exhibit 4 - hudsoncag.ene.com · exhibit 4. interoffice correspondence to: mr. douglas garbarini,...

EXHIBIT 4

INTEROFFICECORRESPONDENCE

To: Mr. Douglas Garbarini, Region 2, USEPAMr. Benny Conetta, Region 2, USEPA

Date: March 15, 2006

From: John B. Mulligan, P.E.

Re: Response to NOAA Comment Letter of October 18, 2005

Comments on the Intermediate Design Report (IDR) were received from NOAA in aletter signed by Lisa Rosman, NOAA Coastal Resource Coordinator, addressed to AlisonHess, USEPA (EPA), and dated October 18, 2005. At EPA’s request, I prepared a draftmemorandum, dated December 20, 2005, which discussed NOAA’s concerns.Subsequent to that date, a number of meetings have been held with General Electric (GE)to address EPA’s comments on the IDR. These meetings have resulted in a resolution tomost of EPA’s comments and, as a result, I am updating my December 20 memorandum.

Most of the comments contained in NOAA’s October 18, 2005 letter identify whatNOAA believes may be shortcomings in the IDR or variations from the recommendationscontained in the Record of Decision (ROD) or supporting documents, such as theFeasibility Study (FS). NOAA’s comment letter, which consists of some 72 pages oftext, quotes extensively from the ROD and supporting documents and argues forrevisions in the IDR or changes to be incorporated in the Final Design Report (FDR)documents for Phase 1 of the project.

NOAA’s comments fall into two general categories:1. Engineering design issues relative to the overall effectiveness of the design in

achieving the goals of the remedy as outlined in the ROD. Specific issues raisedinclude the approach used by GE to develop dredge prisms, which NOAAcontends will result in leaving some polychlorinated biphenyls (PCB)contaminated sediment in areas targeted for remediation; the proposed plan toconstruct engineered caps over areas where dredging does not remove all PCBcontamination; the long-term stability of the proposed backfill and engineeredcaps; and the potential for recontamination of dredged areas through theredistribution of contaminated sediments from areas that are not dredged.

2. Issues related to habitat replacement and reconstruction design including theproposed designs for backfill, underwater caps and shoreline restoration work,and the potential impact of these designs on the long-term recovery of the Riverand its related habitat.

This updated memorandum has been prepared to address NOAA’s expressed concernsabout the design, but does not attempt to respond to each and every point in the commentletter. Rather, it attempts to state how the design addresses the issues raised by NOAAand what I anticipate will occur as the Final Design for Phase 1 is completed and workbegins in the River.

Mr. Douglas Garbarini, Region 2 USEPAMr. Benny Conetta, Region 2 USEPA

March 15, 2006Page 2

COMMENTS ON DREDGE PRISM DEVELOPMENT APPROACH

I believe that the method being used by GE to develop dredge prisms will adequatelydefine the limits of contaminated sediments for the purposes of designing initial dredgecut lines, with the understanding that sampling conducted upon completion of the initialdredging attempt will identify any areas where a second dredging attempt will benecessary to remove missed inventory and that GE will conduct additional dredging toremove it. It is EPA’s intention to review the efficacy of GE’s dredge prism design at theend of Phase 1 and implement any changes to the design techniques, includingconsideration of kriging analysis, in Phase 2 of the project.

GE’s Phase 1 Dredge Area Delineation Report (QEA, February, 2005) estimated thehorizontal and vertical extent of contamination on the basis of a kriging model.However, GE has proposed using a method referred to as Inverse Distance Weighted(IDW) averaging to define the initial dredge cut prisms. This method uses actualsampling results at each sample point to determine the depth of contamination at thatpoint and a procedure for estimating the depth of contamination between that point andadjacent sampling points where the actual depth of contamination is also known. GE hasargued that the IDW method places more importance on the results of individual coresamples than kriging and should be a more accurate means of defining the depth ofcontamination for dredge prism development.

Experts for EPA and GE have discussed and argued the merits of the various methods ofdetermining the depth of contamination on which to base the design of initial dredge cutlines at great length. These arguments considered a number of factors includingthe following:

¯ The individual core samples do not necessarily produce an accurate picture of thetrue thickness of the contaminated layer of sediment at any point. In fact, anevaluation of the laboratory analytical data for 60 pairs of cores, purposelylocated within one meter of each other, shows that any pair of cores located sideby side in the river can produce a different estimate of the thickness of thecontaminated layer at that location. For example, one core of a pair may showthat the thickness of the contaminated sediment layer is 30 inches while thesecond core, located within one meter of the first, shows the thickness to be36 inches. Another pair of cores taken at another location may show a differenceof 12 inches in the thickness of the layer while a third pair in still another locationmay show a difference of only 3 inches. On average, the difference shown by the60 pairs of "co-located" cores was 10 inches. Thus, the thickness of thecontaminated sediment layer determined at any individual core sample locationshould only be considered accurate to plus or minus 10 inches, on average, andestablishing the bottom of the contaminated sediment layer based on the actuallaboratory data for the cores will probably result in dredging too deep into theriver bottom about half the time and too shallow the other half. For the



approximately 92 acres targeted for dredging during Phase 1 of the project, it islikely that about half the area will be dredged too deeply and result in the removalof an unknown volume of clean sediment for shipment to a toxic waste site.Similarly about half the area will be dredged too shallow leaving contaminatedsediment behind after the initial dredging cut is completed. In these locations, GEwill have to develop a new set of dredge cut lines and re-dredge to remove thisinventory. Since the sampling information at any given core location is equallyapt to underestimate the depth of contamination as to overestimate it, it is likelythat the need for re-dredging to remove missed inventory will occur randomlyover the entire Phase 1 project area.

