Environmental Resources Management, Inc. Resources Management, Inc. 855 Springdale Drive • Exton, Pennsylvania 19341 • (215) 524-3500 • Fax One: 524-7335 • Fax Two: 524-7798

Environmental Resources Management, Inc.855 Springdale Drive • Exton, Pennsylvania 19341 • (215) 524-3500 • Fax One: 524-7335 • Fax Two: 524-7798

10 April 1990

Mr. Eugene DennisRegional Project ManagerUS Environmental Protection AgencyRegion III841 Chestnut BuildingPhiladelphia, PA 19107 FILE: 723-01-03

Dear Mr. Dennis:On behalf of Textron Lycoming, enclosed please find four copies of ERM'sresponses to EPA and PADER's comments on the draft RemedialInvestigation, Risk Assessment, and Feasibility Study reports. Theseresponses have been submitted to you on an accelerated schedule tofacilitate your goal of providing these documents to the public for review by15 April.In the course of addressing your comments, we have concluded that ameeting as soon as possible with EPA and PADER staff is essential. Thismeeting will serve to ensure that the key personnel from both regulatorybodies are fully cognizant of conditions at the Textron Lycoming facility, thenature of site-related risks, and the performance of previously implementedremedial actions. Such a meeting will also enable Textron and ERMpersonnel to fully understand the regulatory constraints and technicalrationale for both EPA and PADER's approach to the site.Pursuant to the approved schedule established for the project, TextronLycoming is allotted 45 days for response to agency comments on theRI/RA/FS. Because Textron Lycoming agreed to greatly accelerate itsresponse to comments at the request of EPA, the need for such a meeting ismuch more pressing because of the lack of time available between receipt ofcomments on the RI/RA/FS and EPA's drafting of the Record of Decision.Given the shortened response schedule, Textron and ERM both feel thatmeeting with EPA and PADER as soon as possible is of critical importancein order to allow a mutual understanding of the various outstanding issuesassociated with the proposed remedy. Textron and ERM believe that a face-to-face meeting is needed to facilitate discussion of maps and figurespertaining to the topics addressed in the agencies comments. We arewilling to hold meeting at a location and time of your choice. Thank you in

AR303351*

An affiliate of the Environmental Resources Management Group with offices worldwide —— w—lQr-

Mr. Eugene DennisUSEPA - Region III10 April 1991Page 2

advance for your efforts to arrange such a meeting among the variousinterested parties as soon as possible, preferably within the coming week.

Very truly yours,

Richard T. Wroblewski, P.G.Senior Project Manager

Charles A. Bandoian, Ph.D., P.G.Principal

RTW/rebEnclosurescc: L. Trefsger, Textron Lycoming

P. Boob, Textron LycomingP. Duff, TextronT. Kraig, TextronB. Kelley, TextronR. Baker, ERMU. Tyagi, ERMT. Schuller, ERML. Hoose, ERMS. Sayko, ERM

AR303355

Environmental Resources Management, inc.855 Springdale Drive • Exton, Pennsylvania 19341 • (215) 524-3500 • Telex 4900009249

10 April 1991

Mr. Eugene Dennis FILE: 723-01-01-01Regional Project ManagerUS Environmental Protection AgencyRegion III841 Chestnut BuildingPhiladelphia, PA 19107

Re: Response to USEPA and PADER Comments: 15 January 1991 Draft RemedialInvestigation Report for the Textron Lycoming Williamsport Facility

Dear Mr. Dennis:On behalf of Textron Lycoming, Environmental Resources Management, Inc.(ERM) is submitting a response to the United States Environmental ProtectionAgency (EPA) and the Pennsylvania Department of Environmental Resources(PADER) comments concerning the Draft Remedial Investigation Report for theTextron Lycoming Williamsport Facility. The comments received by TextronLycoming in separate transmittals on 6 and 11 March 1991 are italicized;'responses to the comments follow.Hydrogeological, Sampling, Soil CommentsGenera fP»gfiH)1. There is no mention potential sources of ground water contamination. As

noted during the site visit, there was an odor which may have been a volatileorganic contaminant. It would be in the interest of the Responsible Party(RP) to discuss other possible sources of contamination or providedocumentation that a review of all other potential sources (throughinterviews and reviews of file material from other hazardous wastegenerators in the vicinity) have been identified.It is understood that this paragraph refers to other potential primaryresponsible parties (PRPs) for the ground water contamination within thestudy area. This issue was addressed during the Remedial Investigation(RI) and no other continuing sources to ground water contamination wereidentified from available sources. One wood refmishing company (Brodart),is south of the Textron plant within the defined plume area, but noinformation is available on their operation. Neither ERM nor TextronLycoming has the authority to obtain records from industries active in thearea to check on operations and disposal practices.

AR303356The

An affiliate of the Environmental Resources Management Group with offices worldwide

Mr. Eugene DennisUSEPA10 April 1991Page 2

North of the eastern end of the plant, a dry cleaning establishment existed inthe past. No information regarding this operation is available. Located eastof the WMWA well field is the Williamsport City Dump and an active junkyard. There are several industries and commercial establishments locatedwest of Lycoming Creek, but the names have not been obtained. It is knownthat there was an abandoned disposal lagoon containing liquid tannerywaste located approximately 300 feet west of WMWA pumping well PW-8, asreported by Moody and Associates, Inc. in their 1969 report for the WMWA.Also, there is unsubstantiated information that filling occurred along thebanks of Lycoming Creek during the 1930s to 1940s. The nature of the fillmaterial is unknown. ^It should be noted that portions of the plant were used and/or owned duringWorld War II under U.S. Government auspices. The precise nature of thework and chemicals used are not known, but generally it is believed that theU.S. Government used the same types of chemicals used by TextronLycoming. The Navy has reported to Textron Lycoming that its records forthis time period have been destroyed. Army records are reportedly stillintact back to the 1950s, though these have not yet been made available toTextron.It is not clear where in the study area the EPA noticed odors of volatileorganics. Were the odors from another industry or from the TextronLycoming facility? This will need to be clarified before this comment can befurther addressed.

2. There is a discussion of the former consultant's (Chester) ground watermodeling efforts but no discussion of modelling in the RI by the RP's currentconsultant. The modeling will be necessary for defining volumes of soil to beremediated and the length of time necessary for operation of the groundwater treatment system among other aspects of the Feasibility Study. Thesemodeling efforts may be presented in the Endangerment Assessment orFeasibility Study reports. If so, this comment can be deleted.Modelling efforts have been conducted for soil remediation and arepresented as Appendix B in the Draft Feasibility Study submitted to theagencies on 15 March 1991. Ground water modeling, to evaluate the amountof time to achieve remedial goals, is discussed in detail in the response to thedraft Feasibility Study (FS) comments, specifically in response to GeneralComment 2 and in Attachment A of the response to the FS comments. Oncethe planned on-site remedial system is installed and operating, data will becollected to check its performance. After the system has been operating asdesigned for two to three years, sufficient data would be available to initiatereliable modelling to evaluate clean up times. This would provide a betterbasis to determine if additional remedial measures are warranted. Since

BR303357

Croup

Mr. Eugene DennisUSEPA10 April 1991PageS

the identified plume is in a stable configuration based on the routinequarterly ground water monitoring it would be best to test the effectiveness ofadditional on-site remediation before proceeding so that a firm basis forcomparison can be established.

3. The report should more fully describe the nature of the business that hasoccurred at the facility and provide a list of chemicals, particularlyorganics, that have been used in its operations. A more detailed descriptionof the chemicals used would present a more complete characterization of thesite.ERM believes that the "nature of the business" conducted at the TextronLycoming facility in the past was presented as thoroughly as possible basedon the information obtained from current and former plant employees. Theprocesses presently ongoing at the plant (metal finishing, platingoperations, engine assembly and testing) have remained essentially thesame since the operation started. The primary difference is the volume ofproduction, which has fluctuated dependent upon market conditions.Available records indicate that maximum capacity was approximately 1400engines per month during 1978 compared to the 1989 level of 475 engines permonth. A list is provided as Attachment A for those chemicals presentlyused at the facility, including annual volumes of each purchased. Little tono information is available for past plant usage, but the types of chemicals,for the most part, have remained the same. The use of some compoundssuch as GUNK™ and several varieties of paints have been discontinued atthe plant.During the RI field investigation, expanded analysis (full TCL/TAL) wasconducted on 20 of the Phase I ground water and 12 soil samples distributedthrough the study area to determine if other compounds were a concern atthe facility. Only trace amounts of chemicals other than the compounds ofconcern were detected, as discussed in the RI report. Review of the volatileand semivolatile organic tentatively identified compound (TIC) results forground water and soils yielded no unique parameters that would becharacterized as "signature compounds". Only the presence of nondescripthydrocarbons is discernable. It is the opinion of ERM that the type ofchemical compounds in the environmental media in the study area havebeen adequately characterized through the analytical work completed andthat no "unknown" compounds are present.

General (Pflgfi 2)1. There is some concern that the RI has not adequately evaluated the ground

water contamination with regard to separate phases, specifically Dense Non-Aqueous Phase Liquids (DNAPLs). Consideration should be given to thepositioning of the screen intervals in order to best detect DNAPLs. Most of/\R3033bo

Group


the screens have been placed in permeable units but perhaps not at the baseof such units. Care should be taken to ensure that the screened interval ispositioned at the base of the formation in order to detect DNAPLs. The RIshould provide an assessment of the impact of DNAPLs as they potentiallyreach bedrock or contaminate the bedrock via migration through fracturezones,The commentors noted that though the screened intervals of the monitoringwells in the study area are placed in permeable units, care should be takenthat the screens are placed at the base of these units. Careful considerationwas given to the placement of the well screens in the overburden to evaluatethe potential for dense non-aqueous phase liquid (DNAPL). The majority ofthe screened intervals for the overburden wells extend to the top of bedrock.No free phase liquids have been observed in wells on the Plant site or beyondthe Plant boundary. The bedrock wells in the study area were constructed tomonitor specific intervals along strike and down dip of MW-62 because itcontained the highest concentration of the compound of concern. Theselection of a monitoring interval is also controlled by the occurrence ofwater bearing zones.

TCE concentrations were observed to increase with depth in the well nestcontaining MW-8, MW-8D, and MW-62. This condition may be an indicatorof a DNAPL contaminant source. The wells with the highest concentrationsof TCE, MW-61, MW-62, and MW-63, were checked for DNAPL. AKemmerer™ sampler was used in these three wells to obtain water andsediment samples from the bottoms of the wells. No DNAPL was observed inany of these wells. DNAPL has not been observed in any of the wells on or offsite.

Assuming TCE was disposed of in the old dry well at the facility, it wouldhave migrated through the unsaturated zone and to the water table or,depending on the actual depth of the old dry well (which is unknown), TCEmay have been disposed of directly into the saturated zone. In eitherscenario, TCE could have migrated downward through the unsaturated andsaturated overburden.

A contiguous TCE mass would have been depleted during migration,leaving in its path a discontinuous residual of immobile DNAPL (see DenseChlorinated Solvents in Porous and Fractured Media Model Experiments,1988, by Friedrich Schwille). A contiguous TCE mass could have reachedthe overburden/bedrock interface and migrated horizontally along thatinterface, the direction of migration being controlled by the topography of thebedrock surface. Again, the migrating contiguous mass would have beendepleted during migration leaving an immobile residual in its path. The

AR303359

Group


TCE mass could have pooled in a depression in the bedrock surface, and/ormigrated into the weathered bedrock, along joints, fractures or beddingplanes.

Immobile residual in the saturated portion of the aquifer can act as a sourcewithin the aquifer for continued dissolved phase compound migration in theground water. This residual source can be localized in a discrete locationand/or along a select fracture close to the source, and may avoid detectionunless encountered by accident. Furthermore, this residual source is themost difficult to remediate to present clean-up goals because in most cases itcannot be located, and the technology does not presently exist to remediatesuch a source. Therefore, control of the situation is the best availabletechnology feasible.

The primary concern, if a DNAPL were present, would be to determinewhether it is a mobile residual and is thus potentially recoverable or posesfuture off-site risks; or an immobile residual, representing a continuingcontaminant source, but localized and containable. Investigations to dateprovide the following evidence:

• In approximately 56 monitoring wells and 60 soil borings (includingthose for the UST program) in the study area, no free phase DNAPLproduct has been observed;

• TCE concentrations in the percent of solubility range (5 to 10 percent)indicative of a DNAPL source, have not been observed. If free phaseTCE had migrated off-site during the last 40 years (assuming the plantused it since its introduction in the mid-1940s), observation ofconcentrations in the above referenced percent range would have beenexpected; and

• The TCE plume has remained stable based on routine monitoring withconcentrations decreasing by orders of magnitude in the off-sitebedrock and overburden compared to peak TCE concentrationsobserved on site.

