January 2003

1.0 Introduction

1.1 Indications from Previous Traces

Dye tracing in October-November 2001 indicated that the phreatic zone in the Valhalla area might travel east towards the 4-series wells. The injections in LF-6- 8" indicated that the epikarst in the southeast corner of Lemon Lane may travel southwest into Valhalla. Both injections were detected in well 00-370 in Valhalla. What was left unclear was whether there was another major phreatic pathway out of Valhalla to IC spring or whether the major phreatic pathway was back towards the east side of Lemon Lane as the low flow phreatic water levels suggested. The lack of coincident timing of the storm tracer pulse from LF-64" and the PCB pulse at IC spring indicated that the timing of injected tracer pulses would be a good indication of where the immediate source of the high PCB pulse might or might not be. Dye was introduced into LF-64" just prior to a rain event. The first rain event was small and did not move any tracer from the location. Two storms of greater magnitude followed shortly thereafter and produced dye breakthrough curves at IC spring. But there was a 5 and 6 hour delay, respectively, between the PCB arrival time and the dye arrival time. Based on those results, it was hypothesized that tracer travel times that matched the times predicted by the regression analysis of PCB pulse arrival times might indicate the source location(s) of the high PCB pulse.

1.2 Reasons for Conducting Tracing

• Induction of PCB response
• Comparison of travel times
• Detection of site pathways

Viacom has constructed a relationship between flow rate at IC spring and the time of arrival of the high concentration PCB pulse during storm events. That relationship is an exponential function which has been empirically derived from at least 21 stomm events over a range of rainfalls and is:

Travel Time in hours = 1211.3 (IC flow in gpm)^-0.8591

Although this is a storm derived relationship, we believe that a tracer injection location would be a candidate for the high PCB source area only if the timing of that tracer pulse will match that of the derived equation. If there is a significant lag time between tracer arrival and calculated PCB arrival time then we might presume that location is not a significant contributor to the high PCB storm pulse. Such a situation seemed to be the case in the aforementioned LF-6-4' trace.

It was also reasoned that various epikarst locations could be flushed with water and tagged with tracer. If a PCB pulse could be induced at IC spring arriving coincident with the injected tracer, then this might reveal possible source locations.

The third reason for tracing was to discern pathways around the site. This involved sampling other monitoring wells after tracer injection and trying to discern downgradient from upgradient locations. This was not always apparent from water levels and potentiometric surface interpretations, especially since the phreatic gradient was so flat.

2.0 Description and Results of Traces

A general summary of the dye traces is given in Table 1. The first column shows the date and time of dye injection. The second column shows which dye, either Rhodamine WT (ROOT) or Fluorescein (FLR) was injected. First detection refers to the first sample location the dye was detected, and the hours from injection that the first sample was taken that had the dye positively detected. The fifth column is the date and time of first dye arrival at Illinois Central Emergence and the next column shows that first arrival time in hours. The seventh column is the average flow rate at IC spring between dye injection and first detection at the spring. The eighth column is the travel time predicted by the empirical equation derived from the PCB storm pulses. The last column is the percentage of the mass of dye recovered at the spring. Tables 2-10 show the results of all the traces at the various sample locations.

2.1 MW4i and MW-6 Tracer Test

The scope of work for conducting the MW6-MW4i low flow trace is contained in Attachment 1. The trace was designed to compare travel times to the IC spring in order to evaluate whether MW6 is on an upgradient pathway from MW4i or is on a separate but similar pathway to the spring. MW4i was injected with 225 grams of Rhodamine WT (Cl Acid Red 388) at 12:45 on 7/10/02. MW6 was injected with 50 grams of Fluorescein (Cl Acid Yellow 73) at 13:10 on 7/10/02. The injections were accomplished through an aluminum tube with the outlet set at the 799' elevation. Each well was flushed with 30 gal. of water. Monitoring wells 00-587, NN-300, 00-300, 00-370, 00-125, and NN-12 were sampled twice daily. Quarry B spring and Slaughterhouse spring were sampled twice daily and Illinois Central was sampled with an auto sampler. The monitoring wells were sampled until 7/12. The springs were sampled through the next tracer injection of 8-6-02.

The RWT injected in MW4i was not detected in any other monitored well, but was detected beginning 7/11/02 18:30 hours at IC spring. The RWT rose to a peak of 636 ppb on 7/12/02 02:30 hours and returned to around pre-injection background levels sometime on the morning of 7/24. The mass recovered was about 53% of injected. The first arrival time of 29.75 hours is close to the 27.2 hours predicted by the PCB storm pulse regression analysis.