¯ The kriging method lends itself to an analysis of statistical confidence limits whilethe use of the IDW method does not. The use of kriging to help define the dredgecut lines could reduce the number of areas where the initial dredging might not bedeep enough to remove all of the contaminated sediment provided that thedredging depth is increased at all locations to take into account the inaccuracies inthe core sampling results. Furthermore, the likelihood of success in removing theentire contaminated layer by including some over-cut allowance in the designcould be predicted using statistical formulae incorporated in the kriging model.Both the IDW method and a kriging model were developed to estimate the bottomsurface of the contaminated sediment layer for the Phase 1 areas. A comparisonbetween the two techniques performed by Kern Statistical Services, Inc. yieldedthe observation that the surfaces generated by the two models were fairly similar.This observation suggests that the optimization routines used for the IDW modeland the standard statistical routines used in kriging have converged on a fairlysimilar result, an estimate of what is probably the median surface. While there arelocal differences, due in part to the closer adherence of the IDW to individualpoints and the tendency of kriging to approach the local mean, both approachesgive a reasonable estimate of the median surface.

¯ There is no good way of conducting a statistical evaluation of the likelihood ofsuccess at removing the entire contaminated layer by adding an allowance forover cutting if the IDW method is used. As noted above, this method shouldprovide an adequate depth of cut about 50 percent of the time, assuming noovercut is included in developing the dredge prisms; whereas the kriging methodwould allow a statistical analysis of the probability of dredging to an adequatedepth under various assumptions of overcut depth.

¯ GE contends that the cost of dewatering, shipping and disposing of sediment farexceeds the cost of dredging and, therefore, the Company is opposed to includingany overcut allowance in the dredging design other than a minor (15 percent)allowance to account for sloping the edges of dredge cut areas to achieve stableside slopes. Rather, GE intends to conduct the initial dredging to design cutelevations based on the IDW method, sample to determine where the initial cuthas removed all of the contaminated sediment, and where inventory has been leftbehind, to develop new dredge prism designs for re-dredging to remove themissed inventory.



The Engineering Performance Standard for Residuals provides for re-definingdredge cut lines to remove contaminated sediment missed during initial dredging.Thus, if sampling following the initial dredging attempt shows that inventory hasbeen left behind, GE is required to do additional dredging to remove it. Failing todredge to a deep enough depth to remove the inventory during the first dredgingattempt does not mean that it will remain in the river or that GE will be allowed toconstruct an engineered cap over it.

I am not completely convinced of GE’s assertions concerning cost, as no cost data areprovided in any of the documents that I have seen to date. However, dredging costs aretypically a lot less than the costs for dewatering, transporting and disposing of sedimentsand GE’s assertion seems reasonable. What is not clear to me is whether over cutting bya few inches during the first attempt at removing inventory would result in avoiding theneed to re-dredge a substantial (40 to 60 percent) portion of the targeted area. A re-dredging pass will likely require a minimum cut of about 6 inches, and it is not clearwhether including an overcut of 2 inches or so for the entire area would produce moresediment for disposal than will ultimately be removed if no overcut is included in theinitial pass of the dredge and a 6 inch layer is removed from 40 to 60 percent of thedredging area in a subsequent re-dredging attempt aimed at removing missed inventory.Regardless, EPA has agreed to allow GE to base its dredge prism design for Phase 1 onthe IDW method and is relying on the Residuals Standard to serve as a check on GE’sdesign. If the work in Phase 1 indicates that money can be saved by including someminimal overcut allowance in the development of dredge prisms, I am sure that GE willtake note of this and design the dredge prisms for Phase 2 accordingly.

It is my understanding that EPA generally agrees with NOAA that kriging providesanother useful way to define dredge cut lines and lends itself to an evaluation ofstatistical confidence limits for the dredge cut lines. However, EPA also recognizes thatany delineation scheme may produce dredge cut lines which result in removing asubstantial volume of clean sediment from the river in addition to the contaminatedsediment targeted for removal. GE has argued strongly against incurring additional coststo dredge, dewater and dispose of clean sediment merely to reduce the probability ofhaving to re-dredge some areas to remove inventory missed during the initial dredgingattempt, and has committed to at least one re-dredging attempt to remove missedinventory and up to two additional attempts to remove residual sediments in accordancewith the Residuals Standards.

Re-dredging to remove missed inventory will be based on new dredge prisms developedfrom core samples taken after the first dredging attempt and will presumably involve theremoval of at least 6 inches of sediment. EPA will review these new dredge prismsbefore approving the re-dredging attempt. If the initial dredging attempt results in a largenumber of areas where inventory was missed, EPA may request that an overcutallowance be included in the design of the new dredge prisms.



The volume of sediment removed during re-dredging to capture missed inventory countstoward the targeted removal volume of 265,000 cubic yards for Phase 1 of the project.Thus, to the extent that the initial dredge cut lines established in the FDR may not bedeep enough in all instances to capture all of the inventory, the targeted volume may beachieved before all of the areas identified to be dredged during Phase 1 are actuallydredged. I anticipate that the initial dredging attempt, as defined by the dredge prismsprovided by GE in their FDR, will underestimate the actual depth needed to remove allinventory about 50 percent of the time. Furthermore, I expect this to be the case withineach certification unit, (i.e., each unit will have some areas where inventory is missed,rather than half the certification units being dredged deep enough during the firstdredging attempt and half requiring re-dredging to capture missed inventory). We shouldgain the experience to know whether this is actually the case early in the Phase 1 project,certainly within the first month of dredging and sampling, and can use this experience inreviewing any new dredge prisms developed by GE to remove missed inventory. If GE’sapproach to designing dredge cuts by the IDW method, without any allowance forovercut, turns out to miss inventory in substantially more than 50 percent of the initialcertification areas dredged, we may recommend to EPA that GE be required to include anovercut allowance in prisms developed for re-dredging.