These data indicate that if a residual source of TCE exists, it appears to berestricted to the on site area and feeds TCE to the ground water in a dissolvedphase. Thus, the evidence does not indicate that there is mobile DNAPL, andsuggests that if non-aqueous phase liquids are present, they are located on-site, most likely as an immobile residual.

Soecific (Pace 2)

AR303360

Group


1. Sections 3.1 - 3.4 and 3.7: The RI report does not cite references. Technicalwriting relies on reference material and it is appropriate to give credit tothose who originally compiled the information. In addition, where thereport does cite references (Sections 3.5, 3.5.3, 3.6, Figure 3-4), the referencesare not entirely informative and there is no section listing the citedreferences.References will be added or expanded for the appropriate paragraphs andthese sections will be resubmitted to the EPA. The references section for thedraft RI report was included at the end of Section 6. This is resubmitted as aseparate section (Attachment B).

2. Section 3.8: Regional Hydrogeology: Without providing figures- that presentthe shallow potentiometric surface and bedrock topography in this section,the discussion on page 3-15 about factors that control the hydraulic gradientand the flattening of the gradient toward the south seems somewhatpremature. Perhaps the more specific discussion on gradient and groundwater velocity (on page 3-16) should be omitted here in favor of some generalstatements regarding the increase in aquifer transmissivity toward thesouth.

Section 3.8.1 is resubmitted for EPA review as Attachment B. Also, arevised Table of Contents is included because the document (Section 3) had tobe repaginated.

3. Section 3.9: Pages 3-17 and 3-18: The RI provides the amount of total wateruse for subbasin 10, yet does not give specific amounts for the users ofWilliamsport. The Williamsport Municipal Water Authority (WMWA)obtains most of the water for public supply from surface water reservoirs,but the report does not indicate the amount required and supplied per year.In addition to surface water supplies, the water is pumped from the WMWAfield to supplement the reservoir capacity (especially during the summer).The report provides the early average from the well field, however, there isno indication as to what percentage, of the total, the well fields supply.These figures are important in order to determine the impact that thecontamination in the ground water may have on the overall public watersupply.

The WMWA well water supply is equipped with an air stripping treatmentsystem; furthermore the Third Street recovery well captures the leadingedge of the contaminant plume and prevents it from migrating to theWMWA field. These systems effectively minimize impact to the public watersupply when the well field is used.

Section 3.9 is resubmitted for EPA review as Attachment B.

The"Group

Mr. Eugene DennisUSEPA10 April 1991Page?

4. Section 4.6.1: There is no detailed discussion regarding the procedures usedduring the acquisition of samples. During the EPA oversight occurringduring November 7-9, 1989, the oversight team observed the samplerswithdrawing the first bailer full and discarding the sample (which isusually collected for volatile organic compounds (VOC analyses) as a "finaldecontamination rinse." The second bailer full was collected and submittedfor VOC analyses. Samples for VOC analyses should be collected in amanner to minimize aeration of the sample. The fact that the first bailerfull was discarded, and a sample not collected, reduces the probability ofdetecting possible dissolved VOCs. Samples to be analyzed for highly volatileorganics, and collected via bailer, should be carefully poured form the bailerinto the vials, avoiding turbulence, which might result in the loss of volatileorganics and lor excessive oxygenation of the samples.

Two concerns are noted with the above sampling procedures. The first isthe uncertainty of the decontamination procedures. If the bailer was notdecontaminated properly, contamination may be inadvertently introducedinto the well. The use of equipment rinse blanks should verify the efficiencyof the decontamination procedure.

The second concern is whether there was an accurate determination of theconcentration of VOCs through their sampling procedure as describedabove. The concentration of the volatile organics in the water column is afunction of the time lapse between the purging and actual sample collection.If sample acquisition immediately follows well purging, then the samplecollected, regardless of whether it is the first or second bailer, should berepresentative of the actual contaminant concentration. If formation wateris allowed to cascade down the well screen (in wells of low yield), excessiveaeration may occur resulting in the loss of VOCs. Therefore, the samplecollected from the second bailer may have a decreased concentration ofVOCs due to the aeration or dilution from incoming formation water.

Discarding of the first bailer is performed for the purpose of equilibratingthe equipment to the matrix being sampled. After decontamination of thebailer, there is a possibility that the bailer surfaces may be active to a limitedextent. The first bailer volume, which is discarded, serves to deactivate thebailer with the physical environment specific to that well being sampled. Inaddition, the cascading effect in a low yield well would take place to much agreater extent throughout well purging and in no different a fashion as thatduring sample procurement. Therefore, the discarded bailer volume wouldnot be expected to have any significant impact compared to the well purgingprocedure. Section 4.6 in the RI report details the sampling proceduresconducted during the RI investigation. In addition, the discarding of the

AR303362

Group


first bailer of water is standard procedure as indicated by ERM, Inc.'s"Standard Operating Procedures for The Collection of EnvironmentalSamples, March 1986, revised 1988."

Specific (Page 3>

1. Section 4.6.1.1; Page 4-22: Optimally, field parameters (i.e., pH,temperature, specific conductance) should be measured prior to, during, anfollowing well purging and prior to sample acquisition to establish stability.It is also appropriate to take these measurements prior to sampling andimmediately following sampling to ensure stability of the measurement,which indicates proper sampling conditions.

Ground water sample collection was conducted as per the EPA approvedQAPP (page 4-9 of appendix D in the RISOP). "Field measurements for pH,specific conductance, and temperature will be obtained on ground watersamples immediately following sample collection. A grab sample collectedin a beaker will be used to obtain these measurements."

2. Section 4.6.4; Page 4-25: Rather than collect ground water samples todetermine if separate phases exist, the investigation should consider usingan interface probe to detect the presence of immiscible layers. Thisprocedure should be undertaken before the well is evacuated for sampling.At this site, dense immiscible contaminants are the primary concern. Theinterface probe is lowered into the well, prior to purging, and shouldregister the presence of organic liquids, if present.

Collect the immiscible phase prior to purging. Then, lower the bailer slowlyuntil contact is made with the surface of the immiscible phase, and lower toa depth less than that of the immiscible / water interface depth as determinedwith the interphase probe.

ERM has used an optical interphase probe several times with no success atanother EPA Region III Superfund site with known DNAPL pools in severalwells. A depth-to-water or conductivity meter has proven to be more reliableat this site and therefore will be incorporated in the April 1991 ground watersampling event at the Textron Lycoming facility. These meters use thedifferences in the conductivities of water and DNAPL to detect the depth andthickness of any DNAPL. It is noted that no evidence of DNAPL or NAPLhas been observed to date in the study area when an electronic depth-to-water meter was used to obtain water level information or to sound welldepths.

Specific (Page 4>AR303363

Group


1. Section 5.0: Field Investigation Results: Please be consistent in labelingRW-1, RW-2, and RW-3 among Figures 5-4 and 5-9 and Plate 1.

The recovery well designations on Plate 1 have been changed to the correctdesignations, and are included in the pocket at the end of this responsepackage. The designations on Figures 5-4 and 5-9 are correct.

2. Section 5.1.2.1: Past Plant Operations: This section provides no informationregarding when and how much trichloroethylene was used in the past otherthen the fact that it was the usual solvent for degreasing engine parts.Presumably historical records pertaining to solvent purchase areunavailable, but was any attempt made to estimate an annual quantity ofsolvent used over some period of time?

Present solvent use volumes are presented with the chemical lists requestedfrom the 3rd paragraph on page 1 of the agencies comments. As previouslydiscussed, it is not known how much solvent was used in the past, due to thesignificant fluctuations in production volume and differences in recordkeeping requirements. Therefore, no attempt to estimate past solvent use atthe plant was made during the RI investigation. However, a mass balanceequation using ground water data has been completed. The results(attached) indicate that approximately 3,000 pounds of TCE are in theground water and potentially as much as 30,000 pounds is adhered to thesoil.

3. Section 5.3.2.3; Page 5-17: The report presents analytical data for soil. Thestatement is made that lead and chromium were detected abovebackground, yet the report does not provide either background levels or adescription of how background levels were determined.

The report and Table 5-6 state that lead was detected at 185 mg/kg in soilboring SB-18NB and 169 mgfkg in SB-20. However, the analysis reports, inAppendix D, for these samples indicate the concentrations are 18.5 mg/kgand 16.9 mg/kg, respectively. An explanation, such as dilution factorsapplied to the analytical data, should be provided. Otherwise, the textshould be revised to reflect the actual analytical data. In addition, the textand Table 5-6 provide analytical data for chromium from samples SB-4 andSB-6. However, the analysis reports, in Appendix D, for these samplesindicate that the concentrations are not exactly as reported in the text. The0.1 mgfkg values, as reported in the text, are not indicated on the analysisreport.

Based on these two comments, it appears that there are additional data (rawdata) which may have been reviewed and interpreted for the text discussion.

AR303361*Th«

Group


However, these data were not presented in the RI report for review. On page2-4 of Appendix D, there is no discussion regarding qualifiers for thesebefore mentioned samples, therefore, the reviewers assumed that theresults reported in the text and tables may be transcription errors. The RIshould reevaluate these data, and any other data for accuracy to ensurecorrect values are presented in the text and tables.

Background concentrations were based on data obtained from the tableentitled Typical Heavy Metal Concentrations in Eastern U.S. Soils from thereference:

Connor, Jon J., and Hansford T. Shacklette; Background Geochemistry ofSome Rocks, Soils, Plants, and Vegetables in the Conterminous UnitedStates; United States Geological Survey, Professional Paper 574-F; 1975.

As stated on this table, average soil concentrations for chromium and leadare 34 ppm and 14 ppm, respectively. These values were used loosely asbackground concentrations for these elements. It is noted that the text inthe RI Report primarily discusses only those samples showing elevatedconcentrations above "background".

In addition, New Jersey Department of Environmental Protection (NJDEP)unofficial guidelines for cleanup of contaminated soil were also consideredin evaluating the soil data in the RI. The unofficial cleanup guideline forboth chromium and lead is 100 ppm. NJDEP information was used becauseit is the only available guideline for reference. In completing the draft RAand for evaluating soil clean-up levels the EPA 1989a Final InterimGuidance on Estimated Soil Level Clean-up Levels at Superfund Sites,OSWER 9285.701A which included a range of 200 - 500 ppm as a target levelwas referenced.

The QAR data table values for lead in soil samples SB-18B and SB-20 areincorrect. The values are actually one order of magnitude lower, andshould be 18.5 and 16.9 ppm, respectively, both of which are below the targetlevels referenced above. The raw data has been reviewed and the lead valuesin both of these samples were recalculated. The CLP Form Is in thedeliverables package were incorrect and will be resubmitted by thelaboratory. See the comment below for more information.

The slight differences in results and quantitative limits is the result of theinclusion of the commercial lab results sheets in Appendix D rather thanthe data package Form Is. The commercial lab results sheets are printed todifferent specifications and criteria than the Form Is in the deliverables.

AR303365


Therefore, since the QAR data (which references the Form Is) was used inthe text and data tables, the RI information is accurate.

No additional qualifiers were needed and there is no outstanding data. Seeresponses above.

QAR data Table 5-6 is resubmitted for EPA review as Attachment C.

4. Section 5.3.2.4, Page 5-18: The report states that Arochlor-1254 was found atwhat seems to have been an isolated occurrence.

The report should investigate and provide an explanation for it-s presence.

The isolated occurrence of Arochlor-1254 in soil boring SB-14B cannot beeasily explained. It is possible that cutting oils, which in the past could havecontained PCBs, could have been disposed of in the old dry well. Howeverthis is speculation, because the disposal of cutting oils in this well has neverbeen reported, based on the plant personnel interviews. There are notransformers in this area. The occurrence of PCBs is believed to be anisolated occurrence based on the RI data and recent data obtained from theunderground storage tank investigation being conducted at the plant incooperation with the PADER. Soil samples collected for the USTinvestigation were also analyzed for PCBs. No concentrations were detectedabove analytical method detection limits. A copy of the UndergroundStorage Tank report will be submitted to both the PADER and the EPA whencompleted.

