Miljøprojekt 898, Teknologiudviklingsprogrammet for jord- og grundvandsforurening – Vakuumventilering - erfaring med monitering og optimering af drift

Vacuum Extraction - Monitoring and Optimisation of Operation

Summary and Conclusions

Summary

Technology Development Project

The Danish Environmental Protection Agency has co-sponsored a Technology Development Project about vacuum extraction carried out by Ringkjøbing Amt. The aim of the project was to achieve documentation on the use of vacuum extraction as a soil remediation method and to gain knowledge about specific geological and physical/chemical limitations of the method when used in Denmark. These findings can later be used to form the basis of protocols for investigation, construction, operation, optimization, and post-remediation control of vacuum extraction projects.

The project was carried out at a site where soil and groundwater are heavily polluted with tetrachloroethylene arising from a former dry cleaners. The site is located in Ikast, Denmark. The geology of the site is very heterogeneous but it can generally be described as sand containing clay layers of varying thickness and at various depths. On the main part of the site there is a clay layer of 6.5 meters in thickness close to the surface level. The groundwater level of the upper aquifer is around 17 meters below surface level.

The vacuum extraction project is part of the site remediation scheme which also includes a groundwater pump-and-treat system that is operated at an average pumping rate of 4 m³/h. The vacuum extraction system comprises 6 extraction wells, each with one or two screens that can be operated with a flow rate of up to 100 m³/h per screen. During the project 7 different operating scenarios were tested with pauses between each. In order to investigate different aspects of the vacuum extraction method, each scenario had a specific set-up with regards to the screens that were used, the extraction rate from each of the screens, and in some cases the length of pause and operating periods during intermittent operation. Monitoring consisted of continuous, automatic GC-analysis from all the extraction screens, as well as manual sampling of soil gas and groundwater from a number of monitoring wells. The manual sampling took place before and after each scenario in order to evaluate the specific effect of each scenario. Towards the end of the project period a number of soil samples were collected in the vicinity of an extraction well situated in the most polluted area, with the aim of verifying the relationship between the pollution concentration in the soil gas and in the soil matrix.

Use of air-flow modeling and on-line sampling and measurements

The software program ModAir was used to create a model of the airflow in the soil during vacuum extraction. The model contributed to the conceptual understanding of the airflow in the soil and to the interpretation of the first scenarios. However, fluctuations in the groundwater level due to the operation of the groundwater extraction system made it impossible to achieve a correct simulation of the pressure distribution in the soil during vacuum extraction. On the other hand, this does not imply that air-flow models cannot be used successfully at other sites to gain knowledge of the air-flow patterns and to ensure a proper design of the vacuum extraction system. It is important to remember that building and validating a model requires a lot of exact measurements and if these are not available, or if they vary in quality, it might be better to base the design on more simple calculations together with qualitative assessments.

The GC-system for on-line monitoring of the pollution concentrations in the soil gas was used with good results to observe the development in concentrations in each screen during vacuum extraction and during pauses. There were, however, some measurement errors when the degree of dilution for a specific screen had to be changed to accommodate the change in concentration in the soil gas due to the vacuum extraction. Additionally, large uncertainties occured when comparing one screen with another, or when comparing measurements with more accurate laboratory methods. Therefore, future application of on-line measurements should rely on a measuring technique that is less sensitive than a GC-system with EC-detector.

Changes in pollution concentrations

The soil gas pollution concentrations quickly decreased during the first month of operation, after which the decrease continued at a slower rate. During the periods of pause following vacuum extraction, the concentrations rose again to a level below the initial concentration. This rebound effect seemed to continue for a long time, even after 6 months the concentrations were still rising almost linearly, especially in the hot-spot area. This observation was confirmed by monitoring carried out 15 months after the operation had ended. The average concentration was about 15% of the initial concentration in the hot-spot area and about 9% outside of this area.

Extracting from a single screen compared with several screens showed that by extracting from one screen alone a remediation effect can be achieved in a large area in this type of sandy soil. However, layers of clay or other inhomogeneities may result in local zones where the airflow is reduced, hence it is necessary to have several screens to be able to address these zones as well. The flow rate seemed less important for the remediation effect, for instance a doubling of the flow rate did not yield a significant increase in remediation effect. Prior to designing a vacuum extraction system it is thus very important to know the location of clay layers and other low-permeable soil layers in the whole area, as a soil-venting test only explores the radius of influence in the directions of the few wells used in the test.

Extraction from the deep screens located just above the groundwater level was more effective than extraction from the more shallow screens. This is because the extracted air is taken from the atmosphere through the soil down to each screen, hence both the soil at the surface level and the soil at deeper levels are influenced when extracting from the deep screens. This is not as significant when there are clay layers near the surface, but even here the project showed a remediation effect through the clay layer on the concentrations beneath a building. This effect is probably caused by fractures in the clay layer and not a uniform flow through the whole layer, as the monitoring of soil samples and soil gas before and after the operation of vacuum extraction showed that there was still a significant part of the initial pollution left in the clay layer, which indicates that only a small part of this layer had been influenced by the vacuum extraction.

