Oprensning af klorerede opløsningsmidler ved dampstripning.

Summary and conclusions

Steam stripping is a remediation technique in which the injection of either steam or a mixture of steam and air is combined with extracting pore air and/or groundwater from the soil.

If only steam is injected, a front of condensate forms in the soil so that first cold water, then warmer water and, finally, steam, washes through the contaminated soil, the steam taking the contaminant with it up to the condensation front. Thus the front may contain concentrations up to saturation of the substances present. Steam can be injected both below and above the groundwater table. If only steam is injected, most of the contaminant will be either dissolved in the water phase or found in free phase in the condensation front.

If a mixture of steam and air is injected, a condensation front is formed in the same way as when only steam is injected. However, the front is not quite as well-defined as in steam-only injection because the non-condensable part of the air injected distributes the heat in the front more evenly. If mixtures of steam and air are injected, contaminant will constantly be removed from the front by the non-condensable air injected. For this reason, most of the pollutant is removed in gaseous form if air is mixed with the steam injected into the soil. The injection of steam/air mixtures into the unsaturated zone has been tested in a pilot-scale remediation in Germany. Injecting non-condensable gases below the water table results in a buoyancy effect which may prevent or hinder usage of this method, depending on the geology. Adding air to the steam injected into the soil reduces the amount of energy which can be added to the soil per time unit.

The rate at which the steam spreads in the soil depends on the permeability of the soil. If there are small differences in permeability, the steam front generally spreads at a uniform speed at the entire injection height of the filter: the heat conduction is fast enough relative to the speed of the steam front to equalize the temperature vertically. If there are wide variations in the permeability of the soil, the steam and thus the heat will spread where the permeability is greatest.

Effect of Temperature Increase

The effects of the temperature increase on the chlorinated solvents are a higher steam pressure, a higher Henry’s Law constant, a lower sorption constant, and a slightly higher solubility, along with increasing diffusion constants and greater free-phase mobility (reduced viscosity). In addition, if the soil matrix is heated to the boiling point, "steam drive" may occur, i.e. steam is generated in the soil, flowing from hot low-permeable areas which are boiling inside into more highly permeable areas in which an actual flow of air occurs as a result of the vacuum extraction process. This means it is important to heat the entire contaminated area to the boiling point in order to achieve this effect. Full utilization of the steam drive effect can be accomplished by cycling the injection of steam: alternating between the injection of pure steam, i.e. heating the soil to the boiling point, and the application of as much vacuum as possible to extract substances from the less permeable areas and thus overcome the kinetic limitations caused by diffusion.

It has been documented that when the soil matrix is heated to about 100°C and oxygen is present in the soil, a hydrous pyrolysis oxidation of the chlorinated solvents and other organic compounds in the soil occurs. The half-life for degradation of these compounds is in the range of 5-10 dg^-1.

Identified Problems

A review of the literature on steam stripping as a remedial technique revealed a number of potential problems with using this method. The contamination may spread in a horizontal direction if steam is injected from the middle of the contaminated area out towards its edges. This is especially true if only steam is injected. If a mixture of steam and air is injected, it is vital to apply a vacuum strong enough to keep the contaminants from spreading outside the system, either towards the surface or horizontally away from the contaminated area. The optimal well placement is thus injecting from uncontaminated soil in towards the contaminated area.

