A New Leak Detection System for Long-Distance Pipelines Utilizing Soil-Gas Techniques

Transcription

1 A New Leak Detection System for Long-Distance Pipelines Utilizing Soil-Gas Techniques by K.F. Liang and M.C. Tom Kuo Abstract This paper presents a new leak detection system for a long-distance pipeline using soil-gas techniques. The new leak detection system runs underground along the nearby pipeline and consists of intermittent porous tubes connected in series with impermeable polyethylene tubes. Applying the new detection system to long-distance pipelines for leak detection minimizes the large number of soil-gas samples and analyses that are required when conventional soil-gas probes are employed. A mathematical model was developed for designing the distribution of airflow in the new detection system. Field leak tests were conducted using both conventional soil-gas probes and the new system. Results indicated that the effective detection distance of the new system was at least 30 m, while the effective detection radius from the conventional soil-gas probes is only ~5 m. Introduction Soil-gas surveying refers to the analysis of the soil-air phase as a means to delineate subsurface contamination from volatile organic compounds (VOCs) such as solvents and petroleum products. Soil-gas techniques use a samplecollecting device that is installed beneath the ground surface. The presence of VOCs in soil gas indicates that there is contamination from the observed compounds either in the vadose zone near the device or in the ground water below the device (Kerfoot and Mayer 1986; Marrin and Thompson 1987; Thompson and Marrin 1987; Marrin and Kerfoot 1988; Wolfe and Williams 2002). The aromatic and aliphatic compounds of gasoline and volatile organic compounds have been identified as good candidate compounds for soil-gas analysis (Marrin 1985; Spittler et al. 1985). Chemical analysis of soil gases has also been used to monitor fuel leaks from underground storage tanks and pipelines (Marrin and Thompson 1987; Thompson and Marrin 1987; Marrin 1988; Kerfoot et al. 1988; Marrin and Kerfoot 1988). Conventional sampling techniques used in soil-gas surveying fall into two categories: passive sampling and dynamic grab sampling. Passive sampling uses an adsorbent to trap contaminants that diffuse through the soil gases. Passive adsorbent samplers are buried in the shallow soil for a period of days to weeks. The adsorbent is then retrieved from the soil and transported to a laboratory where desorption and chemical analyses are performed (Kerfoot Copyright ª 2006 The Author(s) Journal compilation ª 2006 National Ground Water Association. and Mayer 1986; Voorhees et al. 1984). Dynamic grab sampling requires the installation of probe or soil boring in the vadose zone followed by the withdrawal of soil gas with a pump (Marrin and Thompson 1987; Thompson and Marrin 1987; Spittler et al. 1985; Swallow and Gschwend 1983). Grab samples typically are analyzed on-site either with mobile laboratories or with portable instruments to provide real-time data. Applying conventional soil-gas techniques to long-distance pipelines for leak detection usually requires a large number of soil-gas samples and analyses. The purpose of this paper is to present a new idea and design to improve soil-gas dynamic grab sampling techniques for monitoring leakage from a long-distance pipeline. The new leak detection system consists of intermittent porous tubes connected in series with impermeable polyethylene tubes and runs along the nearby pipeline. Field leak tests were conducted to compare the distance of detection for conventional soil-gas probes vs. the new leak detection system. A mathematical model was developed to predict the distribution of airflow in the new leak detection system for each porous tube at various distances from the extraction end. The model was verified with experimental data obtained in air extraction tests. The verified model can be useful in designing the new leak detection system to monitor leakage from a long-distance pipeline. Description of New Leak Detection System The new leak detection system consists of intermittent porous tubes connected in series with impermeable Ground Water Monitoring & Remediation 26, no. 3/ Summer 2006/pages

