J. L. Steimke
Westinghouse Savannah River Company
Aiken, SC 29808

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy.

Available for sale to the public, in paper, from:  U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161,  phone: (800) 553-6847,  fax: (703) 605-6900,  email:  orders@ntis.fedworld.gov   online ordering:  http://www.ntis.gov/support/ordering.htm

Available electronically at  http://www.osti.gov/bridge/

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062,  phone: (865 ) 576-8401,  fax: (865) 576-5728,  email:  reports@adonis.osti.gov

A full-scale, transparent replica of a GeoSiphon was constructed in the TFL to test a new concept, using a solar powered vacuum pump to remove accumulated gases from the air chamber. It did not have a treatment cell containing iron filings as do the actual TNX GeoSiphons in the field, but it was accurate in all other respects. The gas generation that is observed in an actual GeoSiphon was simulated by air injection at the inlet of the TFL GeoSiphon. After facility shakedown, three stages of testing were conducted: verification testing, parametric testing and long term testing. In verification testing, the TFL GeoSiphon was used to reproduce a particular test at TNX in which the water flowrate decreased gradually as the result of air accumulation at the crest of a siphon without an air chamber. For this test the vacuum pump was not used and the air chamber was initially filled with air rather than water. Agreement between data from the TNX GeoSiphon and the TFL GeoSiphon was good, which gave confidence that the TFL GeoSiphon was a good hydraulic representation of the TNX GeoSiphon. For the remaining tests, the solar powered vacuum pump and air chamber were used. In parametric testing, steady state runs were made for water flowrates ranging from 1 gpm to 19 gpm, air injection rates ranging from 0 to 77 standard cc/min and outfall line angles ranging from vertical to 60° from vertical. In all cases, the air chamber and vacuum pump removed nearly all of the air and the GeoSiphon operated without problems. In long term testing, the GeoSiphon was allowed to run continuously for 21 days at one set of conditions. During this time the solar cell kept the storage battery fully charged at all times and the control circuit for the vacuum pump operated reliably. The solar panel was observed to have a large excess capacity when used with the vacuum pump. With two changes, the concept of using a solar powered vacuum pump attached to an air chamber should be ready for long term use in the field. Those changes are to insulate the air chamber of the GeoSiphon so it will not freeze in the winter and to make the tank from steel rather than transparent plastic.

The GeoSiphon is a groundwater remediation technology developed at SRS [1,2,3]. As configured at TNX, the TNX GeoSiphon Cells were built in a pre-siphon treatment cell configuration (i.e., treatment occurs prior to water transport within the siphon). The TNX cells are large diameter wells, containing granular cast iron treatment medium in place of gravel pack. Contaminated groundwater flow through the granular cast iron in the treatment cells is induced by use of a siphon from the cells to the X-08 outfall ditch. The siphon must be primed prior to operation. After priming, flow is induced by the natural hydraulic head difference between the cells and the X-08 outfall ditch. The granular cast iron reduces the trichloroethylene to ethane, ethene, and chloride ions. Treated water is subsequently discharged through the X-19 outfall to the Savannah River. This eliminates the requirement for a mechanical pump and a source of power. Environmental Restoration Department constructed a GeoSiphon at TNX and tested several different siphon line configurations. Initially, the siphon line was laid on the surface of the ground. Gas generated by the reaction of water and chlorinated solvents with the iron filings accumulated in high points in the siphon line and broke the siphon. Later, the siphon was laid on engineered grades and an air chamber was added to the high point in the line. During repeated trials the siphon action failed in less than a day because of excessive gas accumulation in the line or in the air chamber. However, the GeoSiphon appeared to run indefinitely when a person periodically removed the gas from the air chamber.

TFL proposed a solution to the problem, which was to automatically remove gas from the air chamber using a solar powered vacuum pump. TFL also proposed building a full-scale facility to test the solution before it was used in the field. This report documents that work.

The work was conducted in compliance with QAP 2-3, "Control of Research and Development Activities". The work was initiated by a Task Technical Request [5], planned with a Task Plan [6] and conducted according to written procedure [7]. Calibrated instruments and a laboratory notebook were used.