The increased mobility of free phases of chlorinated solvents and water may sometimes cause an increased risk of vertical spreading of the contaminants. One analysis of the amount of energy necessary to heat up the soil showed that usually about 75 kg steam/m³ soil is required. This means that the thickness of the condensation front is typically 8-12 % of its distance from the well, providing that the condensate does not move downwards. Since there is a set relationship between the amount of soil and the thickness of the condensation front, the elevated concentration of contaminant in the front can be estimated. From this can be concluded that the concentration of contaminant in the front is four times higher than in the soil that is heated. The mobility of water, PCE, TCE, DCE and 1,1,1 TCA increases about the same amount (3-4 times) in the interval between 10°C and the boiling point of the mixture. With TCM, there is a somewhat stronger effect: its mobility increases up to sixfold, or twice as much as the mobility of water increases. This indicates that if soil concentrations exist that are about one-fourth or more of the residual saturation, there is a potential risk of that free-phase contaminant will move downwards in the soil. This is true in the case of most solvents. One exception is TCM, for which the critical value is one-eighth of the residual saturation. The residual saturation for chlorinated solvents is 1000-100,000 mg/kg depending on soil type, how much of the solvent entered the soil and how fast, and the properties of the solvent.

The difference between the densities of water and the solvent – which is used to compute the pressure head necessary for the penetration of free phase into water-filled cracks – also decreases with the temperature, which reduces the risk of penetration into cracks. Since the injection of only steam gradually creates saturation conditions for water in the condensation front, a vertical percolation of water downwards could also occur. This water would contain elevated levels of solvent and could lead to an increased impact on aquifers deeper down if not extracted.

By mixing air with the steam, this increase in concentration could be avoided completely when the mixture is injected into the unsaturated zone, and a reduction of the elevated concentrations of dissolved solvent in the front would be an added bonus.

In addition to these problems, there are a number of unanswered questions about short-term biological effects and possible impact on the soil mechanics. These subjects are not dealt with in this report.

The Brüel og Kjær Site

The remediation of the Brüel og Kjær site by steam stripping took place between the summer of 1997 and the spring of 1998; final documentation of the remediation was carried out at the end of 1998. Prior to this remediation project, an test was run with steam injection for a period of about 14 days in the early spring of 1997.

Geology and Hydrogeology

The geology of the site consists of alluvial sediments deposited in a braided river system. The particle size distribution varies from silt to fine sand over short distances, both vertical and horizontal. The thermal properties of the sediment at typical heat capacities of 0.4-0.6 kWh/m³ °C and at the water saturations at which the samples were taken can be found in the literature. Coefficients of thermal conductivity on the order of 0.3-2.1 W/mK have were found the lowest values found in the driest and coarsest material and the highest values in the moistest and most finely grained sediment. The specific heat values found correspond to 55 kWh/m³ being necessary to heat the soil to 100°C, corresponding to about 75 kg damp/m³.

There is a regional groundwater reservoir about 15 mbgs (meters below ground surface) in the sand layer. It extends down to 60 mbgs, where the limestone lies either directly under the sand or under a thin layer of moraine clay. Upstream from the contamination at the Brüel & Kjær site is a contamination of unknown origin with TCE and PCE at levels of about 500 m g/L being transported under the site.

Contamination Situation

The contamination consisted of a mixture of TCE and PCE from degreasing operations, a leaking sewer system, and surface spills. When wells were drilled for the remediation project, soil samples were taken at every other meter, which meant there were a total of 137 soil samples subsequently chemically analyzed for solvents. The average concentration was about 5.5 mg/kg in total, with PCE clearly the dominant component (83%). The maximum concentration found was about 500 mg/kg in a single sample. Most of the samples contained solvents at the m g/kg level. The contamination was estimated to be dispersed in about 12,000 m³ of soil with a hot spot about 3300 m³ in size extending down to a depth of 15 meters.

Dimensioning

The extent of the remediation project in various respects was determined on the basis of a preliminary test with steam injection in the middle of the probable hot spot and three extraction wells located at the edge. No water was extracted to check for possible downward loss of contaminant. On the basis of soil samples from before and after this test, it was estimated that steam stripping remediation would be an effective solution to the soil contamination problem at the Brüel & Kjær site. The purpose of the cleanup was to be able to cancel the site’s classification as a waste disposal site.