Following any re-dredging attempt to remove missed inventory, a new round of sedimentsamples will be collected. This round of sampling is designed to identify areas whereresidual contamination still remains and will generally focus on a 6-inch deep sedimentlayer. Where residual contamination is identified, and where it must be removed to meetthe Residuals Standard, a first attempt at re-dredging to remove residuals will be madeand another round of samples collected. This re-dredging will likely result in a cut ofabout 6 inches, as this is the approximate minimum cut that a contractor is likely toattempt using a mechanical dredge. If the new round of samples shows that residualsediments still exceed the criteria established in the Residuals Standard a second, andpresumable final, attempt at removing these residuals will be made, followed by stillanother round of sediment sampling. If this second attempt at removing residualsediment still fails to meet the Residuals Standard, GE may request permission to capnon-compliant areas. However, GE may also decide to continue dredging in non-compliant areas to meet with the Standard if it believes that it is can do so at less costthan that of constructing an engineered cap.

Inasmuch as Phase 1 of the project will be evaluated extensively to test manyassumptions used for design, EPA has accepted GE’s approach to designing dredge cutlines for Phase 1 of the project. As the Phase 1 dredging proceeds and the amount ofre-dredging needed to remove missed inventory can be determined, EPA and GE willevaluate whether the IDW method is successful or whether kriging or some other methodof defining dredge prisms should be employed for the Phase 2 design. In this regard, Inote that the Consent Decree language permits EPA to require changes to the methodused to design dredge prisms for Phase 2, if necessary.



COMMENTS ON DREDGING ADJACENT TO THE SHORELINE

NOAA has expressed concerns about the procedure proposed by GE for dredge prismdevelopment along the shoreline. GE’s proposed procedure consists of cutting a 2 footvertical face at the limit of dredging along the shoreline and then proceeding into deeperwater on a 3 horizontal to 1 vertical slope. The 3:1 slope has been selected by GE as aslope that will generally be stable and not prone to collapse or cave-in. However, wherethe slope of the shoreline sediments is already steeper than 3:1, GE will follow thesteeper slope.

Under GE’s proposed method of designing dredge prisms, where the depth ofcontamination (DOC) is less than 2 feet at the shoreline, the full depth of thecontaminated sediment layer will be removed up to the shoreline, the horizontal limit ofdredging. However, if the contaminated layer extends onto the floodplain, beyond thehorizontal limit of dredging, an exposed face of contaminated material will remain at theedge of the excavation. This exposed face will not be stable, and waves impinging on theshore can be expected to cause it to collapse to a relatively stable slope almostimmediately after dredging. To the extent that the sediment in the exposed face at theedge of the dredge prism may be contaminated, the collapse of the exposed face couldre-contaminate a narrow strip of river bottom along the shoreline (see Figure 1, attached).

¯ This material may be identified during sampling for residuals and removed as part of theresiduals dredging effort. However, the additional dredging needed to remove residualsediments immediately adjacent to the shore may cause additional cave-ins or sloughingof contaminated material at the edge of the dredge cut and start the cycle of sampling forresiduals and re-dredging along the shoreline over again unless some means of stabilizingthe face of the excavation is developed.

Where the DOC is greater than 2 feet at the shoreline, the initial vertical cut will notreach the bottom of the contaminated layer. However, the dredge cut will continuedownward by 1 foot for every 3 feet that the dredge moves away from the shoreline, andit is anticipated that in this manner the dredge cut will reach the bottom of thecontaminated layer relatively close to shore. Nevertheless, a small wedge ofcontaminated sediment could be left behind on the fiver bottom. Furthermore, thevertical face of the dredge cut at the shoreline will expose the edge of the contaminatedsediment layer if it extends inland (i.e., into the floodplain), and this sediment is likely toslough off into the fiver at the edge of the dredge cut. This condition is shownschematically on Figure 1.

Although there is the real possibility that some contaminated sediment may be left at orimmediately adjacent to the shoreline when the initial dredging pass is completed, GE’sproposed design calls for conducting additional dredging if the total PCB concentration isover 50 milligrams per kilogram (mg/kg), even if this means that a temporary sheet pilewall must be installed or that some floodplain sediments must be removed. Sediment



with a total PCB concentration less than 50 mg/kg will be capped or removed at GE’sdiscretion, consistent with the Residuals Standard.

NOAA has pointed out that leaving contaminated sediment in place in areas targeted forremoval may be at variance with the Engineering Performance Standard for Residualsand may require the construction of an engineered cap that will not be suitable to supportthe re-growth of submerged aquatic vegetation. Furthermore, the ROD calls for dredgingand removing contaminated sediments rather than capping them. However, EPA believesthat the approach reflected in the Consent Decree does not represent a significantchange to the remedy selected in the ROD, and also believes that while the ConsentDecree’s approach with respect to the near-shore areas embodies certain changes to theEngineering Performance Standard for Dredging Residuals, the approach is protective ofhuman health and the environment. Further, EPA has noted that it can require changes tothe approach if information gathered during Phase 1 shows such changes are needed.