The test facility is shown in Figure 1 and is documented in Drawing EES-22747-M0-001. There are two tanks, a "well" tank and an "outfall" tank. The water level in the well tank was controlled using an overflow pipe. The siphon enters the well tank from the top. Air was injected into the siphon at a controlled flowrate, which was measured by a rotameter. The upward leg of the siphon line consisted of the following:

An ultrasonic flowmeter was located in the second clear PVC line to measure water flowrate. The end of the clear PVC pipe, which was also its highest point, was connected to a tee whose upward and downward arms connected to the air chamber and to the downward leg of the siphon line, respectively. The downward leg of the siphon line covered an elevation change of 10 feet so its length was 10/cos(q ) where q is the angle between the downward leg of the siphon line and vertical. There was a two-foot long piece of flexible hose in the downward leg of the siphon line so that its angle with respect to vertical could be changed. The remainder of the downward leg of the siphon line consisted of 2" schedule 40 clear PVC pipe. The downward leg of the siphon line emptied into the outfall tank. The air chamber was constructed of 6" schedule 40 clear PVC pipe. The level in the outfall tank was adjusted by adding or removing water. A ¾ hp Teel pump recycled water from the outfall tank to the well tank so that the TFL GeoSiphon could run indefinitely.

A solar panel, Solarex model #MMLBAB000, was mounted on the outside of the TFL pointing west so that it would receive only afternoon sun. The solar panel was connected to a Guardian Heavy Duty Deep Cycle storage battery. The wire connecting the solar panel to the battery contained an electrical resistor to simulate the 1200 feet of double strand wire, that are estimated to span the distance from the solar panel to the GeoSiphon at TNX.

A small vacuum pump, model N05ANI by KNF Neuberger, Inc., was connected to the top of the air chamber to periodically remove air. The operation of the vacuum pump was controlled by a simple circuit that was connected to two float switches in the tank. The circuit is shown in Drawing EES-22747-M0-001. The float switches, Electra Switch by Compac Engineering Inc., part #15-650-PP, were mounted near the top and bottom of the air chamber. When the siphon was first started up, the air chamber contained air but no water. Because neither float switch sensed water, the control circuit energized the vacuum pump. As air was pumped out of the air chamber the water level rose, first past the lower float switch and then past the upper float switch. When both mechanical switches detected water, the control circuit turned off the vacuum pump. The control circuit did not re-energize the vacuum pump again until both float switches detected the absence of water. A toggle switch on the front of the control box allowed the vacuum pump to be temporarily disabled. Valve V5 (see Figure 1) was occasionally opened to allow additional air to vent to the air chamber, which was under a partial vacuum.

Two differential pressure gages were connected to the piping. The first gage measured the differential pressure between the well tank and the outfall tank, D H₁. This differential pressure is defined as the driving head for the siphon. The reading of this differential pressure gage was checked periodically by subtracting the water levels in the well tank and outfall tank. The water levels in the tanks were measured using yardsticks. The second gage measured the differential pressure between the top and bottom of the outfall line, D H₂. The tubing connecting the second gage to the outfall line was filled with water. When the downward leg of the siphon was completely filled with water, the two water legs balanced and the gage reading reflected only the relatively small frictional loss in the outfall line. Void fraction accumulating in the downward leg of the siphon resulted in an increasing differential pressure at the transducer. The vertical elevation of the downward leg of the siphon was 120 inches. For example, a reading of 12 inches of water on the second gage implied a void fraction of approximately 10%.

There were two modes of operation for the siphon, controlled head or controlled flow. For controlled head mode an overflow pipe having the desired length was inserted in the well tank. Increasing the length of the overflow pipe increased DH₁. For controlled head mode, valve V3 was opened fully and V1 was opened sufficiently to assure overflow. For controlled flow valve V3 was closed and V1 was adjusted to achieve the desired flow. Valve V2 was always open unless pump maintenance was required.