The deep screens also showed a significant rebound effect after the end of the operating period, but this was at a slower rate than in the screens near surface in the hot-spot area. The rebound effect in the deep wells could be caused by the concentration in the soil gas being in equilibrium with the concentration in the groundwater. This is partly confirmed by calculations of the theoretical equilibrium, although only in about half of the measurements. The reason for this could be that the equilibrium between groundwater and soil gas is only complete after several years. On the other hand, the concentrations in selected soil samples in the hot-spot area showed a good correspondence with the theoretical concentrations obtained from the equilibrium with the soil gas concentrations in this area. A reason for this could be that in this case the equilibrium is mainly achieved through the process of desorption from the sand to the soil air, whereas the equilibrium with the pollution in the groundwater also involves the process of diffusion from the groundwater level and up through the soil above.

In cases like this, where the groundwater pollution has an effect on the rebound of the pollution in the soil gas, a reduction of the groundwater pollution becomes necessary to achieve a lasting remediation effect from the vacuum extraction. In the present case, the pump-and-treat system reduced the groundwater pollution to about 10% of the start level, and the groundwater extraction will probably have to continue for many years to avoid a rebound of the pollution level in the groundwater and thereby also in the soil gas.

The vacuum extraction reduced the pollution beneath the building from 52 mg/m³ to less than 0.2 mg/m³. There was no rebound effect immediately after the end of the operating period but after 15 months the concentration had risen to 9.2 mg/m³. Vacuum extraction thus reduced, but did not eliminate, the potential problem of soil gas pollution endangering the indoor climate in the building.

Prediction of pollution removal

An empirical model to predict pollution removal was developed on the basis of the changes in average concentrations that were measured during operating and pause periods. To some extent the model could be used to predict the total amount of pollution removed by vacuum extraction in each scenario. The model was constructed around an exponential function that was calibrated according to the measured concentration sequences. However, the model could only fit the exact concentration curve in the scenarios if it was recalibrated with the initial average concentration of each scenario. The reason for this might be that the pauses in this project were generally too short to reach an equilibrium between soil, groundwater and soil gas. However, the necessity of re-calibration before modeling a new scenario is contradictory to the aim of the model, namely to use it to forecast the remediation effect of any given sequence of operation and pause periods on the basis of an initial soil venting test.

It might be possible to improve the model to cope with the non-stationary conditions that prevail during short operation periods and pauses. This should preferably be done using data from other sites so that the model could also be adjusted for more general use.

Economic model for optimization of remediation operation

The appendix shows how the empirical model for predicting the remediation effect can be extended to an economic model for optimization of the remediation, taking into account pollution removal, time necessary to reach the remediation goal, and operating costs of one or several consecutive operating scenarios. The economic model should only be regarded as a prototype because it is highly dependent on the ability to predict the pollution removal, and this ability is still lacking in some aspects. However, the economic model does show how the operating costs depend on the flow rate from each filter and the length of operating and pause periods.

Conclusions

Preliminary investigations and selection of remediation technology

In relation to the selection and design of vacuum extraction as a remediation technology it was found that:

Thorough mapping of clay layers within the remediation area is important in order to place the venting wells and screens in a way that ensures proper spreading of the airflow in the soil.
A prolonged vacuum extraction field test yields information about how the concentrations of pollution in the soil air behave during operation and pause. This can be used to design an effective strategy of operating the vacuum extraction system and to predict to some degree the pollution removal and the changes in concentration during the remediation process.
An airflow model can contribute to designing the placement of the extraction wells and screens and the extraction rates from each screen. The construction and validation of such a model, however, demands a lot of time and effort and in each case it should therefore be considered if the model can be replaced with more simple calculations and qualitative assessments.
The tools for calculation of the pollution equilibrium between soil, groundwater, and soil gas should be further verified in order establish a basis for realistic assessments of the total pollution mass and how it is distributed. This is an important question for the selection of vacuum extraction as part of a remediation strategy.

Construction, operation, and optimization of remediation facilities

With regard to construction, operation and optimization of vacuum extraction the project shows that:

The effectiveness of vacuum extraction in sandy soils is reduced when one or more clay layers are present in the sand. This is due to the slow desorption of pollution from the clay and to the fact that the air does not really flow through the clay. Furthermore, desorption and diffusion of pollution from the groundwater and any possible free phase pollution also slows down the remediation process. The result is that the remediation takes much longer than could be expected from the pollution concentrations in the soil gas alone. It also implies that there is an advantage in operating the vacuum extraction with short operating periods followed by long pause periods, perhaps even using a mobile vacuum extraction unit that can be transferred to other sites when not in use at the specific site.
The indoor climate in a building near the hot-spot did not seem to be influenced by pollution transport from the groundwater because after an operating pause of 9 months there was no rebound in the pollution concentration under the building. But after 15 months of pause a rebound of 18% of the starting concentration was observed. However, the observed rebound effect might just as well be caused by soil pollution in the unsaturated zone. At other sites, where there are no clay layers between the groundwater and the surface and where the groundwater level is closer to the surface, the groundwater pollution probably has an effect on the concentrations under nearby buildings.

Termination of remediation

With regard to the post-remediation control, i.e. measurements taken after the vacuum extraction has ended to verify the level of pollution left in the soil, the project points to the following:

It is necessary to have at least 1-2 years of pause after terminating the vacuum extraction before the soil gas measurements can provide useful information about the level of pollution left in the soil. An alternative method could be to use actual measured concentrations in the groundwater and the soil to calculate the theoretical concentrations in the soil gas. If the measured concentrations in the soil gas are lower than the calculated concentrations, more rebound effect in the soil gas concentrations is to be expected.