Specifics (Page 5)

1. Section 5.3.5: Comparison of Laboratory VOC Analysis to Petrex Soil GasResults

It is presumed that the last paragraph of this section means that all VOCsource areas beneath the facility have been identified.

Based on the data presented in the RI Report, the EPA statement is correct.

2. Section 5.4:

1) The discussion pertaining to overburden thickness on page 5-27 wouldbe clearer if a figure that presented overburden isopachs wasincluded.

AR303366

Group


An overburden isopach map of the study area for the report isincluded as Attachment D. Also, page 5-27 from the draft RA isincluded to show reference to this figure.

2) Please include the five recovery wells on the cross-sections presentedin Plates 3, 4, and 5, as appropriate. Please show the locations ofpotential areas of concern 16 (former coolant well) and 17 (former "drywell" sump) on these cross-sections noting suspected depths. Pleasecheck cross-section C-C' at MW-72; it does not appear as anoverburden well.

Plates 3, 4, and 5 have been modified as requested and are included inthe pocket at the end of this document. In addition, MW-72 on cross-section C-C' will be corrected. Plate 4 did not require any changes.

3. Section 5.5.1.4: Recovery Test - General ObservationsThe unidentified recharge boundary that apparently exists 500 to 1000 feeteast ofRW-2 is troublesome. Regional geologic information should beincorporated to support the apparent increase in the thickness of theoverburden east of the site.

It is not apparent as to why the "unidentified recharge boundary" istroublesome to EPA. The apparent recharge boundary is an observationfrom the recovery test data, as discussed in the RI. ERM felt responsible topresent all observations from the recovery test and a hydraulic boundaryaffect appears in most of the semilog plots. Since the ground water flowdirections in the overburden aquifer are sufficiently understood in the studyarea, ERM does not see any reason for concern. There is no known"regional geologic data" to support the thickening of sediments east of RW-2;however, site specific data is available (see well log for MW-12) whichidentifies the overburden thickness in this direction.

4. Section 5.5.2 Ground Water FlowSection 5.6:

1) The discussion regarding ground water flow and contaminantmigration in the vicinity of Elm Park is not completely clear.Although an upward hydraulic gradient may be indicated by waterlevel elevations at MW-60 and MW-32, little, if any, of the overburdenappears saturateed near the "bedrock ridge" at Lycoming Creek,which means the former statement is irrelevant. Note that waterlevel elevations in MW-14B and MW-72 indicate a downward gradient;presumably this reflects the impact of pumping at the Elm Parkrecovery well. Although operation of this recovery well might appearto affect Lycoming Creek here, it is not clear that contaminants would

AR303367The

Group


also discharge upward given their apparent propensity to movesouthwest along strike in the bedrock. Thus, the data availableregarding TCE-DCE concentrations in bedrock wells provide littleassurance that contaminants have not migrated beneath LycomingCreek at Elm Park.MW-60 and MW-32 are located along the northwestern edge of thebedrock high in an area of thicker saturated overburden sediments.In addition, there are approximately 4 to 5 feet of saturated sedimentson the ridge of the bedrock high, as demonstrated by water levels inMW-72. Since there are saturated sediments, the upward gradient atElm Park discussed is an accurate observation of the hydrogeologicconditions at this location and indicates discharge of contaminatedground water from bedrock into the saturated overburden deposits isapparently occurring in the area.

MW-14B and MW-72 are located far enough away from each other sothat comparing water levels in these wells is not appropriate withoutmaking a correction for the change in head due to the horizontalhydraulic gradient. This correction was made by ERM for thesecomments and a downward gradient is evident in this area of ElmPark. This was expected as a result of the pumping of the Elm ParkRecovery Well and the proximity of MW-14B to this recovery well.

ERM did not state or mean to imply in the RI that all of the groundwater in the bedrock at Elm Park discharges to the overburdensediments. However, it is apparent form vertical gradient data andprimarily overburden concentration data that this area is a dischargearea for significant amounts of ground water from the bedrockoverburden. Contaminant concentrations from Phase II overburdenwell MW-72 and levels of 1000 ppb to 2800 ppb in flood well located inthis area (as obtained by Chester) are apparently from the shallowElm Park bedrock. Additional evidence of this is apparent in the factthat contaminant concentrations for the Elm Park bedrock decreasesignificantly with depth. The deeper bedrock wells in the Elm Parkarea (MW-60, MW-57 and MW-59) show only trace to non-detectablelevels of the contaminants of concern. This indicates "thatcontaminants have not appreciably migrated to the deeper bedrock inthe Elm Park area", as would be expected if there was a downwardgradient in this area (Reference Cross Section C).

ERM did not state or mean to imply in the RI that all of thecontaminants or ground water in the bedrock at Elm Park dischargeto the overburden sediments. However, vertical gradient data andconcentration data (flood relief well concentration data collected byChester and MW-72) indicate that this area is a discharge area fromthe bedrock to the overburden. «r>nn/-»oroAR303Job

They-~r~~<~./-i

Wjt!wGroup


The RI Report does state that the deeper bedrock wells in the ElmPark area (MW-60, MW-57 and MW-59) show only trace to nondetectable levels of the contaminants of concern. This would indicate

. "that contaminants have not appreciably migrated to the deeperbedrock in the Elm Park area" (Reference Cross Section C).

2) Given the increase in TCE-DCE concentrations with depth at MW-8,MW-8D, and MW-62, the existence of a DNAPL at a depth greaterthan 115 feet in this area cannot be discounted. Thus, the verticalextent of contamination has not been characterized here.

It is agreed that the vertical extent of contamination in the area ofwell nest MW-62 has not been fully characterized and that theexistence of DNAPL cannot be fully discounted. As previouslydiscussed the data suggest a residual source of TCE may be present inthe saturated zone on site. It is not apparent as to what advantagesfurther investigation into the bedrock would present, since theexistence of DNAPL can never be fully discounted. Based on the factsdisclosed from the RI investigations, the following can be concluded:

• Due to the likely scenario that spent TCE was disposed of in theold dry well, it is probable that there is a residual source of TCEon-site. The conclusion that this residual is immobile andrestricted to the on-site area is supported by the fact that theconcentration of TCE decreases by orders of magnitude off-sitein the bedrock and in most cases, the overburden.

• The TCE plume configuration is stable based on the routineground water quality monitoring being conducted whichindicates there is no mobile DNAPL causing an off-site spreadofTCE.

• The concentrations of TCE in MW-62 are approximately 1-2% ofthe solubility of TCE, which speculatively could be indicative ofthe presence of a DNAPL. However, the RI and prior Chesterinvestigations have shown no free-phase product in any of theoverburden or bedrock wells, either on-site or off-site. Wellsinstalled along the apparent direction of migration pathway inthe bedrock, along strike and downdip and hydraulicallydowngradient, have non-detect to low concentrations of theorganic compounds.

• Based on the evidence form the RI, residual TCE may bepresent on site and the proposed pumping scenario for the

AR303369

Group


Textron Lycoming facility should eliminate concern, since it isdesigned to contain the migration of both overburden andbedrock ground water from beneath the Plant.

The benefit of drilling additional monitoring wells in thealready identified center of the bedrock contaminant plume isquestioned, since it has already been disclosed that this area iscontaminated and the perimeter bedrock wells exhibit relativelylow to non-detect concentrations.

Bedrock wells located along the strike of the Tully formationand downgradient from MW-62 show decreasingconcentrations. The wells (MW-57, MW-59, MW-60, and MW-65) are all screened at elevations that would be comparable toMW-62. In Elm Park only MW-57 shows trace levels (non-detectto 6.5 ppb), all others are non-detectable concentrations of thecontaminants of concern. In general, water quality improveswith depth in Elm Park. Concentrations in the perimeterbedrock wells located downdip in the Tully formation andsoutheast (the direction of the hydraulic gradient) of theTextron facility are also low to non-detect. Ground waterquality in the surrounding bedrock aquifer being monitored,will not change if higher concentrations were to be encounteredbelow MW-62.

SDecific (t>affe 6)

1. Section 5.6.1.3; Page 5-60: The detection of vinyl beneath the center of thepant property is an indication of the degradation of TCE. However, onewould expect to also find vinyl chloride and further degradation products ofTCE in the ground water from wells at the perimeter of the contaminationplume, since that contamination may have been in the ground water longerand has been able to degrade longer. Through further data interpretation oranalysis, the RI should determine the presence or absence of vinyl chloridein the ground water from wells at the perimeter of the plume.

A detailed description of trichloroethylene biodegradation is presented inSection 5.3.2 in the draft Risk Assessment for the Textron Lycoming Facilityand is resubmitted as Attachment E. This discussion presents a possibleexplanation for the absence of vinyl chloride in the off-site monitoring wellsand is summarized as follows:

TCE/DCE concentrations have been detected in the study area, but vinylchloride concentrations have been detected primarily in wells at the plant.

RR303370


The principal biodegradation pathway beneath the plant appears to bereductive dechlorination, possibly caused by anaerobic conditions enhancedby the plant paving (approximately 85% is paved or covered with building).Different biodegradation pathways are suspected for the non-plant areassince vinyl chloride has not been detected in these wells. These pathwayscould reduce TCE to a less toxic compound and not in the pathway to vinylchloride. One hypothesis for this finding is that microbial populations atthe plant and in the off-site overburden system are different and degradeTCE/DCE differently.

In addition, while vinyl chloride is the second stage decomposition productof TCE after a DCE intermediate, it is not necessarily the end point of theTCE decomposition pathway. Ethene has been documented (Vogel, Griddle,and McCarty: ES&T 21(8) 1987) to be the end point of the abiotic degradationpathway. There is no reason that the pathway must stop at vinyl chloride.

ERM, through data interpretation and analysis, has determined that vinylchloride concentrations are below analytical method detection limits in thewells at the perimeter of the plume. This has been confirmed repeatedlythrough the PADER quarterly ground water sampling events that analyzefor vinyl chloride (EPA method 601) in all the wells sampled each quarter.These include many wells located on the perimeter of the contaminantplume. The analytical detection limits for most of the vinyl chloride groundwater samples is less than 1 ppb.

2. Section 5.6.1.12; Page 5-68: The report states that the reason for the higherpH results from MW-64 and MW-66 is not clear. The investigators shouldclarify the reason or abandon these wells and install two other wells in thatarea.

ERM restates that the reason for the high pHs in these wells is not clear.As an added measure, ERM will redevelop each of the wells with high pHsand re-evaluate the integrity of these wells. However, pH is'not a factor thatwill affect the concentration of the contaminant of concern in the study area,or the ability to detect these compounds.

3. Section 5.7; Page 5-74: The first paragraph on page 5-74 states that "TCEand their compounds were disposed of in this (dry) well in the past." The RIshould provide a list of these "other compounds." This information isneeded in order to provide a complete characterization of the site.

As stated in response to the General Comment (page 1) 3rd paragraph andthe response to comment page 6-6 an effort was made to evaluate thepresence/absence of "other compounds" by completing laboratory analysisory analysis

AK30337 IThe

Group


for nonvolatile compounds in the environmental media in the study area.The findings indicate that there are no unknown compounds of concern.

4. Figures 5-14, 5-16, 5-27, 5-28: Although the deep ground water may beflowing west toward Lycoming Creek, in the zone that is being monitored bybedrock monitoring wells the water levels measured indicated that the flowis to the southeast. The arrows indicating flow direction should be changedso that they are perpendicular to the ground water contour lines.

These Figures all pertain to the potentiometric surface in the bedrockaquifer. As disclosed from the results of the recovery test, the bedrockaquifer is anisotropic. Under anisotropic conditions, the direction of groundwater flow is not necessarily perpendicular to "the potentiometric surface, asexpected in isotropic flow systems. The direction of ground water flow iscorrect as depicted by the flow arrows on the aforementioned figures.Placing the arrows perpendicular to the potentiometric surface would betechnically incorrect since that is the direction of the hydraulic gradient.

5. Figure 5-32: TCE plume configuration of the overburden aquifer, the hotspot around MW-3R should be connected with the rest of the plume and a1000 contour around the 13-- ppb value at M-18 needs to be drawn.

A 1000 ppb contour will be drawn around MW-18 and the figure resubmittedto the EPA. The hot spot around MW-3R could be connected with the rest ofthe plume, however this area is considered a separate potential source areafrom the center of the plant. This interpretation is based on theconcentrations detected in MW-11 and MW-4 over time and the bedrock highlocated in this area of the plant. The bedrock high affects overburdenthickness/aquifer thickness in this area and acts as a divide, locallyseparating ground water flow and contaminant movement in this area.Figure 5-32 is resubmitted as Attachment F.