Neither the ROD nor the Residuals Standard defines the limits of removal at the shorelineor addresses the removal of contaminated sediments from the floodplain. To allow theengineers to move forward with the design, EPA has defined a "limit of dredging" for theproject as the line defined by the water’s edge at a flow of 5,000 cubic feet per second(cfs) measured at the Fort Edward gage. This line is approximately 6 inches above thewater level normally experienced during low, summer time flows.

A small number of samples collected just behind the river bank by EPA as part of anongoing evaluation of floodplain sediments indicates that, in some areas, the layer ofcontaminated sediment found in the river near the shoreline extends onto the floodplainbeyond the limit of dredging. However, it is not known how far this contaminationextends away from the River, how thick the contaminated layer is, or whether it issignificant enough to require remediation under some future project. For designpurposes, GE is estimating the depth of contamination at the shoreline by extrapolationfrom core data collected at sampling points in the river from 10 to 80 feet off shore. Postdredging samples will be collected to determine whether any contaminated material hasbeen left in the river adjacent to the shoreline and will provide the information needed todecide how to restore the shoreline to protect the environment.

Before agreeing to accept GE’s proposedapproach to dredge prism design at theshoreline, EPA considered other approaches to this problem. One approach consideredwas to drive a row of temporary sheet piling along the shoreline at the limit of dredging.The full depth of any contaminated sediment could then be dredged on the river side ofthis temporary sheet pile wall. After dredging and sampling to confirm that thecontaminated material had been removed, backfill would be placed in the river up againstthe sheet pile wall and graded to create a stable slope. The temporary sheet pile wallwould then be removed.



The use of a temporary sheet pile wall to stabilize the fiver bank would prevent bankcave-ins and the sloughing of contaminated sediment into the water from beyond thedredging limit. However, driving the sheet piling would require trimming trees along theriver bank that would interfere with the pile driving work and would create a great deal ofnoise. Inasmuch as the water depth in some areas is not adequate to float a barge with alarge crane needed to drive the sheeting, some access to the shoreline from the adjacentland would be required and additional clearing of trees and brush would be needed.

Whenever a dredging project is designed, decisions must be made on how to terminatethe excavation at the shoreline. Clearly, a transition is required from a 2 or 3 foot deepexcavation just off-shore to the existing ground surface at or just behind the shoreline.While this transition could be made very abruptly by constructing a retaining wall alongthe shoreline, such a structure would limit easy access to the river and would require longterm maintenance. The installation of a temporary retaining structure to stabilize theshoreline until dredging can be completed and backfill placed in the river, followed byremoval of this structure, would be feasible but would require the removal of many treesalong the river’s edge and could have a negative impact on meeting the scheduledeveloped in the Productivity Standard. Furthermore, where the PCB contaminatedsediment layer extends inland, behind the temporary retaining structure, the newshoreline would have to be capped to withstand erosion in the same manner as will berequired if a small wedge of contaminated sediment is left below a sloped cut line asproposed by GE.

In accepting GE’s proposed methodology for designing dredge prisms along theshoreline, EPA considered various factors, including the following:

¯ The average depth of contamination in Phase 1 dredge areas is approximately1.8 feet based on a volume of 265,000 cubic yards and an estimated area to bedredged of 92 acres. The 2 foot vertical cut proposed by GE at the shoreline isgreater than the average depth of contamination and should result in completeremoval of contamination along a substantial portion of the affected shore.Furthermore, if the post dredging sampling identifies areas where total PCBconcentrations are in excess of 50 mg/kg in the wedge of contaminated sedimentremaining along the shoreline below the dredge cut line, GE is required toexcavate deeper to remove it.

¯ With the vertical edge of the dredge cut generally limited to 2 feet at theshoreline, some inventory is likely to remain. However, it is estimated that themass remaining within 10 feet of shore in River Section 1 will be about 1 percentof the mass targeted for removal in this Section. The total PCB concentrationbelow 2 feet was greater than 50 mg/kg in slightly less than 15 percent of allsample nodes within 20 feet of the 5000 cfs shoreline in the Thompson IslandPool. However, the agreement to continue dredging to remove sediment with totalPCB concentrations in excess of 50 mg/kg will reduce the mass remaining. Itshould be noted, however, that these estimates are based on a review of sampleslocated near the shore rather than directly at the limit of dredging.

Mr. Douglas Garbarini, Region 2 USEPAMr, Benny Conetta, Region 2 USEPA


¯ To test the IDW method used by GE to estimate the depth of contamination at theshoreline, EPA requested that GE collect samples along a few transectsperpendicular to the shore where it was anticipated that the depth ofcontamination would exceed 2 feet. GE collected these samples at 3 locationsalong the river bank in the east channel at Rogers Island and provided the resultsto EPA in December, 2005. The sampling results indicated that the depth ofcontamination predicted using the IDW method was significantly greater at twotransects than was actually found in the field and that dredging to the predicteddepth would remove all of the contaminated sediment plus a 6 to 12 inch layer ofclean, underlying material. The sampling results at the other transect were notconclusive, as rocks or other obstructions were encountered at shallow depths andsome samples could not be recovered.