The FLR injected in MW6 was detected on the 7/11 11:00 hr. sample from NN- 300. The dye was not detected in any other well or spring, with the possible exception of MW4i on a 7/31 sample. This dye was not detected at IC spring despite a 0.95n rain on 7/9, a 1.07" rain on 7/23, and a 0.46" rain on 7/29. Figure 1 shows a plot of the RWT and FLR at IC spring for the first week after injection with the RWT concentration on the left scale from 0-700 ppb and the FLR concentration on the right scale from 0-7 ppb and indicate the magnitude of difference of tracer detection of the two injection locations at the spring.

2.2 PZ-F Trace

The scope of work for conducting the PZ-F trace is contained in Attachment 2. The goal of this test was to see if a PCB spike could be.induced and if the travel time to the spring was indicative of the PCB storm pulse travel time. 1000 gallons of water was to be flushed to induce this pulse. After 300 gallons had been flushed, 225 grams of RWT was injected and the rest of the 700 gallons of water was flushed. The dye was poured in at the top. Previous slug testing indicated water exiting major fractures or conduits at elevations 851' and 828'. Time of dye injection was 8/6/02 at 10:00 hours. The following wells were to be sampled 4 hours after the flush and then daily: NN-412, NN-300A, 00-300A, 00-370, and MW-4i. Well NN-300A had too little water in the bottom of the well to get a sample and well NN412 was completely dry. IC spring was sampled with an auto sampler.

The tracer was detected in 00-300A and 00-370 on the first sample after injection. Although that sample was 5.75 hours after injection, there was visible dye in both wells indicating the dye was there before the sample was taken. The tracer was not detected at well MW4i. The tracer first appeared at IC spring on 8/8 at 20:00 hours, 58 hours after injection. This compared well with the predicted travel time of 58.6 hours based on the storm PCB regression. The dye peaked at 284 ppb on 8/8 13:00 hours and receded to background sometime on 8/22. Approximately 35% of the mass injected was recovered at IC spring. Table 11 shows the PCB results at IC spring. Figure 2 shows the plot of the RWT breakthrough curve and the PCBs, and indicates little, if any, PCB response induced by the flush.

2.3 NN-300A Trace

The scope of work for conducting the NN-300A trace is contained in Attachment 3. The goal of this test was to see if a PCB spike could be induced and if the travel time to the spring was indicative of the PCB storm pulse travel time. 1000 gallons of water was to be flushed to induce this pulse. After 300 gallons had been flushed, 225 grams of RWT was injected and the rest of the 700 gallons of water was flushed. The dye was poured in at the top. Time of dye injection was 8/20/02 at 09:15 hours. The following wells were to be sampled 4 hours after the flush and then daily: NN-300, 00-300A, 00-387, MW-6, 00-370, and MW4i. IC spring was sampled with an auto sampler.

The tracer was detected in 00-370 on the first sample taken 30 minutes after injection. The tracer was detected in NN-300 and 00-300A in the first samples taken after injection. The tracer was not detected at wells MW-6, 00-387, or MW-4i. The tracer first appeared at IC spring on 8/23 at 16:00 hours, 79 hours after injection. This was much longer than the predicted travel time of 54.5 hours based on the storm PCB regression. The dye peaked at 8.2 ppb on 8/24 07:00 hours and sampling was discontinued on 8/24 before the tracer receded to background. Approximately 4% of the mass injected was recovered at IC spring. Table 12 shows the PCB results at IC spring and indicates little, if any, PCB response induced by the flush.

Figure 3 shows a plot of the RWT breakthrough curves for both the PZ-F trace and the NN-300A trace. The time scale is in days from dye injection. The left scale shows the RWT concentration at IC spring from the PZ-F flush and the right scale shows the RWT concentration at IC spring from the NN-300A flush and is 1/10 the scale. Note that PZ-F is only 70' north of NN-300A.

2.4 MW16-MW18 Trace

The scope of work for the MW16-MW18 trace is contained in Attachment 4. The goal of the trace was to compare arrival times from each well to IC spring. Low flow water levels taken at phreatic wells around the site would sometimes show MW16 as a local low and it was wondered whether there was a separate path to the spring from this area of the site. MW18 was injected with 225 grams of Rhodamine WT (Cl Acid Red 388) at 13:15 on 10/1/02. MW16 was injected with 50 grams of Fluorescein (Cl Acid Yellow 73) at 14:00 on 10/1/02. The injections were accomplished through an aluminum tube with the outlet set at the 799' elevation. Each well was flushed with 30 gal. of water. Monitoring wells MW4i, MW19, and MW15 were sampled twice daily. Quarry springs combined and Slaughterhouse spring were sampled twice daily and Illinois Central was sampled with an auto sampler. The monitoring wells were sampled until 10/3. The springs were sampled through 10/5 at 09:00.

The RWT from MW18 was detected in MW4i in the first sample taken after injection within 3.25 hours. The first detection of the RWT at IC spring was 10/2 at 20:00 hours or 30.75 hours after injection which compares to 26.6 hours as predicted by the storm PCB regression. The RWT peaked at 398 ppb on 10/3 at 04:00 hours and about 60% of the mass was recovered.

The FLR from MW16 was present in 1\/lW4i and MW19 on the morning sample of those wells on 10/2, about 18.5 hours later. The first occurrence of the FLR at IC spring was 10/3 at 12:00 hours or46 hours after injection as compared to a predicted travel time of 27.2 hours based on the PCB storm regression. The FLR peaked at 15.84 ppb on 10/4 at 01:00 hours and about 27% of the mass was recovered by the time sampling ceased on 10/5. Figure 4 is a plot of the breakthrough curves for the respective dyes. Notice the right scale for the FLR is 1/10 the scale for the ROOT.

2.5 MW6-00370 Trace

The scope of work for the MW6-00370 trace is contained in Attachment 5. No discernable break through curve was seen at IC spring the last time MW6 was flushed. Several dye traces have had intermediate detections in 00370. The goals of this test were:

• Increase the dye amount injected to produce a breakthrough curve at IC spring so travel times could be compared to the PCB storm regression.
• Determine if lack of detection at IC spring was because dye traveled to other springs.
• See if site pathways led back to the 4-series area.

MW6 was injected with 250 grams of Fluorescein (Cl Acid Yellow 73) at 09:15 on 10/22/02. 00370 was injected with 450 grams of Rhodamine WT (Cl Acid Red 388) at 10:00 on 10/22/02. The injections were accomplished through an aluminum tube with the outlet set at the 799' elevation. Each well was flushed with 30 gal. of water. Monitoring wells MW4i, MW4s, MW20, MW-10, MW18, MW19, MW16, and MW15 were sampled twice daily on the eastside of Lemon Lane. 00587, NN625, 00387, NN300, 00300, 00125, and NN12 were sampled twice daily in Valhalla. Quarry springs combined, Detmer, Snoddy- Hinkle combined, Bypass 37, Urban, Crestmont and Slaughterhouse spring were sampled twice daily. Both Illinois Central and Stony East spring were sampled with an auto sampler. Stony West began by being sampled twice daily, and later in the test (beginning on 10/30 16:00) an auto sampler was added there as well. The monitoring wells were sampled until 11/8. The springs were sampled through 11/8 also.

The RWT from 00370 was first detected in MW18 in the sample taken 24 hours after injection at 10/23 10:50. It was then detected in MW20 on the 10/23 14:18 sample. Figure 5 shows the RWT concentration in the east side wells. Note that the Y-axis is log scale. Although reaching concentrations in well MW18 as high as 19,490 ppb, the RWT was never conclusively detected in any springs including IC spring.

The FLR from MW6 was present in MW4i and MW20 on the morning sample of those wells on 10/30, about 190.5 hours after injection, but only at 18.82 and 12.27 ppb respectively. There was 42.09 ppb detected in the 10/30 07:48 sample in MW4s. Figure 6 show the FLR concentration in the east side wells. There was no conclusive detection of FLR at any spring including IC spring.

3.0 Discussion

3.1 Induction of PCB Response

The idea of the induction of a PCB response was based on the premise that flooding a location with water tagged with tracer would force pockets of high PCBs into the phreatic conduits. PCBs in these locations would normally be flushed out during storm events. By forcing them out during non-stomm events we could selectively measure the possible contribution from the flushed locations. The travel times would correspond to those calculated from the PCB storm regression equation.