6. Section 5.7.2.8: Effectiveness of Remedial Program

In various sections of this report, the effectiveness of the recovery wells ispresented in terms of the pounds of TCE and DCE removed to date. Pleaseuse the data available to estimate the total mass of contaminants that may bepresent in the saturated zone, both in the liquid phase and sorbed to the solidphase.

An estimate of the total mass of contaminants present in the saturated zoneis provided as an attachment.

Section 6AR303372

The

Group


1. Section 6, Page 6-2: The report states that "overall VOC concentration in thesoil samples were low." The report should quantify/define the term "low."

The term "low" in this paragraph refers to less than 100 ug/kg total VOCconcentration. The USEPA CLP SOW definition for low level VOCs is lessthan 1000 ug/Kg (USEPA Contract Laboratory Program Statement of Work(SOW) for Organics Analyses Multi Media Multi Concentration, 7-89Revision, Page VGA D-ll).

2. Section 6, Page 6-6: The report states that semi-volatile compounds werefound in the outfall of the Oliver Street storm sewer at concentrations abovebackground, but that semi-volatile compounds are not associated with theplant. This has not been substantiated through a complete sitecharacterization. See comments #3 and #15. The report should provide alist of all chemicals associated with the plant to demonstrate that their areor never have been semi-volatiles associated with the plant.

Though limited amounts of semi-volatile compounds are used at the plant,it is noted that the ground water and soil samples have shown little to noevidence of semi-volatile compounds. Semi-volatiles are not consideredcontaminants of concern based on the TCL volatile, semivolatile,PCB/pesticide, cyanide, and TAL metals analysis conducted on a percentageof the RI ground water and soil samples. In addition, semi-volatilecompounds are not part of the waste treatment plant processes anddischarge.

Specific (Page 7)

1. Appendix D: Remedial Investigation Quality Assurance / Quality ControlReview Report: Further define the "B" qualifier, or qualify data having "B"values using the "Laboratory Data Validation Functional Guidelines forEvaluating Organic Analyses: (USEPA Feb. 1, 1988).

The "The June 1988 Laboratory Data Validation Functional Guidelines forEvaluating Organic Analyses" is referenced in the last sentence of Section 1of the QAR. Also, the "B" qualifier is defined at the bottom of each datasummary table and is thoroughly explained in each QAR bullet dealing withblank contamination.

Biological Assessment Comments

1. The habitat of the creek draining the area appears to be minimally affectedby the site. Lycoming Creek, ultimately receiving the drainage form the

AR303373

Group


site, has a fairly stable ecosystem, and supports a variety of invertebrates aswell as warm and cold water fisheries.

ERM agrees with the EPA comment.

2. Ground water appears to flow to the Susquehanna River which is about amile south of the site; however, no information could be found regardingground water discharge investigation. It is suggested that the dischargepoints be located and tested for any contamination that might be entering theSusquehanna. This concern revolves around the ultimate fate of thecompound 1,1-dichloroethylene as well as the several other metalcontaminants associated with the site.

Ground water flow from the overburden beneath the Textron LycomingFacility should be captured by the Third Street Recovery Well based on theoverburden aquifer ground water flow volumes calculated for the studyarea. Since this overburden flow would be captured by the Third StreetRecovery well, there would be no discharge to the Susquehanna River.

It is noted that prior to the start up of the Third Street Recovery Well,contamination may have reached south of Third Street. Levels ofchlorinated compounds presently detected in the WMWA well field wouldreflect the maximum concentrations which could have migrated furthersouth. These concentrations are low (range from below method detectionlevels to a high for TCE of 1.2 ppb at WMWA well #10 on August 27,1990which is below the 5 ppb MCL for this compound). Given the dilution whichoccurs when the overburden aquifer discharges to the river, theconcentrations in the river would be below detection levels, if discharge tothe river were to occur. The bedrock aquifer in the vicinity of Third Streetdoes not contain contaminants of concern above analytical method detectionlevels.

3. More information should be supplied concerning the impact of groundwater and soil contamination on toxicity. Environmental risk assessmentrelated to the contaminants vinyl chloride, 1,1-dichloroethylene, andhexavalent chromium should be addressed.

This comment has been addressed in the draft Risk Assessment.

Drinking Water Comments

1. Section 5 (pg 74): The contractor mentions the presence of a coolant well, butdoes not provide any discussion concerning the cooling process. Did theplant operate an open or closed cooling process? Was the coolant MMdLwateL

Th«

OP^ Group


analyzed for constituents normally used in cooling system (i.e. biocides,chromates, quats, or ethylene glycol)?

The coolant well, located on the Textron Lycoming facility, was sealed in the1950s and the exact location of this well is presently unknown. It is knownthat there is at least 4 feet of concrete above this well, since a new loadingarea was built in this vicinity. The material under the concrete and abovethe well is unknown. How the well was abandoned and sealed is alsounknown. Therefore, the well has not been and cannot be sampled for anyof the compounds mentioned in this comment.

The cooling system at the plant at the time this well was in operation isbelieved to have been an air-water heat exchanger in which air flows aroundcoils containing water, which was not re-circulated.

2. Tables 5-18, 5-19: Confusion exist over the variation in the detection limitsused for vinyl chloride. Why does the laboratory use levels that range form0.5 ppb to 500 ppb? It is assumed that the laboratory is using the sameanalytical method; therefore, there should be some consistency in theanalytical minimum detection limit.

The regulatory analytical protocols used in this investigation requiredilution of the samples where high concentrations of contaminants areevident. These dilutions drive the final detection limits for all non-detectedcompounds in the diluted analyses. In the present case, detection limitswere mainly dictated by the relatively high TCE levels.

Additional PADER Comments on Draft RI and RA (received 11 March 1991

1. The sample results presented in both the RI (Remedial Investigation) andthe RA (Risk Assessment) for several wells, most notably MW-3 and MW-18,show hexavalent chromium at higher levels than total chromium. This ispresented on page 5-62 of the RI along, with being in the data tables in bothdocuments. Does Textron Lycoming or their consultant, ERM, have anexplanation for this anomaly?

ERM is aware that hexavalent chromium, in most cases, was present inconcentrations greater than total chromium. These two compounds areanalyzed by two different analytical methods (Method USEPA CLPStatement of Work for Inorganic Analyses, 7-88 Revision and StandardMethods for the Examination of Water and Wastewater, Method 312.B 16thEdition, 1985) and the results for the two different compounds in each wellwere below the permissible EPA error range for laboratory analysis.

fiR303375

'Group


Because of the different analytical methods, the results for the twocompounds are not unexpected.

2. Concern also exists regarding the levels of chromium detected in MW-3 andMW-18 which exceed the levels for a characteristic hazardous waste. TheRI report places very little emphasis on the chromium results and seems tofocus only on the organic contamination. While there may be negligible riskposed by these levels due to the lack of ground water users in the immediatearea, what evidence exists demonstrating that chromium will noteventually migrate to the Williamsport Municipal Water Authority wellfield? The presence of hexavalent chromium at 102 ppb at MW-31, locatedsouth (downgradient) of the old wastewater treatment plant would seem toindicate that off-site migration of the chromium is occurring.Isoconcentration maps depicting chromium levels should have beenincluded in the RI report. The FS report should include remedialalternatives which address the chromium as well as the organiccontaminants. The Pennsylvania ARAR for ground water for hazardoussubstances is that all ground water must be remediated to "background:quality as specified by 25 Pa. Code §§264.97(i), (j) and 264.100(a) (9). TheCommonwealth of Pennsylvania also maintains that the requirement toremediate to background is found in other legal authorities.

The hexavalent chromium in the ground water beneath the western portionof the plant facility was not de-emphasized in any way in the RI Report. Theprimary components in the ground water are TCE, DCE and vinyl chloridewith a localized area of hexavalent chromium beneath the western section ofthe plant. It was stated in the RI report that migration of hexavalentchromium beyond the plant boundary is occurring.

An isoconcentration map will be generated in an effort to present thehexavalent chromium contaminant plume. The map is submitted asAttachment G.

Considerable effort has been placed in the Feasibility Study concerning theremediation and treatment of the hexavalent chromium portion of thecontaminant plume. In addition, hexavalent and total chromium sampleswere collected during the January 1991 ground water sampling event assupplemental information for the Feasibility Study. This data is presentedin the Feasibility Study and will also be presented in the First Quarter 1991Remedial Progress Report which is submitted to both the PADER and theEPA in April.

ERM does not agree that EPA should consider the Pennsylvania groundwater "background standard" specified in 25 Pa. Code §§264.90-264.100 to be

AR303376The

Group


an ARAR. In order for a state standard to be an ARAR it must be either"applicable or relevant and appropriate." 40 C.F.R. §300.4. ThePennsylvania background standard is neither "applicable" nor "relevantand appropriate."

In order for a standard to be "applicable" it must "address a hazardoussubstance, pollutant, contaminant remedial action, location, or othercircumstance found at a CERCLA site. 40 C.F.R. §300.5. The Pennsylvaniabackground standard applies only to ground water remediation at RCRAsites, not CERCLA sites, and is therefore not applicable.

A standard that is "relevant and appropriate" must "address problems orsituations sufficiently similar to those encountered at the CERCLA site thattheir use is well suited to the particular site." 40 C. F. R. §300.5. ThePennsylvania background RCRA standard is intended as a prospectivedetection and remediation standard and it is inappropriate to apply such astandard to the situation at the Textron Lycoming facility where long termhistorical contamination caused a ground water plume.

Textron and ERM appreciate EPA's and PADER's review of the Draft RI Report.If there are any additional questions or comments regarding these responses, donot hesitate to contact me at (215) 524-3539.

Sincerely,



RTW/pawAttachmentcc: P. Boob, Textron Lycoming

P. Duff, Textron LycomingL. Trefsger, Textron LycomingT. Kraig,Textron LycomingL. Hoose, ERMR. Baker, ERMT. Schuller, ERMU. Tyagi, ERMS.Sayko,ERM

TheopGroup

Environmental Resources Management, inc.855 Springdale Drive • Exton, Pennsylvania 19341 • (215) 524-3500 • Telex 4900009249

10 April 1991

Mr. Eugene Dennis FILE: 723-01-03-01Regional Project ManagerUS Environmental Protection AgencyRegion III841 Chestnut BuildingPhiladelphia, PA 19107

Re: Response to USEPA and PADER Comments: 31 January 1991 Draft RiskAssessment Report for the Textron Lycoming Williamsport Facility

Dear Mr. Dennis:

On behalf of Textron Lycoming, Environmental Resources Management, Inc. (ERM) issubmitting a response to the United States Environmental Protection Agency (EPA)and the Pennsylvania Department of Environmental Resources (PADER) commentsconcerning the Draft Risk Assessment Report for the Textron Lycoming WilliamsportFacility. The comments received by Textron Lycoming on 5 March 1991 are presented(italicized) as received; responses to each comment follow.General (Page 1)

1. It is the policy of the EPA to present a conservative estimate of site risk asrepresented by reasonable maximum exposure (RME) scenarios. This approachmay be explained as the highest exposure that is reasonably expected to occur ata site. It includes consideration of both exposure parameters and exposure pointconcentrations for their relevance as reasonable maximums. The present reporthas presented a risk range rather than calculating the RME. In this report, therisk numbers reported from some scenarios may exceed the risk calculatedaccording to EPA guidelines, resulting in an overly conservative andunreasonable assessment of site risk This is the result of several philosophicalapproaches to the data which are inconsistent with EPA policy.