2 polyethylene tubes and runs along the nearby pipeline (Kuo 2002). Figure 1 shows the schematic diagram of a new leak detection system installed along a long-distance pipeline. The new leak detection system is configured to obtain composite samples of the soil gas flowing through the intermittent porous tubes located at various distances from the extraction pump. The mode of operation of the new system is intended to be an intermittent monitoring system (i.e., daily or weekly sampling). Pasteris et al. (2002) reported that maximum concentrations of VOCs were found in soil-gas samples within the first 24 hours of fuel leakage. All fuel compounds except methyl tert-butyl ether in soil-gas samples disappeared below detection limits within 70 d. To achieve the best performance of the new leak detection system, daily monitoring is recommended. Monitoring frequency should be at least once a week. Figure 1 also shows that the leak detection system is installed above and nearby the pipeline in the same trench, which is backfilled with coarse sand. The permeability of backfill using coarse sand is in the order of 10 3 Darcy, which is much greater than that of the porous tubes in the order of 1 Darcy. The high permeability of backfill makes the vapor of leakage fuel to be extracted easily to the nearest porous tube of the leak detection system. In addition, the high-permeability backfill allows sufficient air to flow through each porous tube during soil-gas extraction. Figure 2 shows the basic section unit of a new leak detection system. Each section consists of one polyethylene tube followed by one porous tube. The specifications of materials used in this study for constructing the new leak detection system are described as follows. Each porous tube is 5 cm long with outside and inside diameters equal to 1.8 and 1.2 cm, respectively. Two types of porous tubes with permeability of 0.9 and 3 Darcy, respectively, were used in this study to assemble new leak detection systems. Each polyethylene tube is 100 cm long with outside and inside diameters equal to 1.2 and 0.9 cm, respectively. (a) A L h Q p P out 2R Q p Pin P w Q p -Q por (b) r w A A r e Poiseuille law were used to describe the flow behavior of soil gas through the porous tubes and inside the polyethylene tubes, respectively, in the mathematical model. The main objective of mathematical modeling was to calculate the distribution of airflow for each porous tube in the new leak detection system during soil-gas extraction. Model calculations start from the first section near the extraction pump and then proceed section by section toward upstream. Figure 2a shows that each section consists of one polyethylene tube followed by one porous tube. The airflow inside a polyethylene tube is simulated by the Hagen-Poiseuille law as follows (Bird et al. 1960): P w Q por A P atm Figure 2. Basic section unit of a new leak detection system. (a) Basic section unit made of polyethylene tube and porous tube. (b) Cross section of a porous tube. Modeling of Airflow Distribution for New Detection System Soil-gas flow in a new leak detection system falls into two categories of physical processes, i.e., soil-gas flow from soil atmosphere through the porous tubes and soil-gas flow inside the polyethylene tubes. Darcy s law and Hagen- Extraction pump FID Ground surface Trench (coarse sand backfill) Pipeline Natural soils Polyethylene tube Figure 1. Schematic diagram of a new leak detection system installed along a long-distance pipeline. Q pe 5 3: p P 2 in P2 out R 4 P b ll where Q pe is the airflow rate in a polyethylene tube (L/ min), P in is the inlet pressure of a polyethylene tube (atm absolute), P out is the outlet pressure of a polyethylene tube (atm absolute), l is the viscosity of soil gas at 20 C (cp), P b is the reference pressure ¼ 1 atm absolute, R is the inside radius of a polyethylene tube (m), and L is the length of a polyethylene tube (m). For the first polyethylene tube, P out and Q pe are also the pressure and total flow rate observed at the inlet of the extraction pump. After the previous calculation is finished in a polyethylene tube, the next step is to simulate the airflow from the atmosphere through the next porous tube. Figure 2b shows the cross section of a porous tube. The inlet pressure of a polyethylene tube, P in, is assumed equal to the inside pressure of the next upstream porous tube, P w. The airflow from atmosphere through a porous tube is simulated by Darcy s law as follows (Amyx et al. 1960): (1) 54 K.F. Liang and M.C.T. Kuo/ Ground Water Monitoring & Remediation 26, no. 3: 53 59