Prior to the test and before the remediation system was installed, no determinations were made of the permeability or the thermal properties of the sediment. This was done as a part of the technology project after the actual installation of the remediation system. Besides determination of these three parameters, the technology program also included a closer study of temperatures, pressures and concentrations of the chlorinated solvents in three monitoring wells with filters at three depths and located at different distances from a steam injection well.

Permeability determinations revealed both vertical and horizontal differences. To a certain extent, this was reflected in the spread of the steam: it was observed that the segment with the lowest permeability did not reach steam temperatures, in spite of the fact that the steam penetrated the segments above and below it.

The source of steam was a rented oil-fired steam generator connected to five different injection wells at different times, with steam being injected from only one well at a time. The treatment system consisted of a manifold with a flow meter and regulating valves, three parallel vacuum pumps, with downstream heat exchangers and then cyclones to separate condensate and "dry" gas. The cooling water for the heat exchangers was recirculated through industrial cooling units. The condensate was piped into tanks later taken to the Kommunekemi hazardous waste treatment facility. The gaseous phase was put through two alternating granular activated carbon filters with on-site regeneration. The condensate from the filter regeneration was put in pallet tanks, also for transport to Kommunekemi. The cleaned air was emitted into the atmosphere from a high stack.

Operations

Operations consisted of a number of phases. In the first phase (72 hours), only air was used to ventilate the soil, i.e. in a traditional vacuum extraction process. In Phases 2-6, steam was injected into five different areas: a total of about 700 metric tons of steam was injected over 29 days, corresponding to 511,000 kWh of energy. The final phase consisted of traditional vacuum extraction again. The seven phases had a total duration of just under five months. In the first phase, 38 kg of solvent was removed, corresponding to an average extraction rate of 40 g/h per well. In the Phases 2-6, 505 kg of solvent was extracted, corresponding to an average extraction rate of 140 g/h per well. In the last phase, 275 kg of organic compounds were extracted at an average extraction rate of 8 g/h per well. The maximum concentrations observed in the extracted air were as high as 50,000 mg/m³, which corresponds to an extraction rate of over 3 kg/h per well. The vacuum flow in the individual wells was made to vary between 0 and 100 m³/h, depending on when in the five months of the extraction process it was measured. A total of 830 kg of solvent was extracted in the remediation. An additional 600 kg was extracted in connection with the injection test, making the total amount extracted between 1400 and 1500 kg.

Remaining Contamination

The soil gas just below the ground surface was tested before, during and after remediation. It was observed that concentration of contaminants increased dramatically in the soil gas when the soil was heated, due to insufficient vacuum in these areas. The concentration of contaminants in the soil gas was generally reduced by one or two orders of magnitude, but it must also be noted that, after the remediation, the contamination was spread out over a larger area than before. The contaminant concentrations remaining in the soil gas, estimated on the basis of average values from two rounds of testing, are presumed to be in the range of 0.03-185 mg/m³, with an average value of 23 mg/m³. By comparison, the starting values were 0-400 mg/m³, with an average of not quite 50 mg/m³. It must be emphasized that these tests were performed after the injection test, which removed several hundred kilos of solvent. In previous investigations, levels of 100-1000 mg/m³ were found in a large area.

We attempted to assess the remaining concentration of solvents in the soil on the basis of computations of the phase distribution from soil gas concentrations in the filters M1-M3. We estimated that the soil near these filters still contains levels of up to 500 m g/kg dry matter. There are probably higher concentrations locally, where the steam fronts from the different injection wells met.