The Preliminary EA (ERM, 1988) and the Draft RA (ERM, 1991) had calculatedthe same order of magnitude risks from ground water exposures as stated in theExecutive Summary (p. ES-2) and Section 1 (p. 1-13). The Preliminary EA (ERM,1988) was prepared under acceptable RA guidance/1) The Draft RA (ERM, 1991)

USEPA 1986 Superfund Public Health Evaluation Manual (Directive 9285.4-1) and USEPA1986a Draft Superfund Exposure Assessment Manual (Directive 9285.5-1). •nonr>o'7oAnouoo /o

The



was prepared in accordance with the interim final RA guidance 2), except that areasonable and worst-case risk range was used rather than the RME. The RMEscenario was believed to lie within the range calculated in the Draft RA. ERMrecalculated Scenario C to determine the impact to the risk and hazard fromusing the RME calculations. Based on the attached calculations, 1thecarcinogenic risk for the RME scenario equaled the reasonable risk scenario.Similarly, the hazard for RME was almost identical to the reasonable hazardindex. Therefore, ERM requests approval to leave the calculations as a range.Please also note that on other RAs submitted to and approved by Region HI,(William Dicks Lagoons, May 1990 and Cokers Landfill, November 1989)alternative approaches have been accepted.The total vs. unfiltered metals issue specifies in the RA guidance document(December 1989) that total unfiltered sample data should be used to calculaterisks for individual wells. The use of total unfiltered metals represented aconservative, worst-case approach to the risk calculations. To the best of ourknowledge from previous approved RAs (William Dick Lagoons and CokersLandfill),USEPA Region in has required full detection limits be used incalculation of averages for metals. All of these recommendations could beincorporated into the document, but the conclusions to the RA and the mediarequiring remediation would remain the same; therefore, ERM requestsapproval to leave the calculations as reported in the RA.

General

1. It should also be noted that there are several inconsistencies between the datapresented in the RI and that presented in the RA. It appears that there havebeen some transcriptional errors. These should be corrected. Similarly thedefinitions ofET, EF and ED on page 5-16 do not coincide with the calculationspresented on page E-10.

The ET, EF, and ED values on pages 5-16 and E-10 were given for differentscenarios. These values on page 5-16 reference VOC emissions from the airstrippers which were modeled to the nearest population, while the values onpage E-10 reference the inhalation of volatiles from soil at source withoutmodeling. Therefore, the exposure values were different.

The USEPA December 1989 interim final reference for Risk Assessment Guidance forSuperfund, Volume 1, Human Health Evaluation Manual (HHEM) (EPA 540/1-89-002).

AR3 ___Group


USEPA Comments

Geneyaj Cpipments

1. The RA references a version of Risk Assessment Guidance for Superfund.Volume 1. Human Health Evaluation Manual published September 29, 1989.The RA must draw upon and reference the December, 1989 version of thatdocument. Although the differences between the two versions are minimal andare largely confined to clarification of the text and not methods of performingrisk assessments, it must be clear that the most recent guidance documentswere employed.

The interim final reference for Risk Assessment Guidance for Superfund,Volume 1, Human Health Evaluation Manual (HHEM) should be the December1989 document (EPA 540/1-89-002). This manual was used in the preparation ofthe RA.

2. Risks were calculated using the arithmetic mean of the constituentconcentrations and calculated again using the maximum concentrations.According to Risk Assessment Guidance for Superfund. Volume L HumanHealth Evaluation Manual, the 95% confidence interval of the arithmetic meanshould be used with other considerations (described on page 6-19 of the above-mentioned guidance document) to quantify the reasonable maximum exposure(RME). The RME should be used in quantifying risks. Any deviations from theapproach outlined in the guidance document should have EPA approval.

ERM prepared this RA in accordance with the existing CAO for TextronLycoming and according to EPA guidance documents for preparation of riskassessments for human and environmental populations (EPA, 1989a, 1989b).Deviations from the RME were made for the following reason:

• EPA's guidance document (EPA 1989a) states that the upper 95 percentconfidence level of the arithmetic mean be used to calculate the potentialrisks because of the uncertainty associated with any estimate of exposureconcentrations. Instead, an arithmetic mean and maximumconcentration of compound detected in a medium were used to calculate arange of potential risks or hazards. This modification was based on theknowledge that ground water risks calculated in the preliminary riskassessment (ERM, 1988) exceeded EPA's recommended carcinogenic riskrange. Also, ground water data collected after that submission has notshown concentrations significantly different from the previous levels.

ERM would appreciate USEPA's approval for this deviation based on the factspresented and the use of similar conservative approach on previous RAsubmittals made to the agency.

AR303380The

Group


3. Page ES-7: The report states that "it is unlikely that a population has or willinstall a well on site, in Elm or Memorial Park, or on Third Street since publicwater is readily available through the Williamsport Municipal Water Authority(WMWA) and the review of available well records did not indicate a privateuser". This is contradictory to Figure 2-3 of the RA and Table 4-1 of the RI where11 of the 35 wells identified within an approximate three-mile radius of theTextron Lycoming facility were identified as domestic water supply wells. TheRA must provide justification for not considering these wells as potentialreceptors. Ground water sampling analytical data from these wells may benecessary to determine if these wells have been impacted by the contaminantplume.

As depicted in Figure 2-3 and Section 2.3.8, domestic wells are about two to threemiles north and northwest of study area, and south across Susquehanna River.None of these wells are within the defined plume boundaries, nor during the RIwere any migration pathways defined toward these wells. To Textron's/ERM'sknowledge there are no private wells within the defined plume area which isconfined to the north by High Street, to the west within the area of MemorialPark, to the south by Third Street and to the east by Cemetery Street.

4. Section 2.3.8, Pages 2-10 and 2-11: The report states that the well fields of theWMWA supply five to eight percent of the water supply for Williamsport. Acomment provided in the RI review, requested the percentage of the total potablewater usage the WMWA obtains from the well field. The RA appears to providethis information. However, the report should clarify if this percentage is theyearly average. The concern is that the percentage may be higher during thesummer months or during periods of drought. The report should presentseasonal percentages.

In response to comment 4, Table 2-2 has been revised and is attached. This tablenow includes a breakdown of the water usage from the two reservoirs and theWMWA production well field during.months that the well field was used.Additionally, the rainfall for the month is given. This information was suppliedby Mr. Walt Nicholson of WMWA. As evident in the table, a pattern of well fieldusage, e.g. seasonal or drought event related is not clearly evident. However, thedata do show that prior to late 1989 the well field is used more frequently duringlow precipitation periods which coincide with the need to maintain water levelsin the surface water reservoir supply. Beginning in late 1989 and into 1990, thewell field was also used during the time that the reservoir was afflicted with aGiardii bacterial problem.

5. Section 2.5.1, Page 2-14: The RA report identifies five primary area where pastactivity could have impacted the area as well as multiple smaller point sources

Group


throughout the facility. The RI, Table 5-1 identifies approximately 25 "potentialareas of concern." On page 5-73 of the RI, the report states that there are fivesuspected areas which contribute to the contamination in the overburdenaquifer. These same five areas listed in the RA are described as the "primaryareas where past activity could have impacted the environment." The RA goeson to say "In addition, there were likely multiple smaller point sourcesthroughout the facility." Rekj-ses to the overburden aquifer should not beconsidered as the only impact ro the environment at this site since the bedrockaquifer and the soils are also contaminated. These inconsistencies, between thetwo reports, should be clarified.

ERM will change the text on page 2-14 (Section 2.5.1) to read "five suspectedareas." This change will make the RA consistent with the RI and Section 5 ofthe RA. The revised page is attached.

6. Section 2.5.2, Page 2-15: The RA identifies the areas of elevated volatile organiccompounds (VOCs) as primarily in the central section of the plant. According tothe RI, Figure 5-7 and pages 5-16 through 5-17, there appears to be an area, nearunderground storage tanks, along High Street in the north area of the plan,where xylene was detected in the soil at 1200 parts per billion (ppb). It appearsthat no further investigation (additional soil borings or well installations) wasconducted.

The central portion of the site as identified in the RA includes the plant areabetween High Street, Oliver Street, Memorial Avenue, and Park Street. Asdiscussed in the RI and RA reports, the UST investigation is being conducted incoordination with PADER. It should be noted that chlorinated organiccompounds were not the principal compounds detected during the USTinvestigations. Remediation of the USTs will proceed and wiU be in accordancewith the state program. A copy of the UST report will be submitted to the EPAand PADER when completed.

7. Section 2.5.4, Page 2-17: The RA states that the "source of the bedrockcontamination is unclear and "the primary migration pathway forcontaminants in the bedrock aquifer corresponds to the primary ground waterflow direction." The review of the RI indicated that a thorough investigation intothe occurrence and migration patterns of dense non-aqueous phase liquids(DNAPLs) at this site should be conducted. There is a probability that thebedrock aquifer, which is hydraulically connected to the overburden aquifer, iscontaminated with DNAPLs. The migration of DNAPLs does not necessarilyfollow the same flow paths as ground water. Therefore, the statement that "theprimary migration pathway for contaminants in the bedrock aquifercorresponds to the primary ground water flow direction" may need to bereassessed pending further investigation into the occurrence of DNAPLs.

Group


This comment is addressed in the RI response to EPA comments. General (Page2), Specific (Page 3), and Section 5.5.2 Ground Water Flow Section 5.6.

8. Section 4, Table 4-1: References and Qualifier Codes: As footnoted in the table,EPA has withdrawn the slope factors for benzo(a)pyrene. The values listed inTable 4-1 (taken from the Superfund Public Health Evaluation Manual (SPHEM,1986) must not be used in quantifying risks. The risks associated with thiscompound must be assessed qualitatively.

Comment was deleted per discussions with Eugene Dennis.

9. Section 4, Table 4-1: References and Qualifier Codes: The conversion of thereported chronic reference dose (RfD) to the subchronic RfD listed in the Table forCadmium is confusing. The equation requires explanation and justificationsince it appears to be in disagreement with the discussions on this matter onpages 7-8 and 7-9 of the appropriate guidance document Risk AssessmentGuidance for Superfund. Volume L Human Health Evaluation Manual.

ERM has removed the subchronic RFD for cadmium. This removal had noimpact on the conclusions.

10. Section 4, Table 4-1: References and Qualifier Codes: According to thediscussions appearing on pages 7-16 and 8-5 of Risk Assessment Guidance forSuperfund. Volume I. Human Health Evaluation Manual, inhalation toxicityvalues should not be extrapolated from oral toxicity values. The statementappearing as a note under References and Qualifier Codes must be removed andthe oral slope factor not used as an inhalation slope factor unless its use has beenapproved by EPA's Environmental Criteria and Assessment Office (ECAO).

Comment was deleted per discussions with Eugene Dennis.

11. Section 5.3.3, Pages 5-8 through 5-9: The eleven scenarios should be presented interms of current and future land uses as suggested in Chapter 6 of RiskAssessment Guidance for Superfund. Volume I. Human Health EvaluationManual. Once all data on land uses, populations, etc., are available, discussionsand for calculations should be presented to characterize the likelihood thatexposures might actually occur. If it is unlikely that exposures will occur, thescenario may be excluded from further consideration provided sufficientjustification including review and comment from the EPA and PADER regionalproject mangers.

ERM requests that changes in terminology from "feasible" to "actual" and"hypothetical" to "future" not be made in that the text, tables, and appendiceswould require significant revisions. Although some scenarios, such as ScenarioA, could be dropped completely as unlikglg to .occur, removing these scenarioswould again require major changes to tnfe'aBbuiaenc &s previously stated, these

Th

Group


changes could be incorporated into the Final RA, but the conclusions would notbe impacted. Rather than change the document, ERM recommends andrequests approval that an additional page be inserted prior to the ExecutiveSummary that defines the terminology under the old (EPA, 1986) and interimfinal (EPA, 1989) guidance documents.

We appreciate the review of the RA document and trust the above information satisfiesthe reviewers questions. If additional clarification is required please contact us.

Sincerely,

Teresa SchullerEA Project Manager



RTW/pawAttachmentcc: L. Hoose, ERM

L. Trefsger, Textron LycomingP. BoobP. DuffU. TyagiR. Baker

Group

Environmental Resources Management Inc.855 Springdale Drive • Exton, Pennsylvania 19341 • (215) 524-3500 • Fax One: 524-7335 • Fax Two: 524-7798

10 April 1991

Mr. Eugene DennisRegional Project ManagerUS Environmental Protection AgencyRegion III841 Chestnut BuildingPhiladelphia, PA 19107 FILE: 723-01-03

Re: Response to EPA and PADER Comments: 15 March 1991 DraftFeasibility Study Report for the Textron Lycoming WilliamsportFacility

Dear Mr. Dennis:

On behalf of Textron Lycoming, Environmental Resources Management,Inc. (ERM) is submitting this response to the United States Environmental

Protection Agency (EPA) and the Pennsylvania Department of

Environmental Resources (PADER) comments concerning the Draft

Feasibility Study Report for the Textron Lycoming Williamsport Facility.

The comments received by ERM via fax from EPA on 27 March 1991 are

included in the following response. For the reviewer's convenience, the

agency comments are italicized.

Attached please find revised pages for insertion into the Draft Feasibility

Study. These revisions have not changed the pagination for the document;

therefore, the revised pages should directly replace the corresponding ones

of the original draft. Certain of these pages have been changed to reflect

AR303385



comments as noted in the responses below. The remaining pages reflect

changes to the Remedial Investigation report for the facility.