Note: The TR numbers that appear in the listing are TFL instrument identification numbers.

a. Water flow in GeoSiphon. Magnetic flowmeter, TR-03278
b. Siphon driving head, DH1. Differential pressure transducer, TR-03604
c. Water level in the well tank. Yardstick
d. Water level in the outfall tank. Yardstick
e. Water level in the air chamber. Yardstick
f. Pressure drop in downward leg of the siphon from the high point to the point of discharge, DH2. Differential pressure transducer, TR-03536.
g. Air flow into inlet of the siphon. Sho-Rate rotameter
h. Current from solar panel to battery. 1 ohm shunt. Voltage across shunt directly measured by data acquisition system (DAS).
i. Battery voltage. Directly measured by DAS.
j. Voltage to vacuum pump. This indicated when and for how long the vacuum pump was operating. Directly measured by DAS.
k. Current to vacuum pump. Multimeter
l. Temperature in well tank. Thermocouple, TR-1866
m. All data except the vacuum pump current, the airflow and the three water levels were recorded on a DAS at a rate of once per minute. The reading of the airflow rotameter and the three water levels was recorded periodically and manually in a lab book. The current consumed by the vacuum pump was measured and recorded in the lab book. Descriptions of the density and shape of the air bubbles in the discharge line were recorded in the lab book.
n. All instruments were calibrated in the TFL.

In the TFL GeoSiphon air was used to simulate the dissolved gases present in water which was treated by the TNX GeoSiphon Cell and which subsequently degassed within the siphon at TNX. The volume of air was estimated from the fact that the 7.5 gallon air chamber at TNX had to be refilled with water about once a day, equivalent to a flowrate of 20 actual cc/min or 13 scc/min. The chamber pressure was about 2/3 atm. The TFL GeoSiphon was tested at airflows as high as 76 scc/min. The air rotameter was calibrated by bubbling air into an inverted volumetric flask and measuring the time required to displace the water from the flask.

Facility shakedown was performed to find and correct leaks and to verify that all of the instruments were operating correctly. Shakedown was performed according to a written procedure [6]

The symbols in the equations in this section are listed below.

The basic equation relating flows, elevations and pressures in the GeoSiphon was derived from the fact that the pressure at the high point in the GeoSiphon line, P_h, may be calculated approaching that point from either the well tank or the outfall tank.

(1)

Similarly the pressure at the high point may be calculated approaching the high point from the outfall tank.

(2)

Setting equations 1 and 2 equal and solving for water velocity gives the following equation.

(3)

For a GeoSiphon, the miscellaneous losses are frequently negligible compared with the frictional losses in the piping. Note that z₁-z₂ is equal to D H₁. Also, the void fraction in the upward piping was negligible for the airflows used during testing. The void fraction in the downward piping was larger than upward void fraction. The reason that void tends to accumulate in the downflow piping is because the upward rise velocity of air bubbles is approximately the same as the downward water velocity. Neglecting miscellaneous losses and upward void fraction allows for simplifications.

(4)

Note that water velocity increases with increasing values of D H₁, and decreases with increasing void fraction in the outfall line.

Equation 3 was rearranged to allow computation of D H₁ when all other variables were measured and void fraction was negligible.

(5)

The purpose of Verification Testing was to verify that the TFL GeoSiphon behaved hydraulically nearly the same as the TNX GeoSiphon. Figure 8 in WSRC-TR-99-00250 gave the results of a Minimum Flushing Velocity Test in the 2-Inch Siphon Line. The initial flowrate of 6 gpm decreased to 1 gpm over a period of four hours and the driving head for the siphon, D H₁, increased from 0.1 foot to 2.9 feet over the same period. The driving head for the siphon, DH₁, is the water level in the well casing minus the water level at the siphon discharge. The reason that the water flow decreased is that air accumulated in the downward leg of the siphon, increasing a ₂ in equation 4. The reason that the siphon driving head increased is the result of the specific capacity of the well, 0.005 cfs/ft or 2.24 gpm/ft. The initial flowrate of 6 gpm required a head difference from the surface of the groundwater to the water level in the well casing, DH3, of about 6/2.24 or 2.7 feet. As the water flow decreased the well head difference decreased. Over the period of the test the groundwater level was nearly constant so the level in the casing increased. This increased the siphon driving head, DH₁.

