ARCHIVED REPORTS_XR0002666 - Laserfiche WebLink

Home Browse Search

302 HYDROCARBON CONTAMINATED SOILS AND GROUNDWATER GROUNDWATER REMEDIATION LIMITS 303 1 and data points are displayed when the regression is complete, with goodness recovery well ranged from 70 to 300 ppb the nonllnear regression wati fit by Of tit statistics Graphic displays show plots of the data, the regression line, the C = 980 C-0W85t + 70 where t is in days (rz = 0 87) Thus, the asymptotic, Lon- asymptote, residuals, and regression prediLtion intervals, where applicable CLatration was estimatcd to be 70 ppb The linear asymptotic procedure(see Figure Hypothetical applications of the software are illustrated in the following seL(1on 17 3) identified the asymptotic period as Days 210 through 510 The data from day 360(300 ppb)was removed as an outlier following Florida guidance insofar as the concentration exceeded the mean of the asymptotic region by more than APPLICATIONS two standard&viations Data from the asymptotic period exhibit a slope of—0 20 ppb/day which was not significantly different from zero The average concentra- Certain states have adopted or proposed rules that recognize achievement of tion in the asymptotic region was 120 ppb The linear asymptotic, regression an asymptotic limit as a basis for termination of groundwater pump-and-treat predicted that concentrations would not exceed 200 ppb during the next quarter remediation systems Specifically, the State of Florida rule 17-770, Florida Ad- The following quarter's data failed to confirm the linear asymptotic regres- ininistrative Code(F A C ) specifies that after asymptotic conditions have been stun on day 540, the observed concentration of 250 is outside the prediction in- ob,,crved for one year, reinediation may be halted based on a statist1Lal evalua- tLrval Repeating the analysis after day 600, however, shows that the nonlinear tion that includes both the nonlinear and linear regressions described in this Lhapter iegression yielded stable results C = 980 e-000831 + 65 The linear asymptotic These procedures are applied to a hypothetical data set presented in the State of i prOLedure(see Figure 17 4)identified the asymptotic period as days 420 through Florida guidance pursuant to that regulation as well as actual site data provided 600 with an asymptotic Loncentration of 78 ppb,a slope of—020 and predicted by API that LonLcntrationti would not exLeed 100 ppb during the next quarter The final analysis, after day 660, confirmed the predictions made after day Florida Example 600 New data fell within the prediction interval estimated after day 600 The nonlinear regression (see Figure 17 5) yielded C = 980 a-0U083 + 66 (r1 = After 17 months of remediation (see Figure 17 2) concentrations appeared to r 0 88) The linear asymptotic procedure, however, provided much stronger be leveling off, for the previous 12 months the concentration of benzene in the 12001 • raw data used in regression 12001 • raw data used in regre3s-om O = raw dato not used ,n regression O - row doto not used in regress-on A fit - 03ymFltote fel - 1000= asymptote a prediction interval 1000 s a n a Sao 0 0 CL CL 900 • • z Z 0 o • ti a O i 600 a .- 600 L O 7 i w v p ANALYSIS Z ANALYSIS v 400 PERFORMED u 400 PERFORMED k O O i O O O 200 O • - 200 • ----- ----- • + • • • O ---------------------- -------------- - 6 -" 120 240 360 480 600 720 � I _ 120 240 360 4R0 600 720 TIME (DAYS) TIME (DAYS) Figure 17 3 Florida example linear asymptotic regression performed after 510 days of Figure 17 2 Florida example nonlinear reriregaton oerformnd after Stu rfgva of rnmorfi-ifinn romi-H,n1 nn