EPA Comments

Section 2.3.3 Soils

1. Appendix B is referenced in this section. Page B-2 notes that lead was

detected in site soils at levels below those considered protective inresidential areas and refers to a 1989 EPA publication. Recent

recommendations are for lead to be remediated to less than 200 parts

per million (ppm) in residential areas.

The maximum lead concentration detected in site soils was 18.5mg/kg; thus, remediation of the soils for lead is not required.

Reference response to EPA RI comment on Section 5.3.2.3, page 5-17

for additional detail.

Section 2.4 - Media to be Evaluated During the FS

1. Additional remedial measures should be considered in the FS for off-site ground water in the vicinity of Elm. Park. It is not clear that theElm Park recovery well is effectively removing contaminants that are

migrating along strike to the southwest in the bedrock. It is not

adequate for ERM to conclude that additional remedial measures areunnecessary in the Elm Park area based on the statement that the ElmPark well "is thought to intercept that portion of the bedrock flow in theTully member that is contaminated."

AR303386

Group ;


Additional remedial measures have been considered but were notdeemed necessary in the vicinity of Elm Park for the following reasons:

• The on-site ground water recovery system proposed in the FS is

intended to eliminate the source of contamination to the Elm Park

bedrock by intercepting the flow at the Textron plant boundary,

thus enhancing the effectiveness of remediation in Elm Park with

the existing well system.

• Chemical data from the deeper bedrock wells in the Elm Park

area show only trace (ranging from nondetectable to 6.5 parts per

billion (ppb) in MW-57) to nondetectable concentrations (in MW-59and MW-60) of the contaminants of concern in the deeper bedrock.

Therefore, installing a deeper bedrock recovery well in this area

does not appear to be necessary and could induce migration of

more shallow contamination deeper into the bedrock aquifer.

• From the available data, ground water from the bedrock

discharges to the overburden in this area. Any ground waterdischarging to the overburden in this area is within the capture

zone of the Third Street recovery well.

• Removal of contaminants from the bedrock formation is more

effective per volume of ground water recovered on site, where the

mass of contaminants is higher and the migration of compounds

that have not yet moved off site from this area can be halted.

AR303387

Group,


ERM recommends that the proposed on-site remediation system beinstalled and allowed to operate so that its performance and

effectiveness can be properly evaluated relative to both the on-site and

off-site plumes. Data collection and evaluation of the proposed

remedial system will begin within the year after the system is.•>

operational, and the effectiveness of on-site and off-site remediationwill be evaluated two to three years after startup, as explained under

the following comment. At that time, if the effect of on-site

remediation on the cleanup of the off-site plume is deemed inadequate,

additional recovery measures will be considered for implementation byTextron Lycoming.

2. ERM should be referred to EPA 15401G-881003, Guidance on Remedial

Actions for Contaminated Ground Water at Superfund Sites. Exhibit

3-5 provides three methods for estimating the time frame for

remediation of contaminated ground water. The time frame forrestoration of both on-site and off-site ground water should beestimated. It may be that on-site clean-up times could be shortened or

enhanced by additional measures in the off site areas.

Methods for Ground Water Cleanup Time Estimation

The EPA document referenced above discusses three methods for

estimating cleanup time. The three methods are the batch flushing

model, the continuous flushing model, and the advection/dispersion

Group ,


model. In addition, ERM has examined a slug transport model forapplicability to site ground water modeling as well. Attachment A

provides information on the assumptions of each model, their

applicability to the ground water at Textron, and the details of theparameters used for calculating cleanup, times.

Timeframes for On-site and Off-site Ground Water Cleanup

Because there appears to be a residual contaminant source on site, and

because the mass of this source cannot be accurately determined,estimating cleanup time for the on-site overburden and bedrock cannotbe defensibly performed at present. Once the on-site recovery system

proposed in the FS has been operated for a minimum of two to three

years, cleanup times for the on-site plumes can be estimated based on

water quality data collected. In the absence of adequate data toproperly estimate aquifer cleanup times, ERM assumed a 30-year on-site remedial timeframe for purposes of costing. This timeframe is

typically used in FS evaluations where ongoing ground water recovery

for cleanup, rather than containment, has not yet been begun.

Estimates of cleanup time in the off-site bedrock were also not

performed, because the bedrock flow velocities, porosity, and

partitioning coefficients are highly uncertain. It is anticipated that thethree additional bedrock recovery wells proposed in the FS will stop

contaminant migration from the on-site bedrock and allow off-site

remediation to proceed with the existing off-site wells. The rate of off-

AR303389


site bedrock remediation will be determined by observation once the

recovery system proposed in the FS is operating.

Estimates of off-site overburden plume cleanup times are calculable,

and these are detailed in Attachment A. A summary of cleanup times

estimated from applicable models follows. Calculations of cleanup

times are based on the assumption that there will be no migration ofcontaminants from the site once the on-site recovery system is in place.

Using EPA's continuous flushing model and a slug transport model

(Attachment A), the cleanup time estimated for off-site overburden

ground water after installation of the system proposed in the draft FS

ranged from 10 to 420 years.

An important aspect of any cleanup time estimate is the large

uncertainty associated with the estimate. A cleanup time estimate

from either end of the 10 to 420 year range could be justified. However,

until the apparent on-site source of contaminants is contained (the

goal of the recovery system proposed in the FS), and a trend in water

quality response off site is observed, an accurate estimate of the»

cleanup time cannot be determined. Evaluation of the progress of

remediation on site and off site is expected to require at least two to

three years. This is because a trend in water quality in response to the

remedial efforts undertaken must be observable. Natural fluctuations

in water quality due to seasonal changes and uncertainty related to

collection and/or analytical procedures interfere with water quality

1R303390 •'

ItGroup ,


trend analysis. The observable water quality response to remedialefforts must be greater than the data fluctuations due to natural and

analytical causes in order for the data to be useful for modeling. It is

highly unlikely that this magnitude of response will occur in less tb*m

two years.

Within the above range of remedy duration, a reasonable estimate of

cleanup time for the off-site overburden ground water is 13 years, withoperation of the recovery system proposed in the FS. This time wascalculated assuming the off-site plume will migrate as a slug to the

Third Street well at a retarded ground water velocity of 135 ft/day

(Attachment A) due to partitioning of trichloroethylene (TCE) between

the aquifer materials and water. This time is considered a best casefor cleanup of the less mobile of the two principal organiccontaminants (TCE) in the off-site overburden ground water.

Off-site Remediation for On-Site Cleanup

ERM does not believe that expanding off-site ground water recovery

will enhance on-site ground water recovery. On-site ground water

remediation efforts should be performed on site. This is because

remediating on-site ground water via off-site recovery would requirepulling ground water contaminants that have not yet left the site into

an off-site area of lower contamination. By pumping in areas of lower

contaminant concentration (i.e., off site), the mass removal rate per

AR30339I

Group-


volume of ground water recovered decreases as compared to thatachievable with on-site recovery.

The present conditions at the site, consisting of a steady plume shape,

with the peak concentrations located on site, indicate that there is alikely residual source of TCE in the saturated overburden on site.

Until this source is contained, depleted, or remediated, progress on

remediation, both on site and off site, cannot be achieved. The on-site

remedial efforts to date have been insufficient to contain the source.The on-site recovery system proposed in the FS is designed to contain

the on-site contamination. This system will recover ten times the

amount of ground water presently being pumped on site. This on-site

source control also allows the remediation of off-site ground water by

reducing contaminant loadings on the off-site aquifers. Without

source control, neither the existing nor an enhanced off-site recovery

system will have a positive effect on off-site plume remediation.

Section 3.6 - Summary of Detailed Analysis

1. Note that Table 3-2 shows 30 years as an assumed time frame for the

remediation of ground water. The restoration time is an importantparameter that needs to be presented in the ROD for this site, and an

assumed value cannot be used.

See comments under "Timeframes for On-site and Off-site Ground

Water Cleanup" in previous comment.

AR303392


PADER Comments

1. The cleanup level referenced throughout is the MCL. The state'sposition is that all groundwater cleanups should be to backgroundquality if technically feasible.

Regulatory Authority for Background as ARAR

ERM does not agree that EPA should consider the Pennsylvania

ground water "background standard" specified in 25 Pa. Code 264.90-264.100 to be an ARAR. In order for a state standard to be an ARAR, it

must either be applicable or relevant and appropriate (40 CFR 300.4).

The Pennsylvania background standard is neither applicable nor

relevant or appropriate.

In order for a standard to be "applicable", it must "address a

hazardous substance, pollutant, contaminant, remedial action,

location, or other circumstance found at a CERCLA site." (40 CFR

300.5). The Pennsylvania background standard applies only to groundwater remediation at RCRA sites, not CERCLA sites, and is therefore

not applicable at Textron Lycoming.

A standard that is "relevant and appropriate" must "address problems

or situations sufficiently similar to those encountered at the CERCLAsite that their use is well suited to the particular site." (40 CFR 300.5).The Pennsylvania background RCRA standard is intended as a

prospective detection and remediation standard and is inappropriate to

AR303393

Group


apply such a standard to the situation at Textron Lycoming where

long-term historical contamination caused a ground water plume.

Technical Feasibility of Attaining Background in the On-site Aquifers

In practice, both MCLs or background may be equivalent goals andreference points at which remediation will be considered complete.This is because it is uncertain whether ARARs, let alone background

water quality, will ever be reached in the on-site aquifers, despite

expected decades of ground water recovery and treatment. Textron

Lycoming's desire to proceed now using the risk-based MCLs is

therefore reasonable.

The ability of ground water cleanups to achieve background water

quality is presently a topic of intense discussion. Literature reports on

operating ground water recovery systems indicate that cleanups level

off at concentrations above ARARs, such as MCLs, or background. To

our knowledge, the feasibility of reaching background (or even MCLs)

in a reasonable time (less than 20-30 years) for sites with conditions

similar to Textron Lycoming's on-site aquifers has never been

demonstrated. Selection of either MCLs or background as an endpointwould thus not change the ground water recovery system proposed in

the FS, since the remedial system will likely reach its technically

feasible endpoint of recovery above background levels, regardless of

which goal (MCL or background) is established as the endpoint. This

AR30339W

Group


is a reality given available present ground water remedial systems

and will not change unless a new, more effective, method for aquifer

cleanup is developed. Consequently, the Record of Decision (ROD) for

the Textron Lycoming facility should provide that, regardless of thecleanup levels selected, if it is demonstrated that ground watercontamination asymptotically approaches concentrations above the

cleanup levels, EPA shall modify the cleanup levels consistent with

this demonstration.

As an example, the Florida Administrative Code (17-770) includes ananalytical method for evaluating the cleanup endpoint that is based on

the monitored performance of a remedial system.

2. Any calculations of the tower size needed for the air strippers isdependent on the effluent limits set by the facility's NPDES permit.The increased discharge volumes proposed will require modifications

to the permit. Thus, it appears to be too early in the process to

determine specific tower sizes.

As required by EPA guidance for conducting feasibility studies,concept-level cost estimates were prepared for the various treatment

alternatives evaluated. To facilitate cost estimation, preliminary

sizing of all major process equipment, including air strippers, is

typically conducted. ERM has, therefore, sized process equipment toconform to an assumed influent-effluent basis, in the absence of final

AR303395

Group


discharge limits from PADER. As noted on page 3-8 of the FS,

"effluent limitations [will] be determined in coordination with PADER

during the remedial design phase."

3. The proposed system of recovery wells only stretches as far east as thearea ofMW-9 and MW-22. These two wells are areas of high

concentrations of TCE-DCE. Any proposed recovery system must be

able to prevent contamination from leaving the site to the east of thearea ofMW-9 and MW-22.

The on-site recovery system proposed in the FS has been modified to

address this comment to ensure that no migration from the eastern

end of the plant will occur. However, although the water table surface

map suggests that ground water flows south to southeasterly from thearea of MW-9 and MW-20, water quality data from MW-12 and MW-74

downgradient of this area does not indicate that significant off-site

migration has occurred.

