"Performance Testing of Conventional and Innovative Downhole Samplers and Pumps for VOC's in a Laboratory Monitoring Well."

by:

D.F. Baerg R.C. Starr J.A. Cherry D.J.A. Smyth

Waterloo Centre for Groundwater Research

University of Waterloo

Waterloo, Ontario

N2L 3G1

ABSTRACT

Concentrations of groundwater contaminants in samples that differ from concentrations in the aquifer can lead to incorrect conclusions about the extent and severity of contamination and the effectiveness of remedial measures. Sample collection and storage techniques affect measured concentrations in samples, particularly of organic compounds that are subject to loss by sorption or volatilization. Both commercially available VOC samplers and prototypes still under development were evaluated under controlled conditions in a laboratory setting using a vertical stainless steel pipe located in a stairwell, which simulates a monitoring well. Samplers were lowered inside the pipe to the bottom and activated to collect samples, while control samples were collected from a stopcock at the base of the well. Comparison of concentrations in samples collected from the VOC samplers to concentrations in the control samples was used as a measure of the quality of samples provided by various VOC samplers.

INTRODUCTION

This paper describes preliminary results of an ongoing project. Several commercially available and prototype downhole water samplers are being tested under laboratory conditions to evaluate their ability to provide samples for VOC analysis that are representative of the water from which the samples were collected. The laboratory facility includes a simulated well that consists of a 5 3/4 inch (14.6 cm) diameter stainless steel pipe 32 feet (9.75 m) tall, which is located in a stairwell. The simulated well is completely above ground, and the stairs provide convenient access to the entire vertical height of the well. The well was filled to about 3.5 metres below the top with simulated groundwater spiked with a mixture of volatile organic compounds. Water samples were collected by lowering samplers to the bottom of the laboratory well and operating the sampler as if the simulated well was an actual well in the field. These samples were compared against a suite of control samples collected from the bottom of the simulated well by opening a sample port, which allowed water to flow out of the well due to gravity and through stainless steel tubing to an EPA standard 40 ml glass vial. A sampling head allowed the vial to be filled and then flushed with any desired volume of water, which was not exposed to a gas phase after initial filling of the vial.

The mean concentration of the control samples was taken as the true concentration in the well, and a difference between the mean concentration of the control set and the downhole water sampler data set indicates that the sampler produces samples that are not representative of the water being sampled.

The major loss mechanisms for dissolved VOCs from groundwater samples are thought to be sorption onto organic components of samplers and storage containers, such as plastic or rubber, and volatilization into the gas phase. Volatilization can occur whenever the aqueous sample is exposed to a gas phase, such as bubbles or a head space in a sampler or sample container, or exposed to the atmosphere while being transferred from a sampler to a storage container

To evaluate the potential for losses by these mechanisms, the water used in the laboratory experiments was spiked with compounds that have a range of tendency to sorb or volatilize. The octanol - water partitioning coefficient, Kow, is a measure of the tendency of a compound to be sorbed by organic materials, with higher values corresponding to greater tendency to sorb. The Henry's Law constant describes the tendency of a compound to partition from the aqueous phase to the gas phase, with higher values corresponding to a greater tendency to partition into the gas phase. Table 1 lists the values of the octanol - water partitioning coefficient and the Henry's law constant for the compounds used in the laboratory work.

Concentrations in the spiked water in the well were in a convenient range for analysis to maximize the precision and accuracy of the analytical results, and were well above drinking water limits (Table 2).

Several commercially available and prototype samplers have been evaluated to date in this project, and others are currently being evaluated. The samplers evaluated and reported here are listed in Table 3

Both the Solinst and Norwell bailers have check valves at the top and bottom to facilitate collecting a sample from a discrete depth. Water is transferred from the bailers into a separate sample container by opening a valve at the bottom of the bailer and directing a stream of water into a sample vial, which exposes the water to the atmosphere. The Solinst VOA trap sampler is basically a syringe that is lowered to the desired sampling depth. The piston retracts in the cylinder due to the hydrostatic pressure in the well as the syringe fills with water, and a check valve in the piston allows additional water to flush through the syringe cylinder after the piston is fully retracted. The sampler is withdrawn from the well and the syringe is transported to the laboratory for analysis, without transferring the sample into a separate container. The canister sampler, a prototype developed at the University of Waterloo, is designed for use in narrow diameter multilevel sampling systems or conventional wells, and requires a diameter of only one half inch. It consists of a length of stainless steel pipe fitted with manually operated valves at each end. The valves are opened, and a separate check valve is attached to the bottom. The canister is lowered to the desired depth in the well, the check valve is opened to allow water to fill the canister and continue to flush it after initial filling. The check valve is closed, the sampler is removed from the well, and the manual valves are closed. The check valve is removed and the sealed canister is shipped to the laboratory for analysis. The Well Wizard is a bladder pump, the operating principles of which are well known and are not repeated here. The Solinst Double Valve Pump is somewhat similar in operation to a bladder pump except that there is no bladder. Hence, there is some contact between the water sample and a gas phase. The peristaltic pump and sampling head system consists of a sampling head that allows a glass vial to be placed in line on the suction (upstream) side of a peristaltic pump. A suction hose is lowered to the desired sampling depth in the well, and any required volume of water can be pumped through the vial to flush it without exposing the water to a headspace, after the vial is initially filled. The inertial pump is the WaTerra Pump. The operating principles are well known and not repeated here. The procedure for filling sample vials recommended by the supplier was followed, and involves placing a Teflon tube into the polyethylene riser pipe to direct water from the riser pipe into sample vials to minimize exposure to the atmosphere.