¯ EPA and GE are currently discussing the need for additional sampling along theshoreline for Phase 2 dredging areas; some additional limited sampling will alsobe performed in Phase 1 shoreline areas. The inventory remaining farther than10 feet from shore should be removed by the 3:1 dredge cut. The existingshoreline slope in the Phase 1 target areas is estimated at 6.5:1 or flatter for50 percent or more of the dredge areas and 4:1 or flatter for 90 percent or more ofthe dredge areas. With these relatively gradual slopes, the depth of the dredge cutwill be 3 feet at a distance of 6 feet from the shoreline in 50 percent or more ofthe dredge areas. Similarly, the depth of cut will be 3 feet or deeper at 12 feetfrom the shoreline in 90 percent or more of the dredge areas. This approach forthe shoreline should provide comparable results in the remainder of River Section(RS) 1, which will not be dredged until Phase 2 of the Project, and in RS 2 and 3,where the shoreline slopes appear to be even flatter than in RS 1.

¯ GE’s commitment to remove shoreline contamination to levels less than 50 mg/kgtotal PCB is numerically equivalent to the threshold for the ResidualsPerformance Standard, which allows one location to fall between 15 and26 mg/kg Tri÷PCB. Conversion of these Tri+ values to total PCB concentrationsfor comparison purposes depends on the nature of the residual sediment. In-placematerials at 15 to 26 mg/kg Tri+ would be expected to have correspondingtotalPCB concentrations between 30 and 52 mg/kg, essentially twice the Tri+ value.Redeposited material would be expected to have corresponding total PCBconcentrations at 45 to 78 mg/kg, three times the Tri+ value. In both cases,the total PCB criterion to be used by GE falls within the range allowed bythe Residuals standard. Notably, the Residuals Standard only allows oneoccurrence of a level of between 15 and 26 mg/kg Tri+ PCBs in a certificationunit. The use of this allowance while trying to meet the 1 mg/kg mean residualTri+PCB concentration in the Certification Unit (CU) without a near-shore capwould mean the rest of the CU would have to be dredged to lower values toaccommodate one or more high values along the shore.

¯ When a cap is used in a CU, the remaining areas must satisfy stricter residualcontamination criteria, with fewer high values and a mean less than 1 mg/kg Tri+PCBs. Thus, any decision to leave contamination along the shore would result in



additional dredging elsewhere in the CU in order to satisfy the standard. Thiswould likely increase the cost of the initial work plus the cost of long termmaintenance and should provide a strong disincentive for leaving easily removedshoreline contamination behind.EPA has not waived the Residuals Standard for the CUs that contact the shoreline,but GE has agreed to collect and analyze additional core samples at a closerspacing to determine the extent of any wedge of contamination that might be leftbehind beneath the 3:1 slope of the dredge cut adjacent to the shoreline. If, afterdredging, a wedge of contaminated sediment is found immediately adjacent to theshore and it has a total PCB concentration less than 50 mg/kg, GE has the optionof either removing it or capping it. By agreeing to collect additional samplesalong the shoreline to better define the limits of any wedge of contamination leftbehind, the extent of any capping needed to isolate this material from theenvironment should be reduced.The 50 mg/kg total PCB removal commitment must also be viewed in light of theselection criteria for River Sections 2 and 3. The 30 mg/kg Tri+ surfaceconcentration (one of the selection criteria for targeting contaminants for removalin RS 2 and RS 3) represents a total PCB concentration of from 60 to 90 mg/kg,higher than the PCB level which may be capped in the shoreline areas under theConsent Decree approach. Thus, assuming that the current near-shore approach iscontinued in Phase 2, GE may be constructing caps in certain near shore areasover sediment contamination that is lower than the threshold criteria for removal.While constructing an engineered cap in a strip along the shoreline is notdesirable from an ecological standpoint and should generally be avoided, sometreatment will be required to isolate the face of any contaminated sediment layerthat extends onto the floodplain beyond the limit of dredging. Failing to isolatethe exposed edge of such a layer would allow it to erode into the River and,ultimately, to form a new, narrow beach of contaminated material along theshoreline, regardless of whether the depth of cut at the limit of dredging isadequate to remove the full depth of PCB contamination or leaves a small,triangular wedge of contaminated sediment behind. Details on the design ofshoreline restoration work were not provided in the IDR but EPA requested thatthis information be submitted in advance of final design in its comments on thedesign documents. GE provided a briefing to EPA on their proposed designs onMarch 1, 2006. The preliminary drawings presented by GE during the briefingindicate that the Company plans to restore the shoreline using non-structuralmeans wherever possible and will place backfill or a combination of"dirty" riprapand backfill to the full depth of the vertical cut at the shoreline where the fivervelocities are equal to or greater than about 1.5 feet per second. Where lowervelocities are present, backfill alone will be used and the shoreline will be seededor planted to obtain a vegetative cap as soon as possible. The use of biologs wasalso proposed in some locations where these devices can be expected to hold thefiver bank until vegetation can be established. Where the river bank is currentlyarmored with riprap, GE expects to replace this material in kind. The important



point tO note in this regard is that GE’s designs, as currently envisioned, willcover the face of any contaminated layer that extends onto the floodplain and isexposed by dredging. EPA will consider this further in the review of the Phase 1final Design Report. Furthermore, GE is not proposing to armor extensivereaches of shoreline which are currently in a natural state with riprap, concreteretaining walls or other structural means, and the methods presented to EPA in thebriefing appear to have minimal impact on the long term restoration of shorelinehabitat.

While leaving any PCB contamination in the fiver is less than ideal, the difficultiesassociated with the shoreline areas require an adjustment in the engineering approach solong as the ultimate remedial goals are unaffected. The approach agreed to with GE willachieve these goals in a manner that meets the intent of the ROD. EPA will have anopportunity to reconsider the approach based on the experience gained in Phase 1.