A modified version of Figure 1 from Appendix C of the draft FS is

attached showing the revised placement of wells in the eastern area of

the plant. There has been no change in the number of wells required;

rather, the spacing between wells has been increased to include thearea of MW-9 and MW-20, while maintaining acceptable safety factorsfor overlapping cones of depression. An estimate of the additional cost

for this modification in the three FS alternatives involving treatment is

AR303396

Group-,


attached. It should be recognized that the well placements shown onthis figure are preliminary and may require modification at the time ofinstallation, because hydrogeologic conditions vary across the plant

property. The recovery well spacings and pumping rates are a

function of the aquifer saturated thickness, permeability, well yield,and hydraulic gradient. For the FS evaluation of remedialalternatives, these parameters were considered to be uniform;however, ERM does not expect to encounter uniform aquifer conditions

at the time of recovery well system installation. The estimated

saturated thickness and hydraulic gradient assumed for FS design arethe most uniform of parameters along the plant boundary. The aquifer

permeability and well yields are expected to be variable.

The aquifer saturated thickness, permeability, and well performance

will be evaluated upon the completion of each well. The final numberof wells, spacing and pumping rates may be modified to account for

field conditions. In addition to natural variability within the

overburden aquifer, well access for drilling and subsurface utilitiesinterference may influence the final placement and number ofrecovery wells.

4. It appears the Feasibility Study suggests that intensive ground-water

remediation be implemented at the Textron facility property only and

that current off-site remediation processes are sufficient for eventualcomplete abatement. It is important to note that PADER does agree

AR303397

Group ,


that more intensive ground water pumping is needed at the facility

proper. However, off-site remediation currently in operation which

may be containing most of the contamination, is not adequate as far asthe complete "ground-water" resource is concerned.

The primary portion of the ground water resource that PA DER isconcerned with is the extensive sand and gravel aquifer located southof West Fourth Street..... PA DER had relayed the opinion to Textronthat further ground-water recovery in this sand and gravel aquifer

would probably be required once the Third Street recovery well operated

and was studied for aquifer response. PA DER feels that additionalrecovery in the Elm Park area is needed to increase the ongoingground-water cleanup efficiency. It does not seem justifiable to allow

easily attenuated ground-water contamination to continue to migrate

south towards the Third Street recovery well when interception of the

more significantly contaminated ground water within the sand andgravel aquifer can be undertaken. This should also drastically

decrease the overall projected time of the cleanup operation.

PADER is correct that ERM and Textron believe that the additional

remedial efforts should be focused at the Textron facility.

Contaminants continue to migrate from beneath the plant,resupplying mass to the existing off-site plume, at a rate

approximately equal to the rate of mass removal by off-site recovery

wells. With the contaminant supply to the off-site area essentially

AR303398 j

Group ,


eliminated by the proposed intensive on-site ground water recovery

program, off-site ground water cleanup using the existing off-site

systems will proceed much more expeditiously.

Off-site Bedrock in the Elm Park Area

ERM's reasons for not favoring additional off-site ground waterrecovery in the Elm Park area are given under the response to EPA's

Comment 1 for Section 2.4 above.

Off-site Overburden South of Fourth Street

PADER appears to be concerned with the off-site overburden area

south of West Fourth Street in Memorial Park and between the Textronproperty and the Third Street well, as well as the off-site bedrock in the

Elm Park area. PADER implies that there is no equivalent intensive

off-site effort corresponding to the proposed on-site system. The

existing off-site effort, pumping approximately 30 gallons per minute

(gpm) at Elm Park, and between 600 gpm and 1000 gpm from ThirdStreet represents a far greater rate of ground water recovery than is

proposed for total on-site plume containment. Textron has borne all

costs for construction and operations and maintenance (O&M)associated with air stripping towers for the Elm Park and Third Street

wells, as well as stripping towers for the WMWA well field. Analysis

of effluent water from these towers shows that levels of contaminants

are consistently below detection limits. The present off-site system is

AR303399

Group ,


preventing further contaminants from reaching the WMWA well field;

thus, the WMWA system is not subject to future increased VOC

loadings on its treatment systems, and no population is exposed to

additional risk.

Off-site Overburden between the Plant and Third Street

PADER apparently also believes -that additional remediation should

occur closer to the high areas of contamination in the plume to effect

the most efficient remediation of the plume. This was ERM's thoughtin its development of the proposed on-site remedial alternative in the

FS. The proposed on-site system is presently located in the area of

highest concentration; i.e., at the southern plant boundary. It is

anticipated that the on-site ground water recovery system proposed

will eliminate off-site migration of contaminants in the overburden

and bedrock aquifers.

The off-site recovery systems presently in place are deemed adequate to

handle the affected off-site ground water after the proposed on-siterecovery system is implemented. The overburden system was modeled

by Chester Engineers in 1985. They determined that the Third Streetwell could effectively contain the overburden plume by pumping at a

rate of 500 gpm. This well has pumped at rates of 1,000 gpm in the past

and is presently pumping at 650 gpm. Further, water quality in the

Group


WMWA well field has improved since implementation of the off-siterecovery system.

Hypothetical Additional Off-site Ground Water Recovery Scenario

A sound technical basis for the need to increase the contaminant mass

removal rate from the subsurface must be defined before additional off-

site remediation is considered. Furthermore, the need for additionaloff-site recovery and the effectiveness of additional recovery can only bedetermined after the proposed on-site system is made operational.

Because there is no unacceptable risk to municipal water users in the

area and there are no other known ground water users in the area

overlying the plume, adequate time should be taken to properlyevaluate the need for further off-site ground water recovery. Such a

proper evaluation would include detailed examination of water quality

response to operation of the recovery program proposed in the FS and

detailed ground water flow and transport modeling.

However, as requested by EPA to assist the agencies in their

consideration of the benefits of installing additional off-site recovery

versus the cost of such measures, a hypothetical off-site ground water

recovery option has been developed. Two pumping wells and threeinjection wells are assumed to comprise this option. This system

would be located approximately midway between the plant, the Elm

Park recovery well, and the Third Street recovery well, in the

'AR303i*OITht


approximate center of the off-site plume. The goal of this option wouldbe to split the plume approximately in half. This system would recovercontaminants in the area north of it to the plant boundary as well ascapturing contaminants migrating from the area of the Elm Park well

toward the Third Street recovery well. The Third Street well would

then recover water from the area between the hypothetical system and

Third Street. Injection wells are included in the assumed system tomaintain water levels and ground water velocities between the

hypothetical system and the Third Street well and to avoid further

pumping of Lycoming Creek water. Given the centralized location ofthe hypothetical system, it is estimated that this system could reduce

the off-site remediation time by about a factor of two. However,

additional off-site recovery provides no additional reduction in risk or

protection of the WMWA well field compared to the existing system.

This is because the WMWA well field is provided with treatment and

there are no other known users of ground water in the area. In

addition, the estimated capital cost for such a system represents a 50

percent increase over that for the on-site recovery system. Further

discussion of this hypothetical off-site system is found in

Attachment B.

A time for off-site cleanup, assuming the ground water remedial

scenario presented in the FS was approximated as 13 years, based on

an assumed slug transport model (Attachment A) for ground water

Group


flow between the southern plant boundary and the Third Street well.Because of uncertainties in the model, a factor of safety of two wasapplied to this estimate and rounded to the nearest decade, giving anassumed cleanup time of 30 years. The addition of the hypothetical

supplemental off-site recovery system would decrease this cleanup

time to approximately 15 years, as the wells would be located midway

between the Third Street well and the plant boundary. However, datacollected after the expanded on-site system proposed in the FS isoperational are needed to accurately estimate cleanup times, as noted

above. Future evaluation of off-site remedial alternatives to

complement the existing off-site recovery program can be performed

once these data are collected.

5. Page 3-18, State Acceptance - PADER comments on the draft feasibility

study should be addressed in the responsiveness summary section of

the ROD.

Page 3-18 has been modified, as attached, to reflect this change.

6. Page 3-33, Administrative Feasibility - The third and fourth sentences

of this section are very confusing.

The two sentences in question are as follows: "That portion of thedischarge routed to direct discharge will need to be treated to the more

stringent VOC limits for reinjection to avoid redundancy in

AR303W3Tht

Group


equipment. The upflow filtration process will only be used to treat the

flow to be reinjected."

From the design basis for ground water treatment given in Table 3-1, it

may be seen that the ARARs/TBCs estimated for reinjected water aremore stringent than those for surface discharge, assuming limited

stream dilution. Consequently, a treatment system sized for organics

removal for water recharged to the aquifer will be larger or require

greater chemical dosages to meet the discharge limits than that for

surface discharge. Because in Alternative GW-5 a portion of the

treated flow is reinjected, whereas the remainder is direct discharged,two different treatment system sizings would be appropriate to treat

the split streams to their individual effluent limitations. Since two

separate but similar organics treatment systems are not thought to be

cost effective for the on-site recovery system, a single organics

treatment system was sized for the more restrictive effluent basis (for

reinjection).

Similarly, the effluent to be direct discharged is expected to provide a

clear effluent that will not contain unacceptable levels of suspended

solids. For reinjection, even extremely low levels of suspended solids

(under 10 ppm) can be problematic over time, because the solids can

clog the well screens in the recharge wells. Thus, the water to be

discharged via reinjection must be treated to a greater extent for

AR303UOI*Tin

Group


suspended solids removal (e.g. via filtration) than water for directdischarge.

7. Appendix B--Can ERM provide references for the model used in this

section? Has it been used previously at other NPL sites?

The equations used in calculating soil cleanup levels at Textron

Lycoming are based on simple physical principles and have been used

regularly for determining soil cleanup levels at NPL sites.

The general approach and equations used has been compiled

previously and named the Summers model. This model is presented

in EPA/600/7-80-117 (1980). Some examples of NPL sites for which

these equations were used include:

• William Dick Lagoons, PA,

• Geiger/C&M Oil, SC,

• Sand Springs Petrochemical Complex, OK, and

• Pristine, Inc., OH.

For further descriptions of the Summers model and NPL sites where it

has been applied, please refer to "Determining Soil Response ActionLevels Based on Potential Contaminant Migration to Ground Water: A

Compendium of Examples", EPA/540/2-89/057, EPA, 1989.

AR303U05ThttmGroup


The Summers model consists of a simple linear partitioning equation

(Equations B-3 and B-4 in Appendix B of the FS) and a simple dilution

equation (Equation B-ll in Appendix B of the FS). These equations arederived based on simple physical processes. The partitioning equation

is described fully in many texts, for example:

• "Seminar Publication: Transport and Fate of Contaminants in

the Subsurface", EPA/625/4-89/019, EPA, 1989; and

• "Superfund Exposure Assessment Manual", EPA/540/1-88/001,

EPA, 1988.

The dilution equation is derived from a simple application of theprinciple of mass conservation, and is described in the Summers

model.

Use of Vapor-phase Carbon versus Fume Incineration for Off-gas Controls

In your telephone conversation with Mr. Lee Trefsger of Textron

Lycoming and Richard Wroblewski of ERM on 8 April 1991, you noted

that there were inquiries on PADER's part regarding the

undesirability of fume incineration for stripper off-gas controls. For

stripper emissions control, two technologies can be employed: fume

incineration and vapor phase carbon. As noted in Table 2-5 and page2-47 of the draft FS, vapor-phase carbon is not an appropriate

technology for removal of two of the principal organic constituents that

will be present in the stripper off-gas. These compounds are vinylAR303!i06

.'Group ,


chloride and 1,2-dichloroethylene. Vinyl chloride, because of its small

molecular size and chemical properties, is very poorly adsorbed on

vapor-phase carbon and breaks through the carbon bed quickly. The

1,2-dichloroethylene is also relatively poorly adsorbed, relative totrichloroethylene, for example. Consequently, carbon usage is high ifcompound removals commensurate with those from fume

•

incineration are to be maintained. Thus, for the compounds of

concern at Textron, fume incineration is a much more appropriate

technology.

In addition, process control is easier with fume incineration for the

constituents to be present in the off-gas. This is because the operationof the fume incinerator can be electronically linked to the recoverysystem. In this way, should the fume incinerator fall below its design

temperature range (as determined during system design) for

destroying the VOCs present in the off-gas, no untreated air emissionswill be released. There is no equivalent process control parameterwith a vapor-phase carbon system, except perhaps on-line organics

analyzers, which are a less reliable control system.

From the standpoint of mobility, toxicity, and volume destruction,fume incineration provides efficient destruction of the vinyl chloride

over the long term, whereas vapor-phase carbon will not adsorbi

significant quantities of vinyl chloride. Thus, for the reasons cited

Group


above, the selected off-gas control technology of fume incineration setforth in the FS is the better of the two alternatives.