The goal of the TFL GeoSiphon Verification Test was to reproduce that TNX test. The discharge pipe was inclined at an angle of 30 degrees from vertical to match the TNX test. The specific capacity of the well is an important feature of the TNX GeoSiphon that was not simulated in the TFL GeoSiphon. Therefore, the TFL GeoSiphon was run in the following way, which accounted for the specific capacity of the well. Air flowrates of 17 and 36.5 scc/min were set for different runs. The vacuum pump was used to partially fill the air chamber with water and then disabled. Constant flow mode was used and the valve on the discharge of the pump, which recycled water from the discharge tank to the well casing tank, was adjusted to give a flow of 6 gpm. Air was bled into the top of the air chamber until the water level was even with the top of the siphon pipe. Then the head difference between the casing and the discharge tank was recorded. Air accumulated in the siphon downward leg because the water velocity for 6 gpm in a 2 inch pipe, 0.6 ft/sec, was insufficient to sweep out air bubbles, which have a typical rise velocity of 0.8 ft/sec. The accumulation of air caused the flow to decrease. The incremental flow reduction was divided by the specific capacity of the well to compute the decrease in the well driving head, DH3. That decrease in head corresponded to an increase in the head available to drive the siphon, DH1. The valve on the outlet of the recirculation pump was then adjusted to give a new appropriately larger siphon driving head, DH₁. The process was repeated until the flow had decreased to about 1 gpm.

The results for the two runs at airflows of 36.5 and 17 scc/min are shown in Figures 2 and 3. Figure 2 shows that for a TFL airflow of 36.5 scc/min, the TFL and TNX water flowrates decreased at the same rate. The driving head increased somewhat faster in the TNX test than in the TFL test. Figure 3 shows that for a TFL airflow of 17 scc/min, the TFL and TNX water flowrates decreased at about the same rate and the driving head increased faster in the TNX test. The verification was judged successful because the velocity and driving head transients for the TNX siphon and the TFL siphon are comparable. They were not expected to be exactly the same because the TNX gas flow is not known, only estimated and because the effect of well capacity was manually simulated.

All parametric tests listed in Table 1 plus a few additional conditions were run in controlled flow mode with the vacuum pump enabled. The air chamber effectively removed nearly all air bubbles. Each set of conditions was run until the instruments indicated that hydraulic steady state had been achieved. Note that the airflows are in standard cc/min where standard conditions are 1 atmosphere and 20° C. Some of the GeoSiphon piping is at a significantly lower pressure so that the actual air volume is larger.

Tabular data from the Parametric Testing are given in Table 2. DH1 and DH2 are the differential pressures in inches of water measured by the two differential pressure gages. H_w and H_o and the levels measured in the well tank and the outfall tank, respectively. H_w-H_o is the arithmetic difference between those two levels. The superficial velocities of water and air were computed by dividing the volumetric flows of water and air, respectively, by the cross-sectional area of the pipe. The significance of water superficial velocity is that a typical bubble rise velocity is 0.8 ft/sec. Note that most of the water superficial velocities are less than that. Therefore, water flowrates less than about 12 gpm are not expected to be effective for flushing air bubbles out of the downflow pipe. For horizontal sections of pipe or tubing the ratio of air superficial velocity to water superficial velocity is approximately equal to the void fraction. Therefore, the maximum expected void fraction in horizontal or nearly horizontal pipe was generally much less than 1%. The Reynolds number is listed in Table 2 and is defined by equation 6.

(6)

Reynolds numbers less than 2000 are in the laminar flow regime, Reynolds numbers greater than 3000 are turbulent and intermediate Reynolds numbers are in the transition zone. Water flows of 1 gpm are laminar and all other flows tested are turbulent.

Figure 4 is a check on the differential pressure gage and compares the output of the gage with the differential head computed by subtracting the measured levels of the two tanks. The two measurement methods agree well. Figure 5 plots D H₁ and D H₂ for all cases including air injection rates ranging from zero to 76 scc/min. Both differential heads tended to increase with increasing water flowrate, but D H₂ was always small, never exceeding 1.2 inches of water. Since the overall height of the siphon downward leg was 120" the maximum void fraction was 1%.