COMMENTS ON DESIGN OF ENGINEERED CAPS AND BACKFILL

NOAA has expressed concerns about the potential for excessive use of engineered capsto cover PCB contaminated sediment that GE cannot successfully remove throughdredging, the design of these caps to resist erosion, and the selection of material forbackfill and capping that will support aquatic vegetation.

EPA has commented on GE’s proposed design of engineered caps in its review of theIDR. These comments convey NOAA’s concerns about the assumptions made by GErelative to the extent of capping and the need for the inclusion of fine grained material inthe backfill and capping designs to facilitate habitat restoration (see EPA Comments 20,21, 23, 113, 114, 115, 119, 125, & 126 on GE’s IDR).

As noted in NOAA’s comments, all backfill systems and underwater caps are vulnerableto damage or destruction by floods and other catastrophic events and, for this reason, adredging alternative was selected in the ROD rather than a capping alternative. Becauseit is inevitable that some contamination will be missed during dredging, the ROD alsocalls for 1 foot of clean backfill to be placed in dredged areas to provide a temporaryprotective layer over any residual contamination. This backfill layer is not designed towithstand major flood events, as such a design would require an armored layer of stoneand would cause a major change in the River’s habitat. Rather, the backfill layer isdesigned to be reasonably compatible with the River’s existing sediments and to supportthe restoration of the existing biological community. It is expected that the backfill willundergo some erosion and redistribution over time, as does the sediment in any river bedunder changing conditions of flow, floods and ice movements, and that the small amountsof residual contamination left in the dredged areas, if eroded, will become mixed with thebackfill material so that the PCB concentration is diluted to a very low level. The gradualredistribution of backfill can also be expected to produce a relatively clean layer of new



sediment over areas that were not dredged and reduce the average concentration of PCBsat the sediment surface throughout the upper Hudson.

GE has selected backfill material that is slightly coarser than the existing sediment foundin the areas identified for dredging. This backfill will be acquired from local gravel pitsand typically will contain from 5 to 10 percent silt and clay sized particles unless it isprocessed through a sand washing plant. GE is looking into the sources of the backfill todetermine what is readily available and will make a decision as to whether the backfillwill be washed to remove fines in the final design. At this time, it is anticipated that thematerial will not be washed and will contain a low percentage of silt and clay sizedparticles. When placed in areas of the river with moderate velocities, the silt and claysized particles may be eroded from the surface of the backfill layer over time, leaving anaturally armored surface of sand and gravel sized particles behind, and resettle indepositional areas downstream to replenish silt and clay sediments removed by dredging.This is the naturally occurring process by which river beds achieve and maintain stabilityand will result, over a number of years, in the distribution of sediment types throughoutthe remediation area which is similar to that which currently exists. The backfill shouldnot be considered as a significant source of silt and clay to the river, however, as thepercentage of silt and clay sized particles in the material is likely to be quite low.

In those locations where multiple dredging attempts fail to achieve the ResidualsStandard, an engineered cap may be constructed over the river bed. GE is required tomonitor and maintain all engineered caps for an extensive period of time, and capmonitoring and maintenance is costly. Therefore, there is a very real financial incentivefor GE to achieve the residuals standard and avoid the need for engineered caps.

Two cap designs have been developed for use in certification units where dredging hasfailed to meet the residuals standard of an average of 1 mg/kg Tri+ PCB with only onesample exceeding 15 mg/kg and no sample exceeding 26 mg/kg. A Type A cap designwill be employed to isolate a portion of the certification unit where the average Tri+PCBconcentration is 6 mg/kg or less. A Type B cap will be employed to isolate portions of acertification unit with a Tri+PCB concentration greater than 6 mg/kg and where GE andEPA agree that additional dredging is not required. In the case of both cap types, the capswill be placed over a large enough portion of a certification unit to achieve an arithmeticaverage Tri+PCB concentration in the uncapped nodes of 1 mg/kg or less with noindividual uncapped node having a Tri+PCB concentration of 15 mg/kg or above.

The Type A cap is designed to withstand a minimum 10 year recurrent interval floodflow event and provide resistance to ice damage. It will be at least 12 inches thick andwill contain an armored surface and a layer of finer material to prevent the transport ofresidual sediment up through the armor material. The Type B cap is designed towithstand a minimum 100 year recurrent interval flood flow, ice damage, and vesselwakes and prop wash in areas likely to be subject to such events. The Type B cap will beat least 12 inches thick and will contain an armored layer and a separate layer at least



6 inches thick with a high total organic carbon concentration designed to minimize theflux of PCB from the residual sediment below the cap to the water column.

EPA considered the following facts in deciding whether two types of caps would beappropriate and, if so, where a Type A cap could be employed and where it would benecessary to construct a Type B cap:

¯ The Residuals Standard permits one sample node in a certification unit to have aTri+PCB concentration of between 15 mg/kg and 26 mg/kg, provided that thearithmetic average Tri+PCB concentration in the top 6 inches of the sediment isless than or equal to 1 mg/kg. If these criteria are met, backfill will be installedover the entire certification unit.