3 Q por 5 6:0 pkh P 2 atm P2 w re (2) lp b ln r w where Q por is the air rate flowing through a permeable porous tube (L/min), k is the permeability of a porous tube (Darcy), h is the length of a porous tube (m), l is the viscosity of soil gas at 20 C (cp), P atm is ambient pressure ¼ 1 atm absolute, P w is the inside pressure of a porous tube (atm absolute), P b is the reference pressure ¼ 1 atm absolute, r e is the outside radius of a porous tube (m), and r w is the inside radius of a porous tube (m). After finishing the calculations for one section, the simulation continues to the next section toward upstream. The airflow rate inside the polyethylene tube of the next section is then calculated as follows: Q pe; next section 5 Q pe Q por (3) In addition, the outlet pressure of the polyethylene tube of the next section is calculated as follows (Liang 2003): P out; next section 5 P w 1 Q por Q pe ðp atm P w Þ (4) Simulation of airflow distribution in the new system is accomplished by repeating the calculations for every section using Equations 1 to 4. The distribution of airflow during soil-gas extraction was calculated for the 12- and 30-m-long systems using Equations 1 to 4. The parameters used in the model calculations are as follows: P b ¼ 1 atm absolute, l ¼ 0.07 cp, L ¼ 1m,R ¼ m, r e ¼ m, and r w ¼ m. For the 12-m-long system, the permeability of porous tubes, the total airflow rate, and the extraction pump pressure are as follows: k ¼ 3 Darcy, Q total ¼ 114 L/min, and P out ¼ 0.73 atm absolute. For the 30-m-long unit, the permeability of porous tubes, the total airflow rate, and the extraction pump pressure are as follows: k ¼ 0.9 Darcy, Q total ¼ 35 L/min, and P out ¼ 0.91 atm absolute. The calculated distributions of airflow for the 12- and 30-m-long systems are shown in Figure 4. Furthermore, Figure 4 also compares the calculated airflow distributions with experimental data measured in air extraction tests. The verified model of airflow distribution would be useful in designing the new leak detection system, such as the permeability of porous tubes, the total airflow rate, and the extraction pump pressure. Air Extraction Tests for New Leak Detection System Figure 3 shows the schematic diagram of air extraction experiments. During the air extraction tests, one end of the detection system is sealed and the other end is equipped Flow meter Polyethylene tube Flow meter Polyethylene tube Extraction pump Figure 3. Schematic diagram of air extraction tests for a new leak detection system. with an extraction pump. The extraction tests were conducted with an extraction pump. The airflow rates were measured between the neighboring two porous tubes at the midpoints of polyethylene tubes. The main objective of the air extraction tests was to measure the distribution of airflow through each porous tube in the new leak detection system. The percentage of airflow through the last porous tube located at the far end away from the extraction pump is critical to determine the effective distance of detection for the new system. Two new systems were used in the air extraction experiments. One was a 12-m-long new leak detection system with total airflow rate ¼ 114 L/min, extraction pressure ¼ 0.73 atm absolute, and permeability of porous tubes ¼ 3 Darcy. The other was a 30-m-long new leak detection system with total airflow rate ¼ 35 L/min, extraction pressure ¼ 0.91 atm absolute, and permeability of porous tubes ¼ 0.9 Darcy. Figure 4 presents the distribution of airflow measured for the 12-m system and the 30-m system. Figure 4 indicates that the percentage of airflow through each porous tube decreases with an increase in the distance between the porous tube and the extraction pump. The air rate decreases from 41.0 L/min flowing through the 1st porous tube to 0.1 L/min flowing through the 12th porous tube for the 12-m system. The air rate decreases from 5 L/min flowing through the 1st porous tube to 0.2 L/min flowing through the 30th porous tube for the 30-m system. It is also evident that the distance of detection for the new system increases as the permeability of porous tubes decreases (Figure 4). The percentage of air flowing through the 12th porous tube is 0.087% for the 12-m system with high-permeability porous tubes (3 Darcy). For the 30-m system with low-permeability porous tubes (0.9 Darcy), the percentage of air flowing through the 30th porous tube is 0.6%. Leak Detection Test Using New Leak Detection System A field leak test was conducted using a 30-m-long new detection system installed at a depth of 1 m below ground surface. The main objective of leak detection tests was to demonstrate that the approximate location of a simulated leakage could be determined from the measured breakthrough time using the airflow distribution model presented previously. The simulated leakage was a source of 3 L gasoline released at a depth of 1 m below ground surface near the end of the 30-m-long new detection system. The K.F. Liang and M.C.T. Kuo/ Ground Water Monitoring & Remediation 26, no. 3:

4 Percentage of air flow through individual porous tube ( ) model prediction (3-Darcy system) 30 observation (3-Darcy system) 25 model prediction (0.9-Darcy system) 20 observation (0.9-Darcy system) Distance of individual porous tube from the extraction point (m) Figure 4. Distributions of airflow in a 12-m-long, new leak detection system (total airflow rate = 114 L/min, extraction pressure = 0.73 atm absolute, permeability of porous tubes = 3 Darcy) and a 30-m-long new leak detection system (total airflow rate = 35 L/min, extraction pressure = 0.91 atm absolute, permeability of porous tubes = 0.9 Darcy). permeability of the porous tubes used in the 30-m-long new detection system was 0.9 Darcy. An extraction pump with a total flow rate of 7 L/min and a portable FID detector (Foxboro TVA 1000 model) were used in the leak detection test. Figure 5 presents the results of the leak detection test with a breakthrough curve. The results demonstrated that the simulated leakage of 3 L gasoline located at 30 m away from the extraction point could be detected by the new detection system. The peak concentration at 428 ppm was 1430 times higher than the background concentration at 0.3 ppm. With the enhancement in the effective distance of detection, the new method significantly reduces the large amount of soil-gas samples and analyses that are required when conventional soil-gas probes are applied to detect a leakage along a long-distance pipeline. In addition, the breakthrough curve in Figure 5 shows a breakthrough time of 19 min, which could be related to the distance of 30 m between the leakage source and the extraction pump. The mathematical model verified in the air extraction tests can be used to calculate the percentage of airflow for each porous tube in the 30-m-long, new leak detection system. Table 1 shows the airflow rate predicted for each Concentration of total hydrocarbon (ppm) Extraction time (min) Figure 5. Breakthrough curve of total soil-gas hydrocarbons measured in leak detection test using a 30-m-long new leak detection system. porous tube, Q por,i. Assume that a leakage source is located nearby section K of a new leak detection system. Section k is any section between the extraction point and section K. The airflow rate inside any section k, Q pe,k, can be calculated by summing up the air rates flowing through all the porous tubes starting from any section k upstream toward section n. Section n is the end section away from the extraction point. Q pe;k 5 Xn i5k Q por;i (5) where Q por,i is the airflow rate in porous tube i (L/min), Q pe,k is the airflow rate in polyethylene tube k (L/min), and n is the total number of porous tubes. The travel time of the contaminant in section k can be calculated as follows: t k A kl k Q pe;k (6) where t k is the travel time of the contaminant in section k (min), A k is the cross-sectional area of section k (m 2 ), and L k is the length of section k (m). Assume that a leakage source is located nearby section K of a new leak detection system. The contaminant breakthrough time for a leakage source located nearby section K can be calculated by summing up the all the travel times from section 1 to section K as follows: t BT 5 XK k 51 t k (7) Use the leak detection experiment for example. The leakage source is located nearby section 30, i.e., K ¼ 30. Tables 1 and 2 show the calculations of airflow distribution and breakthrough time, respectively. The airflow distribution 56 K.F. Liang and M.C.T. Kuo/ Ground Water Monitoring & Remediation 26, no. 3: 53 59

5 Table 1 Distribution of Airflow Predicted in a 30-m-Long, New Leak Detection System 1 Order of Porous Tube Distance of Porous Tube i from Extraction Point (m) Airflow Rate Predicted for Porous Tube i, Q por,i (L/min) Permeability of porous tubes ¼ 0.9 Darcy, total airflow rate ¼ 7 L/min, extraction pressure ¼ atm absolute. was first calculated for the 30-m leak detection system using Equations 1 through 4. Table 1 shows the calculated airflow distribution for each porous tube in the 30-m leak detection system with extraction pressure ¼ atm absolute and total airflow rate ¼ 7 L/min. The breakthrough time was then calculated using Equations 5 through 7. Table 2 demonstrates that the calculated breakthrough time for the leak detection test is 21.9 min, which is in close agreement with that observed in the field, 19 min. Assuming VOCs volatilize from the leakage point to the nearest porous tube, the mathematical model for calculating the airflow distribution, Equations 1 through 4, and the mathematical model for calculating the breakthrough time, Equations 5 through 7, would be useful in determining the approximate location of a pipeline leakage. For a leakage event, the breakthrough time can be calculated for various assumed leak locations. By comparing the observed Table 2 Breakthrough Time