We trust that the above information satisfies your questions on thedraft FS for Textron Lycoming. We appreciate the cooperation andunderstanding you have shown in the past and look forward tocontinuing our work with you in implementing the on-site recovery

system set forth in the FS.

Very truly yours,


Ruth E. Baker, P.E.FS Project Manager


RTW/pawAttachmentcc: L. Trefsger, Textron Lycoming

P. Boob, Textron LycomingP. Duff, TextronT. Kraig, TextronT. Schuller, ERMU. Tyagi, ERML. Hoose, ERMS. Sayko, ERM

AR3Q3l*08

Group

ATTACHMENT ASUMMARY OF CLEAN UPTIME CALCULATIONS

FOR THE OFF-SITE OVERBURDEN PLUME

The document referenced in EPA's comments on the draft FS for theTextron Lycoming site discusses three methods for estimating cleanuptime. The following text summarizes each with respect to itsapplicability for the off-site overburden ground water at Textron

Lycoming. In addition, a slug transport model is presented as an

alternative method of cleanup time calculation for the off-siteoverburden plume.

The Batch Flushing Model

This model is an exponential, or first-order, decay model. It assumesthat the rate of mass removal from the system is proportional to the

mass remaining in the system. In a situation where the mass is beingconvectively transported toward the point of removal, i.e., migration ofthe contaminant mass toward a recovery well, this model mayoverestimate cleanup time. This overestimation is because the massremoval rate achieved through off-site ground water recovery should

remain relatively stable as the plume moves into the recovery wells,

rather than decreasing in proportion to the remaining contaminantmass in the aquifer.

TtnA-l ~~ts»* Group .

The Continuous Flushing Model

The EPA document is somewhat confusing regarding the differencebetween this model and the Batch Flushing model. Both are first-order decay models; however, Exhibit 3-5 of EPA/540/G-88/003states that the difference is in the soil-water partitioning coefficient

(Kd) used in the models, with the Batch Flushing model usingtheoretically derived values and the Continuous Flushing model using•laboratory-derived values. However, Appendix D of that documentindicates that the difference between the two models is that the BatchFlushing model solves the mass removal in steps of pore volumes,whereas the Continuous Flushing model solves the mass removal intime steps. In either case, laboratory evaluation for partitioningcoefficients was not included as a work element in the RI. Therefore,only theoretical partition coefficients can be used. The ContinuousFlushing model would have the same problem as the Batch Flushing

model in predicting cleanup time for a convectively transportedplume.

Both the Batch Flushing and Continuous Flushing models can bereduced from the iterative form presented to a closed form:

Assumption of first-order decaydM/dt = -kM (1)

where M = mass remaining in the system,t = time, andk = decay constant.

AR303U10A-2

Group

Solution

M(t) = M(t=0) exp(-kt) (2)

and if concentration, C, is directly proportional to mass then:

C(t)= C(t=0) exp(-kt) (3)

The Batch and Continous Flushing models are theoretically identical.Therefore, for simplicity in the following discussion, only the

Continuous Flushing model has been referenced below.

The Advection/Dispersion Model

This model describes dispersion of a mass introduced instantaneouslyat a point in the aquifer. It accounts for reduction of concentrations

due to dispersion, but does not account for mass removal by pumping.

Because all of the mass is introduced at a single point, the initialconcentration is infinite. This concentration decreases rapidly withtime; however, the results obtained do not come close to representingthe contaminant distribution that exists in the off-site overburden at

Textron. Therefore, this model was not used for cleanup time

calculation.

Slug Transport Model

In the slug transport model, it is assumed that the off-site plume willclean up as soon as the trailing edge of the plume, Initially assumed to

be at the plant boundary, reaches and is removed from the off-siterecovery well system. The inputs to the model are the distance from

AR303UI ITh«

A-3Group ,

the trailing edge of the plume to the recovery well system and theground water velocity.

Off-site Overburden Cleanup Time Calculations

Overview

Uncertainty in the cleanup time calculations presented below will

result from uncertainty in the estimates of dissolved phasetrichloroethylene (TCE) and dichloroethylene (DCE) (representingapproximately a factor of two uncertainty) in the off-site overburdenaquifer, and in the soil-water partition coefficients (Kd). The Kdvalues were estimated based on two values for the fraction of organiccarbon (foe) assumed for the overburden aquifer. The foe values usedwere 2% and 0.14%. The 2% foe was taken from the calculations ofsite-specific soil cleanup levels presented in Appendix B of the FS; the0.14% foe was determined based on an average Kd value of 0.18 ml/g,presented by Mehran, et. al. (1987) Distribution Coefficient of

Trichloroethvlene in Soil-Water Systems in Ground Water Journal, Vol.

25, No. 3.

ERM has performed calculations of cleanup times by two methods.

The first of these (Continuous Flushing Model) assumes first-orderdecay using estimates of the TCE and DCE mass. A mass removal rate,dM/dt, was set equal to the removal rates at the Third Street recoverywell and has little uncertainty. For the second model (SlugTransport), a range of cleanup times was developed by evaluatinjgtwo

A-4-Group .

soil-water partition coefficients (assuming foc=2% and foc=0.14%).

The final estimated cleanup time presented in this Attachment isbased on the Slug Transport model, because the Slug model is not asoverly conservative as the Continuous Flushing model.

It should be recognized that neither of the models discussed above."» "*"

account for the complex flow conditions at the site or for chemical

and biological decay, factors that may be significant in the degradationof volatile organic compounds (VOCs).

Continuous Flushing Model

Using equation (1) above, the decay constant (k) can be determined ifthe rate of mass removal, dM/dt, and the total mass are known values.

Based on data from the Third Street recovery well, the mass removalrate of TCE and DCE from the off-site overburden is approximately400 pounds per year. Estimated total dissolved TCE and DCE masses

in both the on- and off-site overburden are 2,500 and 600 pounds,

respectively. The amount of TCE and DCE sorbed to the overburdenaquifer sediments was estimated to range as follows for the cases of0.2% and 2% foe:

sorbed TCE : 1,900 to 28,000 pounds

sorbed DCE : 190 to 2,800 pounds

These figures represent only the sorbed on-site and off-site mass. ForDCE, approximately 50% of the mass is off-site; for TCE,

AR303U3

A-5Group.,

approximately 80% is off-site. It should be noted that these figures donot include the apparent residual TCE mass on site.

Two values of the decay constant (k) were determined using the 400pounds per year mass removal rate and the range for the total TCE andDCE mass range. The calculated k values were 0.015/yr and 0.10/yr.

Assuming that concentration is directly proportional to mass, i.e.,equation 3 is valid, cleanup times for the off-site plume with theexisting off-site recovery system were calculated for well MW-59,where TCE concentrations were 2,900 in August 1990, and theacceptable level was assumed to be 5 ppb. Calculated clean up time fortotal VOCs ranged from about 60 to 420 years. As discussed

previously, the first-order decay model (Continuous Flushing) isconsidered overly conservative for conditions at Textron, but is usefulin providing an upper bound on cleanup time.

Slug Transport Model:

The slug transport model assumes that the plume will migrate withoutdispersion and with a velocity equal to the ground water seepagevelocity (v) divided by the retardation coefficient (R). Clean up time(t) is equal to the distance (d) from the trailing edge of the plume to

the recovery well(s):

t = d / (v/R) (4)

From this equation, estimated clean up times for TCE and DCE were:AR303U1U

A-6

TCE : 13 to 90 years

DCE : 10 to 42 years

The retardation factors used in the calculations were 1.8 and 12 forTCE, and 1.3 and 5,i for DCE (corresponding to the foe values of0.14% and 2%). An unretaraed ground water seepage velocity of 243feet per year and a distance from the plant boundary to the Third

* •

Street recovery well of 1,800 feet were assumed.

Summary

Estimated cleanup times for the off-site overburden plume rangedfrom 10 to 420 years. There are other similar approaches toevaluating cleanup time in addition to those covered above; however,the results should fall into the above range and other available modelswould probably not reduce the uncertainty inherent in cleanup timesfurther. The uncertainty in the cleanup time could be narrowed with

additional testing to evaluate soil-water partitioning in combination

with numerical flow and transport modeling. These evaluations should

occur during the implementation and operation of the on-site recoveryprogram detailed in the FS. This is because developing accurateestimates of cleanup times will require the observation of transient

changes in water quality under known aquifer conditions. Such known

conditions are only obtainable from the operation of full-scale groundwater recovery systems over a period of at least two to three years.

AR303M5

A-7r _Group

ATTACHMENT BPRELIMINARY EVALUATION OF FURTHER OFF-SITE REMEDIAL

MEASURES

As requested by EPA, ERM has examined preliminarily a supplementalground water recovery system for the off-site overburden plume. Thefollowing brief evaluation has been prepared to assist EPA in assessingthe value of additional off-site ground water recovery and treatment.As noted in ERM's response to FS comments, ERM reemphasizes thatthere is no technical need to implement additional off-site groundwater recovery at this time, as the existing recovery wells incombination with the proposed on-site recovery system will result inan equivalent off-site remediation, albeit somewhat slower.

As explained in Comment 4 of the response to PADER, thehypothetical ground water recovery system examined to supplementexisting off-site ground water recovery and treatment consists of two

overburden extraction wells and three reinjection wells. The

extraction wells would recover approximately 200 gpm of flow fromthe vicinity of Elm Park and the central portion of the off-site plumenear MW-67. To maintain ground water levels and flow velocities, the

water would be reinjected following treatment. This flow is assumedto be treated via air stripping and fume incineration, with filtrationbefore reinjection. Residuals would be limited to solids recovered viafiltration after organics treatment and gross solids removed before airstripping to minimize tower packing fouling. These residuals would

AR303M6B-l

Group .

be disposed of off site at an approved facility. The ground waterrecovered and treated in this supplemental off-site system wouldundergo a permanent reduction in toxicity and volume of organics.Stripped organics would be destroyed via fume incineration, if off-gascontrols are employed.

Installation of such a system would not cause additional risk tosurrounding populations or workers or adverse impacts on the

environment. Conversely, the addition of more ground water recovery

wells in the off-site overburden would not provide increasedprotection of human health and the environment, because the existingmunicipal water supply wells are furnished with air stripping to

reduce organics in the ground water to safe levels. Thus, there is no

presently exposed population for untreated water. The hypotheticalrecovery system could possibly cut in half the 30-year cleanup durationconsidered for off-site overburden ground water followingimplementation of the recovery program proposed In the FS.

Acheivement of ARARs in the off-site ground water is a function of theeffectiveness of the on-site and off-site ground water recovery systems.The addition of more off-site wells, as assumed in this hypotheticalsystem, would not reduce the source of contamination, which isbelieved to be on site in the form of residual TCE on aquifer materials.

The hypothetical remedial system is able to be constructed and usesexisting proven technologies. The assumed location of the recoverywells and reinjection points would need to be further refined, based

RR303UI7B-2

Group..

on field studies of the aquifer thickness and ground water flow

patterns in the area. Siting the hypothetical recovery and treatment

system is impeded by of the presence of ball fields and a park in thearea. Utilities, including power and fuel, would need to be supplied.Provision for fuel storage for the fume incinerator would need to be

made near the treatment system. The treatment system may need tobe located some distance from the wells to avoid occupying heavilyused areas of the park. Twenty-four-hour access over the long term towells and the treatment system would need to be secured for

maintenance and sampling, because the system would be located onproperty not owned by Textron Lycoming. Underground piping wouldneed to be laid among the recovery and reinjection wells and thetreatment system. The installation of this piping would disrupt theexisting park in the short term.

Permitting requirements could include an air permit for the stripperand limitations for discharge to ground water, both from PADER.

The cost for the hypothetical treatment system is expected to consistof a capital cost of approximately $800,000, with estimated annualcosts of $140,000, resulting in a 15-year present worth ofapproximately $2.3 million. The hypothetical system would reducecleanup time to half of that for the on-site recovery system specified inthe FS, but would increase the capital cost of the overall remedy by anestimated 50 percent. Thus, the benefit of more rapid plume recovery

is equally balanced by the increased cost of the remedy. No increased

AR303M8B-3 LliL.Group .

protection from risk to public water users would be afforded by thissupplemental recovery system, as the remedial measures already takenoff site include cleanup of recovered drinking water to acceptablestandards.

AR303UI9

B-4X,-Group ,

Documents

Environmental Resources Management, Inc. Resources Management, Inc. 855 Springdale Drive • Exton, Pennsylvania 19341 • (215) 524-3500 • Fax One: 524-7335 • Fax Two: 524-7798