However, neither the flush test of LF-6-8" in April, nor the flush test of PZ-F or NN-300A produced any significant PCB response at IC spring. Whether this means that the locations themselves are not contributors to the high PCB storm pulse, or the low flow flushing does not adequately simulate a storm event is not completely clear. Response in phreatic wells to flushing PZ-F and NN-300A was nearly instantaneous, as shown on Figures 7-9, which may indicate that the water, rather than flushing the epikarst around the borehole, descended straight into the phreatic zone.

3.2 Comparison of Travel Times

There seem to be three categories of traces with regard to travel time from the site to IC spring. They would be:

• Those tracer injections which are never detected.
• Tracer travel times significantly longer than predicted.
• Tracer travel times comparable to that predicted by the empirically derived equation based on high PCB storm pulses.

1. The October 2001 injection of Phloxine B into SP-1.
2. The October 2001 injection of Eosine into NN-700.
3. The July 2002 injection of Fluorescein into MW-6.
4. The October 2002 injection of Fluorescein into MW-6.
5. The October 2002 injection of Rhodamine WT into 00-370.

1. The May 1996 injection of RWT into North Sink.
2. The October 2001 injection of FLR into LF-6-8".
3. The May 2002 injection of FLR into LF-64".
4. The August 2002 injection of RWT into NN-300A.
5. The October 2002 injection of FLR into MW-16.

1. The May 1996 injection of FLR into MW-7 area macropores.
2. The April 2002 injection of FLR into LF-6-8".
3. The July 2002 injection of RWT into MW4i.
4. The August 2002 injection of RWT into PZ-F.
5. The October 2002 injections of RWT into MW-18.

Our interpretation, at the present time, is that these categories represent hierarchal organization within the Illinois Central Spring basin karst conduit branchwork and the proximity or open connection of the injection location to a major tributary conduit. Traces that never produce break through curves at the spring represent areas that are not drained by major phreatic tributaries, perhaps because they are at the "upper reach" of the drainage basin. The "upper reach" aspect may area-related, that is, far away in distance, or vertical-related, in that the vadose segments may not be fed by significant recharge. They may also take long, tortuous pathways so that dye concentrations are longitudinally dispersed by the time they reach the spring. Another possibility, of anastomotic flood mazes, is discussed in Palmer (1991) (Palmer, Arthur N., 1991, Origin and Morphology of Limestone Caves, Geological Society of America Bulletin, v.103, p. 1-21.). This refers to a conduit collapse or blockage which results in flood waters dissolving an anastomotic maze around the blockage. Palmer cites Blue Springs Cavern in Indiana as an example. The dye disperses in the maze to concentrations not detectable at the spring.

Tracer travel times that are significantly longer than predicted by the PCB storm pulse regression represent locations that are "farther" from the spring, for example - North Sink, in terms of length of phreatic conduit. Another aspect of "distance" may also be lateral travel in the vadose zone. This may explain why the October 2001 injection of FLR into LF-6-8" and the storm injection of FLR in LF-6-4" of May 2002 took longer than the April 2002 injection of FLR into LF-6-8". The difference in those injections was the amount of flushing water - 1032 gallons for the April 2002 trace. This above normal amount of water may have flushed the dye into a shorter pathway than it would have normally taken. However, since the volume of water altered the travel time results, caution should be exercised in extrapolating the results of these non-storm dye tests to storm conditions.

Locations where the tracer travel times match the PCB storm pulse are candidate locations for investigation as possible source areas for the PCB storm pulse. It may be instructive to note that tracer locations in Valhalla have either produced no break through curves or delayed passage to the spring. Refer back to Figure 3 - Comparison of Breakthrough Curves for PZ-F and NN-300A Traces as an illustration. This implies that the high PCB pulse does not originate in the Valhalla area, but lies north of the tracks under the Lemon Lane site itself.

3.3 Detection of Site Pathways

It is pumling why there is a lack of tracer detection at IC spring from MW6 and why the 00370 trace would show up in MW18 at concentrations as high as 19,400 ppb but not produce a break through curve at the spring. It is probably necessary to repeat this trace. Nevertheless, a pattern does emerge as shown on Figure 10, which is a summary of the tracer detections in the wells around the site. Although some traces were detected in well 00370, that well was conclusively traced to MW18. From a simple Connect the dots" perspective, it seems apparent that the near site flow directions are from west to east and towards the 4 series wells. This would imply that a major collector tributary conduit exists along the east side of Lemon Lane, possibly just east or north of the 4-series wells. We suspect most of the near-site drainage, and in particular high PCB source areas, converges toward this junction before being conveyed to IC spring.