Calculation for Leak Detection Test Using a 30-m-Long, New Detection System Order of Section Flow Rate of Air through Porous Tube i, Q por,i (L/min) Flow Rate of Air Inside Section k, Q pe,k (L/min) Travel Time in Section k, t k (min) Total sum t BT (min) 21.9 breakthrough time with those calculated, the approximate location of a pipeline leakage can be estimated. Figure 6 shows the calculated distribution of airflow on a semilogarithmic paper for a 0.9-Darcy leak detection system with total airflow rate ¼ 35 L/min and extraction pressure ¼ 0.9 atm absolute. The permeability of porous tubes is 0.9 Darcy. The porous tubes are 1 m apart from one another. Figure 6 allows one to read the percentage of airflow through individual porous tubes in the composite sample toward the end segments and to design the maximum line length. The percentage of airflow through the 24th, 27th, 30th, 32nd, and 34th porous tube in the composite sample is 0.5%, 0.3%, 0.2%, 0.1%, and 0.05%, respectively. The most challenging situation in leak detection would be a leak located at the last porous tube since its volumetric contribution to the composite sample would be the smallest. For example, given an acceptable minimum percentage of airflow through the last porous tube at 0.2%, the maximum line length of a composite sampling unit would be 30 m long using the design curve shown on Figure 6. To K.F. Liang and M.C.T. Kuo/ Ground Water Monitoring & Remediation 26, no. 3:

6 Percentage of air flow through individual porous tube ( ) (0.5, 24 m) (0.3, 27 m) (0.2, 30 m) (0.1, 32 m) (0.05, 34 m) Distance of individual porous tube from the extraction point (m) Figure 6. Calculated distribution of airflow in a 0.9-Darcy, new leak detection system (total airflow rate = 35 L/min, extraction pressure = 0.91 atm absolute). Table 3 Concentration of Total Soil-Gas Hydrocarbons Measured in Leak Detection Test Using Conventional Soil-Gas Probes Probe Number Distance of Each Soil-Gas Probe from Leakage Source (m) Concentration of Total Soil-Gas Hydrocarbons (ppmv) Background ¼ 0.3 ppmv. install a leak detection system along a 1-km-long pipeline, a total of 33 independent 30-m-long composite sampling units would be required. Leak Detection Test Using Soil-Gas Probes The main objective of the leak detection test using conventional soil-gas probes was to evaluate the effective detection distance of conventional soil-gas probes compared with the leak detection system. The simulated leakage was a source of 3 L gasoline released at a depth of 1 m below ground surface (Figure 7). As shown in Figure 7, three conventional soil-gas probes were installed to a depth of 1 m below ground surface and located at 2, 3, and 5 m away from the source of a simulated leakage, respectively. An extraction pump with a total flow rate of 7 L/min and a portable FID detector (Foxboro TVA 1000 model) was used in the leak detection test. Table 3 presents the results of leak detection tests using conventional soil-gas probes. The results show that the concentration of total hydrocarbons in soil gas decreased dramatically when the distance between the ground probe and the leakage source only increased from 2 to 5 m. The leak detection test confirmed that the effective detection Soil-gas probe Extraction pump (7 L/min) FID distance is ~5 m for conventional soil-gas probes using an extraction pump with a 7 L/min flow rate. Thus, applying conventional soil-gas probes to detect a leakage along a long-distance pipeline would require a large number of soil-gas sampling points and analyses because of the small effective detection distance of each soil-gas probe. While the new leak detection system was originally designed and tested for long-distance pipelines to save a large number of soil-gas samples and analyses, it is also applicable to other situations of various geometric shapes, such as large-diameter storage tanks and plumes of contaminated ground water. Figure 8 shows the schematic diagram of a new leak detection system installed around the periphery of a large-diameter petroleum storage tank. In addition, if measuring VOCs in soil gas indicates the concentrations of VOCs in ground water, the new leak detection system can be employed to monitor the progress of natural attenuation and bioremediation of ground water contaminated by petroleum leakage. Conclusions This paper presents the idea of a new leak detection system and demonstrates its application for a long-distance pipeline using soil-gas techniques. The new detection system consists of intermittent porous tubes connected in series with impermeable polyethylene tubes. The following Storage tank Polyethylene tube Extraction pump FID Leakage source 2 m 3 m 5 m Ground surface Coarse sand backfill Natural soils Figure 7. Schematic diagram of leak detection test using conventional soil-gas probes. Figure 8. Schematic diagram of a new leak detection system installed around the periphery of a large-diameter petroleum storage tank. 58 K.F. Liang and M.C.T. Kuo/ Ground Water Monitoring & Remediation 26, no. 3: 53 59