¯ Habitat replacement is a very important aspectof the project, and an engineeredcap that is "environmentally friendly" is desirable wherever the use of such a capwill not unduly jeopardize the primary objective of removing PCBs from theenvironment. If all caps are designed to withstand a 100 year flood event, icedamage, prop wash, and damage from boat anchors, boats running aground, andsimilar unpredictable incidents, all caps will have to be heavily armored and maynot be conducive to habitat replacement. On the other hand, by designing capswith minimal armoring that are more suitable to the growth of SAV and wetlandplants for those areas where the mass of PCBs left behind is low, the potentialimpact of an engineered cap on the river bottom environment can be minimized.Accordingly, EPA agrees with GE’s approach to cap selection and design. Underthis approach, a Type A cap is not as heavily armored as a Type B cap and willprovide more favorable conditions for the growth of SAV and wetland plants thana Type B cap.

¯ Where necessary for the re-establishment of SAV beds or other aquatic plants thatrequire fine grained sediment, a layer of suitable, fined grained material can beplaced on top of any backfill or engineered cap to improve the conditionsnecessary for habitat replacement as part of the habitat treatment.

¯ Engineered caps must be designed to withstand a design maximum flow velocityin the river. For a flood of any given magnitude, the maximum flow velocity willonly occur at one point in the River and will be a function of the river’s crosssectional area, its slope and the roughness of the fiver bed. Other locations in thefiver will experience lower velocities under the same flood flow. Even at a singlecross section of the River, the velocities may differ significantly along oppositeshorelines and the velocity at either shore may be much less than the velocity inthe main channel of the river. Therefore, a number of different designs, eachsuitable for a narrow range of local velocities, can reduce the impact of cappingcontaminated sediment left behind after dredging.

¯ Inasmuch as it is not practical to design and install a great variety of caps to meetdifferent flow velocities and it is difficult to accurately predict the velocity at anypoint in the fiver under different flood magnitudes, GE has limited its design ofType A caps to two prototypes: a first design for use in locations where thevelocity during the 10 year flood will be less than 1.5 feet per second (fps) and a



second design for use in locations where the velocity will be between 1.5 fps and5 fps, the maximum velocity predicted to occur at any point in River Section 1during a flood of this magnitude. Similarly, a limited number of Type B capprototypes have been proposed for use in different ranges of velocity. For Type Bcaps, three prototype designs have been developed: one for velocities of less than1.5 fps, a medium velocity cap for use where velocities are between 1.5 and3.5 fps, and a high velocity cap for use where the velocity is predicted to rangefrom 3.5 to 6 fps. Thus, a Type A cap may be installed in an area where the flowvelocity is only 2 or 3 fps even though it is designed to be stable under a 5 fpsflow velocity. Similarly, a Type B cap may be installed in an area with a velocitysignificantly less than its design velocity. Since it is unlikely that Type A orType B caps will always be needed in areas where the velocities are at the highend of their respective design ranges, these caps should, on average, withstandfloods of greater magnitude than the 10 year and 100 year floods selected fordesign purposes.

COMMENTS ON HABITAT REPLACEMENT/RESTORATION

NOAA asserts in its comment letter of October 18, 2005 that the design presented in theIDR appears to differ from the ROD and post-ROD documents in regard to some aspectsof habitat replacement. These issues are summarized in NOAA’s comments as follows:

¯ Cap design does not contain a benthic substrate layer or specifically identifiedhabitat layer.

¯ No requirement for at least a 1:1 habitat replacement program for fringingwetlands and submerged aquatic vegetation (SAV).

¯ No requirement to provide adequate backfill to restore fringing wetlands andSAV.

¯ Presumptive natural recovery approach for SAV rather than active recovery plan.¯ Habitat replacement and reconstruction design elements preclude serious

consideration of future input by natural resource agencies and the public.¯ Design is not responsive to National Remedy Review Board (NRRB) or Trustee

recommendations on especially sensitive or unique habitats (ESUHs).

The IDR presents conceptual designs for habitat replacement and reconstruction.Significantly more detail will be presented in the FDR. Cap design in the Phase 1 FDRwill be done in accordance with the Critical Phase 1 Design Elements attachment to theCD, which does not include a specific benthic substrate layer or a habitat layer forreestablishment of vegetation. The IDR does, however, provide for the installation ofadditional fine-grained backfill over an engineered cap to assist in the re-establishment ofhabitat where river velocities make this feasible. Backfill specified in the IDR(Section 3.9) followed most of the conceptual recommendations presented in the RODand Feasibility Study, specifying sand and gravel as the two types of material to be usedfor backfill. This material is anticipated to be run-of-bank material and is currentlyexpected to contain some percentage of fine grained (passing the 200 sieve) soil. GE’s



design also anticipates that, over time, silt and fine sands will be transported into thebackfilled areas by currents, gradually increasing the heterogeneity of the substrates andthat the deposition of these fine grained materials will play a role in shaping post-&edging habitats (see IDR page 3-122).

For shallow wetland areas, the Feasibility Study envisioned pre-removal water depthsbeing re-established using a combination of sand and fine sand blended with siltymaterial. However, the IDR did not include any specific provisions for use of fine-grained material in backfill, other than noting that the material would come from localpits which contain some percentage of fine soils. This issue was raised by EPA in itscomments on the IDR (Comment No. 20) submitted to GE. GE has responded that it willconsider use of fine-grained material for SAV and wetland restoration in the final design.Subsequently, during discussions with GE on habitat replacement, the Companyindicated that the final design would include the use of fined grained backfill in lowenergy areas of the fiver where SAV and or riverine fringing wetlands are to be restored.