Conclusions

On the basis of the literature review and the remediation performed, the following conclusions can be drawn about steam stripping as a remediation technique:

• Steam stripping is a simple way to heat up soil with a sufficiently high permeability. The steam spreads along high-permeable zones and may through heat conduction penetrate less permeable areas.
• Heat conduction is a relatively slow process compared to the transport of steam by flow. Heating a three-meter-thick body of soil to the boiling point using heat conduction requires the injection of steam around the area for a period of about one month. At the Brüel & Kjær site, the injections in the various areas had a typical duration of one week; as a result, some parts of each area did not reach steam temperature.
• Steam stripping is an extremely effective way to accelerate a remediation process. In this case, a rise in the extraction rate from 200 g/h to 3000 g/h was observed: an increase of more than ten times in areas where there was full steam penetration.
• Concentrations in the condensation front may become so high that free phase compounds may move in a vertical direction. For this to happen, there must be high concentrations of the substance in the soil prior to remediation, presumably at least 1000 mg/kg before the concentration is high enough to allow the free phase to become mobile. According to our computations, TCM has the greatest potential for vertical mobilization. With the concentrations found, however, we estimate there was no great risk of vertical transport of free phase during the remediation of the Brüel & Kjær site.
• The best placement of steam injection wells is injection from the edges of a contaminated area in towards the center of the area, combined with extraction of both air and water from the center. This keeps the contaminant from spreading away from the extraction area. At the Brüel & Kjær site, steam was injected into the center of the contaminated area out towards extraction wells at the edges – i.e. the opposite strategy. No groundwater was pumped up from the aquifer below the center of the treated area. Groundwater was extracted from existing remedial wells next to the area, to the east. This resulted in a minor spread of the contamination, which can be seen from the elevated levels of contaminant in soil gas below ground surface found after termination of the remediation project, in areas that were less contaminated, just as there was presumably also some transport of contaminant into the groundwater below.
• It is possible to control to some extent the phase in which the contaminant is when extracted by mixing air with the injection steam. Two things are to be gained by this: 1) higher levels of contaminant in the condensation front are avoided and 2) the contaminant is transformed into its gaseous phase, which makes it easier to remove by vacuum extraction. With the injection of steam only, most of the contaminant remains in the water phase and must then diffuse into the gaseous phase before it can be transported away by vacuum extraction. Below the water table, the first choice for injection would not usually be air mixed with steam. After breakthrough into the extraction wells of the air/steam mixture, then only steam is injected so that the soil reaches the boiling point. This technique was not used at the Brüel & Kjær site, which meant that after the heating process had stopped, there was a relatively long phase of vacuum extraction in which there were kinetic limitations to the extraction rate. Since water was not extracted from a well in the center of the contaminated area during the remediation, we cannot estimate with any precision the amount of dissolved contaminant that was transported downwards. Of the solvent extracted from the soil in the vacuum extraction process, not quite 25% was removed in the condensation of the water in the cooler and 75% in the activated carbon filter.
• Cyclic steam injection can be used with good results once the soil has been heated to the boiling point. In this technique, steam injection is alternated with the application of vacuum so that the soil "boils" when the pressure drops. The steam produced can function as a transport route for contaminant trapped in less permeable areas of the soil and any contaminant that may remain in the condensation front. This technique was not used at the Brüel & Kjær site, which was part of the reason why there was a relatively long period of vacuum extraction after heating was concluded.
• When steam was injected into deeper-lying areas, there was only a modest loss of heat into the surroundings. At the Brüel & Kjær site, it was found that about 5% of the energy injected during the heating stage was lost to the surface. The loss was typically 1-5 kWh/m³ per month, depending on the temperature gradient. By comparison, with the heat capacities found, it was necessary to use 55 kWh/m³ to heat the soil up to 100°C. Heat removal during the extraction process was 1-5 kWh/m³ of soil per month, i.e. about the same as the passive loss into the surroundings.
• The preliminary assessment of the extent of the contamination was an underestimate. Before the remediation started, 137 soil samples were taken, which is a relatively high number. The average content of chlorinated solvents was 4.5 mg/kg. Remediation removed a total of about 1400-1500 kg. With a total contaminated soil volume of about 12,000 m³, the amount extracted corresponds to an average concentration of about 70 mg/kg, or more than ten times the amount estimated, in spite of the large number of samples analyzed.