5 Gumbel, E J Sum vic s of Errnetner(New York,NY Columbia University Press, 1958)
<br /> regression can be used to estimate the time to achieve specific,concentration tar- 6 Meyers, R H Classical and Modern Regression with Application~ (Duxbury, MA
<br /> gets The user may subjectively identify the asymptotic region and perform linear r Duxbury Press, 1986)
<br /> regression analysis(distinct from the objective procedure referred to as the linear
<br /> asymptotic analysis) The exponential regression analysis is still supported, but ;
<br /> is of limited utility (authors' opinion) The first-order removal rate constant, k, -
<br /> Calculated by the nonlinear regression has broader implications regarding the Oper-
<br /> ating. conditions of the remedial system and contaminant fate and transport 12
<br /> 1 lie development of the asymptotic limit statistical approach was prompted by
<br /> the observation that groundwater pump-and-treat systems at petroleum conlanmi-
<br /> nated sites are effective initial] at reduce groundwater contaminant levels but
<br /> Y reducing li
<br /> eventually reach a technological limit of reduced effectiveness Once an asymp-
<br /> totic region of a groundwater data record is defined, the average asymptotic con- 1'
<br /> ceniration can be compared to target cleanup levels to ev mluate the restoration ,
<br />
|