RESULTS AND DISCUSSION

Several experiments were conducted, with two downhole samplers compared against control samplers in each experiment. In each case, control samples were collected either at the same time as or alternating with the downhole sampler samples. Preliminary results are shown in Tables 4 - 11. Concentrations of DCE and MeCl are given in mg/L, and all others in ug/L. Columns 2 and 3 give the mean concentrations in the control samples and in samples collected using the downhole samplers. Column 4 is the ratio of the sampler mean concentration to the control case mean concentration, expressed as a percentage.

The samples collected with a downhole sampler are a subset of a population, and the control samples are a subset of a second population. If a downhole sampler operates properly, the mean concentration of the downhole sampler population is the same as the mean concentration of the control population. Column 5 lists the probability that the mean concentration of the downhole sampler population is the same as that of the control case population. If the probability is less than 0.05, i.e. 5 percent, then in at least 95 times out of 100 one would be correct in claiming that the means of the two populations are actually different.

Several observations can be made about the data. First, inspection of column 4 shows that 7 of 8 samplers produced samples with a mean concentration within 12 percent of the mean concentration on the control data set. This level of accuracy is sufficient for many practical problems, regardless of whether there is a statistically significant difference between sampler and control case mean concentrations. In spite of this, the mean Concentration of downhole sampler data sets is significantly different from that of the control case data sets for some samplers. Although the stainless steel bailer data set has mean concentrations that are greater than 95 percent of the mean control concentration, the two data sets are significantly different for 4 of 5 compounds. The Teflon bailer produced samples with a mean concentration greater than 96.9 percent of the mean control concentrations, and the mean concentrations were different for only 1 of 5 compounds.

The VOA trap sampler produced samples that generally had higher concentrations than the control samples, and the concentrations were significantly different from the controls for 4 of 5 compounds. The canister sampler also produced samples that had higher concentrations than the controls, and that were significantly different from the control samples for all five compounds. It is surprising to observe concentrations in the downhole samples that are greater than the control samples, and additional work is being conducted to determine if this finding is repeatable, and if so, the cause. Also note that fewer replicates were used in this experiment than the 15 used in evaluations of other samplers. Only 10 controls and 10 replicate samples were collected using the canister sampler and 5 using the VOA trap sampler. Fewer replicates were used in these cases because there was insufficient equipment to collect fifteen replicates.

The Well Wizard produced samples that did not differ from the control samples for 4 of 5 compounds. The double valve sampler produced samples that did not differ significantly from the control case samples.

Mean concentrations of samples collected with the peristaltic pump plus sampling head were significantly different from the mean control concentrations for all compounds. The WaTerra pump data set means were less than 89 percent of the control case means and were significantly different from the control case means for 4 of 5 compounds. One might conclude from this that this equipment is not suitable for VOC sampling. However, it must be emphasized that the riser tube used in this evaluation was polyethylene, which has a large tendency to sorb organic solutes and hence is not the ideal material for this application. Additional testing will be conducted using Teflon tubing instead of polyethylene in the near future, and it is expected that the performance will be better.

CONCLUSIONS

Seven of the eight samplers evaluated produce samples whose mean concentration is within 12 percent of the mean concentrations of the control case, which is suitable for many practical situations. Based on the criterion of statistically different mean concentrations, the samplers evaluated here that have the best performance are the Teflon bailer, double valve sampler, and the Well Wizard.

Additional work is underway to determine if sampler performance can be improved by modifying sampling and handling techniques, using different materials (such as Teflon instead of polyethylene), or utilizing ancillary equipment, such as sampling heads for transferring water between samplers and storage containers. Samplers other than the ones described here are being evaluated, and developers of new equipment are encouraged to test their devices in this program.

Solinst.com > Text Section > Resources and Information

"Performance Testing of Conventional and Innovative Downhole Samplers and Pumps for VOC's in a Laboratory Monitoring Well."

by:

D.F. Baerg R.C. Starr J.A. Cherry D.J.A. Smyth

Waterloo Centre for Groundwater Research

University of Waterloo

Waterloo, Ontario

N2L 3G1

ABSTRACT

INTRODUCTION

Concentrations in the spiked water in the well were in a convenient range for analysis to maximize the precision and accuracy of the analytical results, and were well above drinking water limits (Table 2).

Several commercially available and prototype samplers have been evaluated to date in this project, and others are currently being evaluated. The samplers evaluated and reported here are listed in Table 3

RESULTS AND DISCUSSION

The Well Wizard produced samples that did not differ from the control samples for 4 of 5 compounds. The double valve sampler produced samples that did not differ significantly from the control case samples.

CONCLUSIONS