7 conclusions are based on the results of air extraction and leak detection tests. 1. Field leak tests were conducted using both conventional soil-gas probes and the new detection system. Results indicated that the effective detection distance of the new system was at least 30 m, while the effective detection radius from the conventional soil-gas probes was only ~5 m. 2. Applying the new leak detection system to longdistance pipelines, a large amount of soil-gas samples and analyses can be saved. 3. A mathematical model to predict the distribution of airflow in the new detection system was developed and verified with the air extraction tests. Darcy s law and Hagen-Poiseuille law were used in the model to describe the flow behavior of soil gas through the porous tubes and inside the polyethylene tubes, respectively. The model is useful in designing the distribution of airflow in a new detection system to monitor fuel leakage from a long-distance pipeline. 4. The approximate location of a detected leakage event could be determined from the measured breakthrough time using the airflow distribution model. Acknowledgment This research has been supported by NSC of Taiwan (NSC E , NSC E , NSC E , NSC E , and NSC M ). References Amyx, J.W., D.M. Bass, and R.L. Whiting Petroleum Reservoir Engineering. New York: McGraw-Hill Inc. Bird, R.B., W.E. Stewart, and E.N. Lightfoot Transport Phenomena. New York: John Wiley Sons Inc. Kerfoot, H.B., and C.L. Mayer The use of industrial hygiene samplers for soil-gas surveying. Ground Water Monitoring Review 6, no. 4: Kerfoot, H.B., C.L. Mayer, P.B. Durgin, and J.J. D Lugosz Measurement of carbon dioxide in soil gases for indication of subsurface. Ground Water Monitoring Review 8, no. 2: 67. Kuo, M.C.T Applied research of horizontal porous injection tubes for soil in-situ bioremediation. NSC E , National Science Council, Taipei, Taiwan. Liang, K.F Applied research of horizontal porous injection tubes for pipeline leak detection and soil in-situ bioremediation. M.S. thesis, Department of Mineral and Petroleum Engineering, National Cheng Kung University, Tainan, Taiwan. Marrin, D.L Soil-gas sampling and misinterpretation. Ground Water Monitoring Review 8, no. 2: Marrin, D.L Delineation of gasoline hydrocarbons in groundwater by soil gas analysis. In Proceedings of the 1985 Hazardous Materials Management Conference, West Tower Conference Management Company, Wheaton, Illinois: Marrin, D.L., and H.B. Kerfoot Soil-gas surveying techniques. Environmental Science and Technology 22, no. 7: Marrin, D.L., and G.M. Thompson Gaseous behavior of TCE overlying a contaminated aquifer. Ground Water 25, no. 1: Pasteris, G., D. Werner, K. Kaufmann, and P. Hohener Vapor phase transport and biodegradation of volatile fuel compounds in the unsaturated zone: A large scale lysimeter experiment. Environmental Science and Technology 36, no. 1: Spittler, T.M., L. Fitch, and S. Clifford A new method for detection of organic vapors in the vadose zone. In Proceedings of the Annual Symposium on Characterization and Monitoring of the Vadose Zone; National Water Well Association, Dublin, Ohio: Swallow, J.A., and P.M. Gschwend Volatilization of organic compounds from unconfined aquifers. In Proceedings of the 3rd National Symposium on Aquifer Restoration and Groundwater Monitoring; National Water Well Association, Dublin, Ohio: Thompson, G.M., and D.L. Marrin Soil gas contaminant investigations: A dynamic approach. Ground Water Monitoring Review 7, no. 3: Voorhees, K.J., J.C. Hickey, and R.W. Klusman Analysis of groundwater contamination by a new surface static trapping/ mass spectrometry technique. Analytical Chemistry 56, no. 13: Wolfe, W.J., and S.D. Williams Soil gas screening for chlorinated solvents at three contaminated karst sites in Tennessee. Ground Water Monitoring Review 22, no. 4: Biographical Sketches K.F. Liang is a Ph.D. candidate at the Department of Mineral and Petroleum Engineering, National Cheng Kung University, Tainan, Taiwan. M.C. Tom Kuo, Ph.D., Environmental Engineering, corresponding author, is a professor at the Department of Mineral and Petroleum Engineering, National Cheng Kung University, University Avenue, Tainan, Taiwan; (886) ext.62827; fax (886) ; mctkuobe@mail.ncku.edu.tw. K.F. Liang and M.C.T. Kuo/ Ground Water Monitoring & Remediation 26, no. 3: