Ellicott's Experience in Dredging PCB's 

The Remediation via Dredging of Lake Järnsjön in Sweden: Project Implementation for PCB Clean-up:

"The Dredging Equipment Performed Well in Limiting PCB Resuspension"The remediation of Lake Järnsjön in Emån was carried out during the period 1993-1994. PCB-contaminated sediments were dredged, dewatered and disposed of in a landfill. In order to reduce the spreading of suspended solids and PCB to the watercourse, a suction dredger constructed specially for dredging with minimal spread of suspended solids was used. In addition, dredging of the most contaminated part of the lake was carried out within a protective geotextile screen. An extensive environmental monitoring program, running throughout the operation showed that the spread of suspended solids and PCB during dredging was limited. Monitoring of PCB also showed that the use of geotextile screens considerably reduced the exposure of the river water to PCB. Dewatering was carried out with filter presses and the surplus water was returned to Järnsjön after cleaning by flocculation, flotation and sedimentation. The landfill was covered with 1.2 m of sand and gravel from the site. No external hydraulic barriers were constructed. Instead, the relatively low permeability of the fine-grained and low-contaminated sediments in the topmost meter of the disposed sediments was considered sufficient, also taking into account the low mobility of PCB from the deposited sediments, as measured in the pilot study. Thus, the calculated sediments, as measured in the pilot study. Thus the calculated leakage from the landfill is small compared to the remaining transport of PCB in the river. During the first years of monitoring after construction, only minor leakage has been detected from the landfill.

INTRODUCTION
In the early 1990s, it was decided that remedial action should be carried out in order to protect Emån from the leakage of PCB from the contaminated sediments in Lake Järnsjön. Based on a feasibility study, a primary alternative was selected, which included hydraulic dredging, mechanical dewatering and disposal of the sediments in a landfill. Investigations of the presence and transport of PCB in the water system followed by detailed planning of the action were conducted on behalf of the Swedish Environmental Protection Agency during the years 1990-1992. During this phase, the municipality of Hultsfred and the Kalmar County Administration also participated in the work of the executive committee for the project. During the remediation program, the municipality of Hultsfred was responsible for the work. To carry out the project, a separate project organization with the municipality of Hultsfred as the client was built up at the end of 1991. The project planning was completed in mid-1992, a contractor was appointed at the end of the same year and the remedial actions were finally carried out during 1993 and 1994.

During the planning phase of the project, experiences from other sediment remediation projects abroad were surveyed. A number of removal technologies were available, but very few studies of the impact of dredging and the spreading of contaminants during the dredging process were found. Environment Canada started researching technologies in 1990 within the Contaminated Sediment Removal Programme (CSRP) with the aim of developing and demonstrating sediment removal technologies causing minimal adverse environmental impact. An outcome of the CSRP was the Welland River Contaminated Sediment Removal Demonstration carried out in November 1991. During this demonstration, approximately 230 m³ of industrial deposits and contaminated soft sediments were removed by a modified Mud Cat™ horizontal auger suction dredger from Ellicott International. This equipment was shown to be very efficient in removing soft sediments with a minimum loss, thereby ensuring low turbidity, low concentrations of suspended solids and low exposure of the river to contaminants ⁽¹⁾.

From a survey of ten field studies of dredging of contaminated sediments with different types of equipment, it was concluded that modified conventional dredging equipment may be used to handle contaminated material; however, in most cases special purpose dredgers are employed ⁽³⁾. Two factors were considered especially important to observe in selection of equipment for removal of contaminated sediments, a low rate of sediment resuspension during dredging to reduce secondary contamination, and a low water content in dredged sediments in order to reduce the volume of slurry and consequently the volume of excess water to be treated.

In some demonstration projects, silt curtains were used to enclose the area that was dredged, thereby preventing further spreading of resuspended sediments. The silt curtains generally behaved satisfactorily in preventing sediment transport, but occasional failures occurred due to wind and wind-generated currents ^(1,3).

HYDROLOGICAL AND GEOTECHNICAL CONDITIONS
Lake Järnsjön is a small, shallow lake in Sweden with an area of 25 ha and a typical depth of 1.2-1.8 m. A small part of the lake is deeper, the maximum depth being about 4.5 m (prior to dredging). The flow through the lake may vary from a few m³ s-1 up to about 100 m³ s-1. Subsequently, the total exchange of the water volume in the lake is rapid, and a typical rate is calculated to be about 8 to 12 hrs. Another consequence of the occasionally very high water flow is that the water level can vary within a 2-m range, which occasionally leads to flooding of the land area close to the river.

The sediments containing PCB were mainly characterized as very soft organic sediments (gyttja) mixed with mineral silty sediments ⁽⁴⁾. A significant part of the organics consisted of partly decomposed fibers containing PCB. This contaminated layer covered the whole bottom of the lake. In the major part of the lake, the thickness of the contaminated layer was less than 0.4 m, but within a smaller area the thickness was greater and extended up to 2 m. The contamination varied, with highly contaminated sediments within a limited area in the eastern part, and less-contaminated sediments in the rest of the lake. This area of the lake was also the deepest and contained the thickest layers of contaminated sediment.

The content of dry solids in the sediments was typically in the range 20-35%. In some areas, however, the soft sediments were mixed with sand, most probably an effect of sedimentation and erosion during varying flow velocities and, consequently the sediments in these areas showed significantly lower water contents. In minor areas, sand was the dominating element and a content of dry solids as high as 70% was detected.

The contaminated sediments were located on thick layers of fine-grained mineral sediments over glaciofluvial sand and gravel deposited along the river valley. Sand and gravel are exposed locally at the lakeside. The glaciofluvial formation is deposited on a moraine, which is the dominating soil in this part of Sweden. The principal geological conditions in the area are shown in Figure 1.

Figure 1. The principal geological conditions in the area

The lake is mainly surrounded by flat areas where the glaciofluvial deposits are covered with relatively thick layers of sedimentary deposits of silt and sand. In minor areas, the sediments are overlaid by peat. At varying distances from the lake, the ground level rises and the glaciofluvial deposit is exposed. Outside the glaciofluvial deposit, closer to the sides of the valley, the underlying moraine is exposed.

The landfill site is situated on glaciofluvial sand and gravel about 250 m west of the lake. At the western border of the site, the moraine is exposed.

REMEDIAL ACTION METHODS
Based on a feasibility study, including a review of the methods available for remedial actions of PCB-contaminated sediments, a primary alternative was selected, which included hydraulic dredging, mechanical dewatering and immediate disposal of the sediments in a landfill. The primary reasons for removal of the sediments were that no in situ treatment technology had proved successful for PCB-contaminated sediments and covering in situ was considered less favorable with respect to the soft character of the sediments, the shallow water and the risk of erosion.

One of the major obstacles that had to be dealt with regarding the chosen alternative was the risk of exposure of PCB when dredging contaminated sediments. The risk of serious spreading of PCB was obvious, as Järnsjön is a small lake and Emån creates a significant flow through the lake. Nevertheless, with respect to the large volume, it was necessary to maintain a reasonably high removal capacity during dredging. To minimize resuspension while maintaining high capacity, a specially designed hydraulic dredger for dredging contaminated material with minimal spillage was found to be the most suitable equipment. Although very few investigations of the amount of material spread during dredging with such equipment could be found at the time, the available information on common hydraulic dredgers showed that spillage should be considerably less than 1%. Based on this conservative figure and investigations of the settling properties of the material ⁽⁵⁾, it was estimated that dredging of the highly contaminated sediments without further protective measures involved a certain risk of unacceptable spreading of PCB to Emån downstream of Järnsjön. The possibilities of preventing the spread of contaminated suspended solids by dredging within a geotextile screen were thus investigated. These investigations were mainly based on studies in a 6 x 8 m² scale model of the lake built up in the water laboratory of the Royal Swedish Institute of Technology ⁽⁶⁾. In this model, the natural currents in the lake at different discharges were studied, as well as the effects on the currents of geotextile screen in a different positions, and spreading of suspended solids from the enclosed area to the river. Based on the investigations and theoretical calculations of suspension, transport and settling of the sediment particles, it was decided that the eastern part of the lake should be dredged within a protective geotextile screen, positioned in such a way that the heavily contaminated sediments were enclosed, as shown in Figure 2. It was also suggested that experience and measurements of the actual suspension of particles from the dredging of this area would form a basis for decisions on protective measures when dredging the western part of the lake.

Figure 2. Map of Lake Järnsjön showing the position of the geotextile screen used during dredging in 1993, and the location of the landfill.

During the feasibility study, different methods for disposal of the dredged sediments were investigated, after which landfilling was chosen. The arguments for choosing landfilling were primarily the relatively low degree of contamination and calculated slow release of PCB from the landfill, in combination with the economic advantages. Destruction or separation techniques were not found to be economically feasible due to the large volumes and characteristics of the sediments, which were fine-grained and had a high water content, and the limited resources available.

THE OPERATION
Hydraulic Dredging
Dredging was carried out with a Mud Cat™ suction dredger using an auger head designed especially by Ellicott to reduce the spread of suspended particles. The auger is 3.5 m long and 0.5 m in diameter, and is mounted in the longitudinal direction of the dredger (Fig. 3). During dredging, the auger is swung successively from right to left. The auger is divided into two separate parts which rotate in opposite directions and, thereby, transport sediment to the center of the head during the swing. In order to further decrease the suspension of particles, the auger is protected by a cap of adjustable steel plates. Normally, the auger will be exposed only in the direction of the swing, i.e. the cap will be closed to the surrounding water in order to protect spillage. 

Figure 3. Photo showing the Ellicott Mud Cat™ suction dredger with the auger elevated. Photo. T. Svahn.

In order to further reduce suspension of contaminated materials, disturbance from pipes, wires, service boats, anchoring, etc. were minimized by stringent restrictions. To control operation, the dredger is equipped with an advanced positioning system permitting high precision, with a vertical accuracy of about 10 cm and a horizontal accuracy of 5 cm. The position was determined by a land-based precision total station and a computer in which a terrain model from the previous echo-sounding had been stored. To maneuver the dredger, floating wires anchored on land were used. The high accuracy of the auger was considered important to completely remove sediments with PCB levels above the target and at the same time minimize mixing with uncontaminated sediments to be handled in the dewatering plant and disposed of in the landfill.

The dredger is designed primarily for soft sediments, which implies certain difficulties if denser material appears. In minor areas in the southern part of Järnsjön, the sediments were dominated by dense sand and gravel and could not be cut with the suction dredger. Instead, these sediments were cut with a bucket dredger. Patches of dense sand and gravel occurred both inside and outside the enclosed area, and buck dredging was therefore carried out both with and without the protection of a geotextile screen. The occurrence of sand layers mixed in the sediments caused difficulties also in some other areas, although it was possible to cut them with the Mud Cat™ suction dredger. However, dredging with the suction dredger in dense sediments required the addition of more water to the sediments and, consequently, a higher load on the dewatering plant. Similar problems were reported in the removal of contaminated sediments at Cold Spring in Putnam County, New York ⁽³⁾.

Dredging of the heavily contaminated sediments within the enclosed area (eastern part of the lake) was carried out by Milman during May to November 1993. The total thickness of the sediments to be dredged in this area varied from 0.4m to 1.6m and dredging was carried out in layers of 0.4m. At the end of this period, in November 1993, bucket dredging outside the protective screen was also carried out.

In order to reduce the risk to aquatic life in the river, due to spreading of suspended solids containing PCB, dredging was stopped during the most susceptible period, from December to April. Consequently, the western part of the lake was dredged in summer 1994. In this area, the sediments were of limited thickness and it was possible to limit dredging to a single layer with a thickness of 0.4m. When dredging was completed, a contracted volume of about 120 000 m³ had been dredged. Including "overcut", the dredged amount in reality was close to 150 000 m³. According to the follow-up, an amount of 394 kg of PCB was cut, which constitutes about 97% of the estimated total amount of PCB prior to the dredging operation. About 80% of the dredged amount of PCB originated from the enclosed area in the eastern part of the lake containing the heavily contaminated material. About 2.9 kg PCB was left in the areas in the remediation project. The major part of the remaining PCB (7.4 kg) is located at or near the lakeside, an area not included in the remediation. 

The Dredging Equipment Performed Very Well Limiting Resuspension of Solids.
During the dredging operation an extensive program for environmental monitoring was run ⁽⁷⁾. As part of this program, the turbidity and concentrations of suspended solids and PCB were measured upstream, within the enclosed area and downstream of the lake. When dredging heavily contaminated sediments, the measured concentrations of suspended solids in the water volume inside the enclosed area varied between 2 and 6 mg L^-1.  Upstream, the concentrations varied between 2 and 5 mg L^-1. And downstream the concentrations varied within the range 1-4 mg L^-1. Only occasionally was a higher concentration measured downstream compared to the upstream value. The generally lower values downstream reflect the fact that the lake acts as a sedimentation basin during normal discharges. The variations within the enclosed area showed no correlation with the intensity of the work, but were probably a function of the varying distance between the measuring station and the dredger. The generally low values that were measured show that the dredging equipment performed very well in limiting the suspension of solids. The spillage from dredging was estimated to be less than 0.5%. However, it should be noted, that in spite of the relatively low concentrations of suspended solids, fairly high concentrations of PCB were measured within the enclosed area.

Also during the dredging without protective screens in 1994, concentrations of suspended solids were generally higher upstream than downstream of the lake, emphasizing the Mud Cat's capability of cutting with minimum spillage. The mean concentration of suspended solids in 1994 was close to the mean concentration measured in 1993. Taking into account the discharge of water, it is estimated that the total transport of suspended solids in the river was notable higher in 1994, but that the reduction in the amount of suspended solids in the river during passage through the lake was more efficient. This was probably an effect of the closure of a considerable part of the lake during 1993, which decreased the effective wet area and thus created a faster current. Moreover, due to dredging in 1993, which produced greater depths and, consequently a larger water volume in the lake, even more favorable conditions for sedimentation in the lake had been created.

Geotextile Screen
Concentrations of PCB were measured at the same stations as the concentrations of suspended solids. It is difficult to quantify the influence of the dredging operation on the transport of PCB in the river, as this is determined to a great extent by other factors such as temperature and water discharge. However, from the recorded measurements of PCB it is suggested that the amount of transported PCB in Emån downstream of Lake Järnsjön was lower during dredging in 1993 than during dredging in 1994, in spite of the larger amounts of PCB that were removed as shown in Figure 5 ⁽⁸⁾. This probably reflects the fact that the protective geotextile screen was used in 1993, but not in 1994. The effect of the screen is also evident from the fact that PCB concentrations of up to 134 ng L^-1, with a mean value of about 60 ng L^-1 were recorded within the enclosed area. The corresponding concentrations upstream and downstream of the lake varied between 1-4 ng L^-1 and 10-15 ng L^-1, respectively. Thus, the concentrations of PCB recorded in the river were no higher during dredging than "normal" levels recorded prior to remediation. The geotextile screen was removed early in December 1993 without any considerable increase in the transport of suspended solids and PCB. When planning the procedure, there was some concern as to whether the removal would lead to a high peak in PCB transport downstream, as sediments that had been suspended during dredging, and which had only just settled, would be exposed to the current.

Bucket Dredging
During the dredging periods the concentrations of suspended solids downstream of the lake, only occasionally exceeded the concentrations upstream. From the measurements carried out over longer periods, it can be concluded that this was a normal and natural effect of a high discharge. However, an increase in the volume of suspended solids that left the lake was also observed during the period when bucket dredging as in progress outside the protective screen, and at the end of this period the transport from the lake was higher than the transport into the lake.

DEWATERING AND DISPOSAL OF SEDIMENTS
Landfill design
A localization study was carried out for the landfill, which resulted in field investigations of four possible sites. Based on the study, the landfill site was chosen and a detailed investigation carried out. The primary reason for the final choice was the proximity to the lake (250 m) and the direction of groundwater flow, which implied that no other recipient than Lake Järnsjön could be influenced by a possible leakage of PCB. The investigations showed that leachate from the landfill would be transported with superficial groundwater directed towards the lake.

In order to save volume and obtain material for the protective covering, sand and gravel were excavated at the site, so that the landfill bottom lies below the original ground level over most of the area. In the landfill, heavily contaminated sediments are separated from slightly contaminated sediments by a geotextile. The reason for this being to make it possible to localize the heavily contaminated sediments in the future, when remedial techniques have evolved, and it may be feasible to carry out more definite actions.

The topmost meter of the disposed sediments consists of the most fine-grained and least contaminated sediments from the lake. Calculations of the hydraulic conductivity from oedometer tests by means of CRS ⁽⁹⁾, performed on fine-grained sediments after dredging and dewatering, indicated that the hydraulic conductivity of these sediments, after consolidation for the pressure from the landfill cover, should be less than 5x10-9 m s^-1. This is not normally low enough to serve as a sealing layer for landfills, but was considered to reduce the leakage sufficiently with regard to the calculated low leakage of PCB. Finally, the landfill is covered with a 1.2 m protective cover of sand and gravel from the site. The total area occupied by the landfill is about 32,000 m², the maximum height is 6.5 m and the volume is about 90,000 m3.

To calculate possible leakage and concentrations of PCB from the landfill during planning, a lysimeter test was carried out using a lysimeter column with a diameter of 1 m. The column was filled with sediments with a concentration of PCB close to the mean value of the contaminated sediments. Parallel to this, concentrations of PCB in pore water from the most contaminated sediments were also determined. The investigations showed that PCB concentrations in leachate water from the landfill could be expected to vary between 400 and 2000 ng L^-1. Taking the water balance for the landfill into account, which was calculated according to an approximate model ⁽¹⁰⁾, the total leakage of PCB in the short term was estimated at between 1 and 10 g yr^-1. Theoretical modeling of the long-term behavior of the landfill showed similar results with a calculated release of less than 6 g yr^{-1 (11)}.

Dewatering
For dewatering, an alternative based on mechanical dewatering by filter presses was used in the set-up shown in Figure 7. Based on stability analyses, calculations of deformation and water balance, performed with data from laboratory investigations on dewatered sediments of shear strength, deformation properties, and hydraulic conductivity, it was suggested that the content of dry solids in the sediments had to be at least 36% (corresponding to a water content of less than 180%) before landfilling. This was necessary, primarily, for landfill stability.

Figure 7. A view over the dewatering plant. Dredged sludge was transported in a pipeline to one of two lagoons used for separation of coarse material (in the background). The sediment slurry was then discharged to the sludge basin (in the foreground). From the sludge basin, the sediment slurry was pumped to the filter presses for dewatering of sediments (in the middle). The dewatered sediment was transported from the presses to the landfill site by a conveyor belt (to the right) and the separated water was discharged to the water treatment plant (to the left) and finally back to the lake. Photo: T. Svahn

Certain difficulties were associated with dewatering of the sediments according to the requirements. The capacity appeared to be less than expected, which caused some delay compared to the initial time schedule. One major reason for this was that the dilution of sediments with water during dredging in some areas had to be much higher than calculated, due to the occurrence of dense sediments. The actual dilution with water varied between 1:2 and 1:10 and, thus, a total volume of more than 500 000 m³ had to be dewatered. Furthermore, for the fine-grained fraction of the sediments it was impossible to achieve the required water content with the type of filter presses used. This was partly a consequence of an effective separation of sand in a settling lagoon prior to the filter presses, which led to low hydraulic conductivity in the remaining sediments. In order to meet the requirement for water content, it was thus necessary to remix the dewatered fine-grained sediments with sand. This was not permissible for the topmost meter in the landfill, as this was intended to form a low-permeable layer. For this layer, however, stability analyses showed that a higher water content could be accepted, providing special precautions were taken, and the landfill was completed in accordance with these procedures.

The treatment of the water that emanated from dewatering of sediments and that was returned to the lake included flocculation, flotation, and sedimentation, in order to reduce the volume of suspended material returned to the lake. This was considered t be sufficient also with respect to PCB. During the remediation, about 0.2 kg of PCB was returned to the lake with returning water.

Landfill Behavior
During handling and landfilling, PCB concentrations in the air were elevated. Concentrations in air were determined by the volume of handled sediments and the temperature ⁽¹²⁾. The highest concentrations occurred above the sludge basin and in the center of the landfill. The median concentration above the sludge basin was 5.9 ng/m³, which was 45 times higher than the background level, but far below health limit values. Stations in the central part of the landfill to the NW showed an exponential decline in PCB concentrations with distance from the landfill. At a distance of 350 m from the center of the landfill, the concentrations had decreased by one order of magnitude. However, at a distance of 850 m from the landfill, about 5% of the elevated PCB level remained, which was significant compared to the reference station at a distance of 15 km. After the landfill had been closed and the contaminated sediments covered by uncontaminated soil, PCB levels decreased to reference levels.

So far, the follow-up indicates that the landfill is behaving satisfactorily. No increase in PCB concentration has been measured in groundwater under the landfill, compared to reference levels. Despite this, groundwater closer to the lake show slightly enhanced levels of PCB, in the range 1-4 ng L^-1. One explanation may be the seasonal flooding over time, which has probably involved deposition of PCB in these areas.

References:

1. Buchberger, C. 1993. Environment Canada Demonstrations. Remediation technologies for the removal of contaminated sediment in the Great Lakes. Terra et Aqua No. 50, January 1993.
2. Herbich, J.B. 1994. Removal of contaminated sediments: equipment and recent field studies. Dredging, Remediation and Containment of Contaminated Sediments. American Society for Testing and Materials ASTM STP1293, pp. 770-111.
3. Pelletier, J.P. 1994. Demonstrations and commercial applications of innovative sediment removal technologies. Dredging, remediation and Containment of Contaminated Sediments. American Society for Testing and Materials ASTM STP 1292, pp. 112-127.
4. Von Post, H. and Wik, O. 1992. Remediation of Lake Järnsjön in River Ema. Investigation of contaminated sediments. Swedish Environmental Protection Agency, Report 3998. (In Swedish, summary in English).
5. Nilsson, B. 1992. Remediation of Lake Järnsjön in River Ema. Estimations of erosion and settling properties of sediments. Swedish Environmental Protection Agency. Report 4115. (In Swedish, summary in English).
6. Kullbert, S. and Berg, H. 1992. Remediation of Lake Järnsjön in River Ema. Investigation of currents through Lake Järnsjön by studies in a physical model. Swedish Environmental Protection Agency, Report 4164. (In Swedish, summary in English).
7. Helgee, A., Troedsson, B., Elander, P. and Hammar, T. 1997. Remediation of Lake Järnsjön. Monitoring program. Swedish Environmental Protection Agency, Report 4746. (In Swedish, summary in English).
8. Bremle, G., Larsson, P., Hammar, T., Helgee, A. and Troedsson, B. 1998. PCB in a river system during sediment remediation. Water Air Soil Polut. (In press).
9. Larsson, R. and Sallfors, G. 1985 Automatic continuous consolidation testing in Sweden. ASTM Symposium on Consolidation of Soils: Testing and Evaluation: Fort Lauderdale, American Society for Testing and Materials ASTM STP 892. Consolidation Behaviour of Soil.
10. Lundgren, T. and Elander, P. 1986. Environmental control in disposal of residues from coal and peat combustion. Swedish Environmental Protection Agency, Report 3144, (In Swedish).
11. Elert, M. Hoglund, L-O and Lindgren , M. 1992. Release of PCB and mercury of fibre sediments. Swedish Environmental Protection Agency, Report 4076.
12. Bremle, G. and Larsson, P. 1998. PCB in the air during landfilling of a contaminated lake sediment. Atmos. Environ.

AMBIO: A Journal of the Human Environment Volume XXVII Number 5, August 1998

Pär Elander and Tommy Hammar