The need for 1:1 wetland mitigation was raised in EPA’s comments on the IDR submittedto GE (Comment No. 130). GE responded that the intent of the project is to replace theriverine fringing wetlands removed by dredging. EPA and GE have agreed to thefollowing language regarding the replacement of riverine fringing wetlands: "For riverinefringing wetlands unavoidably lost or adversely affected by the dredging project, the goalis to replace the functions provided by those wetlands, i.e., no net loss of functions.Functional replacement of the riverine fringing wetlands will be accomplished by thereplacement of the riverine fringing wetlands in their original locations, to the extentpracticable and appropriate, consistent with the remedy. For locations where it is notpracticable or appropriate to replace the wetland in its original location, and where it isdetermined appropriate by EPA to do so, additional mitigation activities will beundertaken in other dredge areas to replace the lost functions of that wetland."

The Consent Decree (Appendix B, Critical Design Elements, Section 2.7) contains theapproach that has been agreed upon for SAV restoration. The replacement of SAV bedswill be dictated by the post-dredging river bathymetry (depth) and other factors thatcontrol the occurrence of SAV beds (e.g., current velocity, light availability) within theproject area. Where water depths and other controlling factors support the replacement ofSAV beds, the goal is to replace those beds. Consistent with agreements between GE andUSEPA, areas in the project area that supported SAV beds prior to dredging andbackfilling will be evaluated to determine if the resulting water depth has increased to apoint where these beds would no longer be supported. Where necessary, additionalbackfill will be placed to reduce the water depth so that SAV beds can be supported. Thedesign includes an allocation of up to 15 percent of the total backfill volume estimated tobe placed as part of the entire project (1 foot over all dredge areas), for restoring waterdepths as needed for the creation of aquatic vegetation beds. An evaluation by EPAindicates that the additional 15 percent allowance for additional backfill material will beadequate to return areas that currently support SAV beds to their pre-dredging water



depth. However, EPA has requested that GE conduct its own evaluation to determinehow much backfill would be needed to re-establish pre-dredging contours in the photiczone to assist in deciding, ultimately, how restoration should occur in specific areas(Comment No. 134 in EPA’s comments on the IDR). GE subsequently provided thisestimate to EPA, which shows that the allowance should be more than adequate to restoredredged areas currently containing SAV to their pre-dredging water depth.Approximately 9000 cubic yards of the approximately 13,000 cubic yards included in theallowance will be required to bring the river bed back to its pre-dredging depth in SAVareas, assuming that dredging extends to the depth indicated in the dredge prisms. The4000 cubic yards remaining in the allowance will be available for use where dredging iscarried out to a deeper depth to remove missed inventory or residuals.

Section 3.10.2.2 of the IDR outlines two replacement and reconstruction optionsfor aquatic vegetation beds: one that uses natural recovery and another that uses bothnatural recovery and human engineering approaches. In both, backfill and/or cap materialare used as the growing medium. The specific areas at which natural and humanengineering approaches will be implemented will be determined using the aquaticvegetation model currently being developed as a part of the Adaptive Management Plan.This model incorporates parameters related to the pre-remediation distribution of aquaticvegetation to identify areas suitable for the establishment of aquatic vegetation followingdredging. The model will be provided in the Adaptive Management Plan accompanyingthe Phase 1 FDR.

The precise areas where SAV will be planted will be identified in the FDR. EPA hasrequested further information on sources of the aquatic material that will be used forplanting (IDR Comment 134). GE has responded that aquatic vegetation plantingmaterial (both submerged and emergent species) will be acquired from commercialsuppliers and that specifications will be provided with the FDR.

The NOAA comment also touched on the importance of re-establishing native HudsonRiver plant species.The importance of maintaining native aquatic vegetation species isrecognized in the Habitat Assessment Report (HAR) for Phase 1 Candidate Areas and isincorporated into the functional capacity indices (FCIs). SAV FCIs for support ofphytophilous macroinvertebrates/benthic macroinvertebrates populations (FCIsAvMACROS)

and for providing habitat for fish populations (FCIsAvFISH) both include a variable forplant species composition (% native) (see Section 4.2.2 of HAR). Establishment of exoticspecies would lower functionality of aquatic vegetation beds, .thereby requiring moreadaptive management. Therefore, reestablishment of native species at the earliest possibledate would assist in restoring or replacing habitat function to within the range offunctions found in similar physical settings in the Upper Hudson River, which is theprimary goal of the habitat replacement and reconstruction. This issue was also raised byEPA in comments provided to GE on the IDR (see Comments 121 and 129), and GE hasagreed to this, where practicable.



The habitat replacement and reconstruction program includes submittals at variousplanning stages of the project that natural resource agencies may comment upon,including the Phase 1 FDR and associated Adaptive Management Plan. There will also bedocuments associated with the Phase 2 program, such as a habitat assessment report anddesign documents.

The dredge prism design is based on the Statement of Work (SOW) for Remedial Actionand Operations, Maintenance and Monitoring attached to the CD (Appendix B). As notedin this document, EPA may determine if dredge prism boundaries should be modified toavoid impacting unique or sensitive habitats. EPA and the Trustees are currentlydiscussing concerns that the Trustees have related to especially sensitive and unique areasin the Phase 2 program.

caw

AttachmentsF:~PROJECT\4553013kDOCkRevisedresponse toNOAA3-13 -06jbrrLdoc

X

o.o "o orm

N ~6

//

/

12..

0

rY0.__J.__JLI_

0

m

i!!E

exhibit 4 - hudsoncag.ene.com · exhibit 4. interoffice correspondence to: mr. douglas garbarini,...

Documents