The predicted driving head was computed from the parametric data using equation 5. Equation 5 requires the total number of velocity heads loss for fittings. The Crane Manual [8] was used to estimate that loss at (1.5+90 f) velocity heads, where f is the friction factor. Friction factors for pipe with a roughness factor of 0.005 were also taken from the Crane Manual. The tubing is slightly corrugated and it is difficult to relate the amplitude of the corrugation to conventional pipe roughness. Measured and calculated driving heads are plotted in Figure 6 versus water flowrate. Agreement is good.

Long Term Flow Testing was conducted from February 4 to February 25, 2000. The hardware was configured in controlled head mode, resulting in a flow of about 6 gpm. The air injection rate was set at 36.5 scc/min and the downward leg was set to an angle of 30° from vertical. Everything ran well over this period. Figure 7 plots water flowrate and the differential head between the well tank and the outfall tank, DH₁ for the first three days, which were typical of the 21 day run. The sawtooth variation is the result of accumulation and release of water from the air chamber. Referring to the abscissa of Figure 7, from 0.05 day to 0.4 day air was bubbling into the chamber displacing water. The displaced water raised the level in the outfall tank while the water level in the well tank was controlled and therefore, could not change. As a result the head difference between the two tanks decreased and the flow decreased. This sawtooth variation would not occur in an actual GeoSiphon. The average time required to displace all of the water from the chamber with air was 8.5 hours which is in good agreement with calculated time to accomplish this, 8.0 hours, at the air addition rate and the sub-atmospheric pressure in the chamber. Figure 8 plots battery voltage, pump voltage and solar cell current. The battery voltage was nearly constant and decreased about 0.2 volts when the vacuum pump was energized. The pump voltage shows when the vacuum pump was energized. A period of about ten minutes was required to remove the air from the chamber. During that period the pump consumed 13 watts. The solar cell has a controller that allows it to send current to the battery when the battery voltage is less than a 12.2 volt setpoint. The solar cell sends a current of 0.5 amperes only briefly and, of course, only during daylight hours. Recall that abscissa values on Figure 8 of 0, 1.0 and 2.0 days all correspond to noon. The pump current was measured separately and was 1.06 amps.

1. The TFL GeoSiphon satisfactorily duplicated the hydraulic characteristics of the TNX GeoSiphon.

2. Over the wide range of conditions for which testing was performed, the combination of the gas accumulation chamber and the solar powered vacuum pump effectively removed almost all of the air flowing in the GeoSiphon.

- siphon downward leg angle ranging from 0° to 60° from vertical
- airflows ranging from 0 to 77 scc/min
- water flows ranging from 1 to 19 gpm

3. When the vacuum pump was used only a small number of bubbles flowed past the chamber to the outfall line. The void fraction was visually estimated to be much less than 1%.

4. The solar panel provided much more electrical power than was needed by the vacuum pump. Therefore, the battery remained fully charged at all times.

5. The chamber for the TNX GeoSiphon should be insulated with fiberglass insulation to provide freeze protection.

1. M. A. Phifer, F. C. Sappington and M. E. Denham, "TNX GeoSiphon Cell (TGSC-1) Phase I, Deployment / Demonstration Final Report", WSRC-TR-98-00032, February 27, 1998.

2. M. A. Phifer, F. C. Sappington, R. L. Nichols and K. L. Dixon, "TNX GeoSiphon Cell (TGSC-1) Phase II Single Cell, Deployment / Demonstration Final Report", WSRC-TR-98-00032, January 12, 1999.

3. M. A. Phifer, F. C. Sappington and R. L. Nichols, "TNX GeoSiphon Cell (TGSC-1) Phase II Minimum Flushing Velocity, Deployment / Demonstration Final Report", WSRC-TR-99-00250, July 30, 1999.

4. Cassandra Bayer, "TNX GeoSiphon Support and Laboratory Mockup", ERE-027, October 5, 1999.

5. Steimke, J. L., "Task Technical and QA Plan: GeoSiphon Testing", SRT-ETF-990023, October 20, 1999.

6. Steimke, J. L, "GeoSiphon Operating Procedure", Rev. 0, FP-827, January 18, 2000.

7. Crane Co., Flow of Fluids through Valves, Fittings and Pipe, 1991.

Figure 1 Schematic of TFL GeoSiphon

Figure 2

Figure 3

Figure 4

Figure 6

Figure 7

Figure 8

Table 1 Operating Conditions for Parametric Testing