FOUNDRY GREEN SANDS AS HYDRAULIC BARRIERS: LABORATORY STUDY By Tarek Abichou,1 Craig H. Benson,2 and Tuncer B. Edil,3 Members, ASCE ABSTRACT: A laboratory testing program was conducted to assess the use of foundry sands from gray iron foundries, typically called green sands, as hydraulic barrier materials. Foundry green sands are mixtures of fine uniform sand, bentonite, and other additives. Specimens of foundry sand were compacted in the laboratory at a variety of water contents and compactive efforts and then permeated in rigid-wall and flexible-wall permeameters to define relationships between hydraulic conductivity, compaction water content, and dry unit weight. Additional tests were conducted to assess how hydraulic conductivity of compacted foundry sand is affected by environmental stresses such as desiccation, freeze-thaw, and chemical permeation. Results of the tests show that the hydraulic conductivity of foundry sand is sensitive to the same variables that affect hydraulic conductivity of compacted clays (i.e., compaction water content, and compactive effort). However, hydraulic conductivities 20.
INTRODUCTION Wisconsin gray-iron foundries generate about 800,000 Mg of byproducts per year, most of which are landfilled. High costs of landfilling and the potential uses of foundry byproducts have prompted research into their beneficial reuse (Abichou et al. 1998a,b). One byproduct of particular interest is excess ‘‘green sand,’’ a byproduct of most gray-iron foundries. Foundry green sand, which is referred to simply as ‘‘foundry sand’’ in this paper, is primarily a mixture of sand and sodium bentonite, and thus has potential for use in construction of hydraulic barrier layers in applications where leachate emanating from the foundry sand will ultimately be collected (e.g., in landfill covers). Some excess foundry sands that are lightly or not contaminated may also be considered for landfill liners. Abichou et al. (1998c) presents a case study where one foundry sand from a foundry in central Wisconsin was successfully used in the construction of a landfill liner and cover. The objective of this study was to assess the feasibility of using foundry sands for hydraulic barriers. Twelve foundry sands were collected from 11 foundries in Wisconsin and Illinois. Specimens of each foundry sand were compacted in the laboratory over a range of water contents and compactive efforts and then permeated with tap water to measure their hydraulic conductivity. Results of these tests were used to determine how hydraulic conductivity of foundry sands is affected by compaction conditions and composition of the sand. Additional tests were conducted to assess how hydraulic conductivity of foundry sand is affected by desiccation, freeze-thaw, and permeation with chemicals other than water that might be encountered in municipal solid waste landfills. 1
Res. Sci., Dept. of Civ. and Envir. Engrg., Univ. of Wisconsin-Madison, Madison, WI 53706. E-mail:
[email protected] 2 Prof., Dept. of Civ. and Envir. Engrg., Univ. of Wisconsin-Madison, Madison, WI. 3 Prof., Dept. of Civ. and Envir. Engrg., Univ. of Wisconsin-Madison, Madison, WI. Note. Discussion open until May 1, 2001. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was submitted for review and possible publication on November 23, 1999. This paper is part of the Journal of Geotechnical and Geoenvironmental Engineering, Vol. 126, No. 12, December, 2000. 䉷ASCE, ISSN 1090-0241/00/0012-1174–1183/ $8.00 ⫹ $.50 per page. Paper No. 22166.
BACKGROUND Foundry Sand Generation Foundries use sand in two ways: for molds that form the outside of the casting and in cores that form the internal shapes and cavities within the casting (Javed and Lovell 1994). The sand is bonded by natural clays (e.g., bentonite) or with chemical agents, such as phenolic urethane. Carbon additives, such as coal dust, are added to control gas permeability, strength, and other properties of the mixture (Bralower 1988). When natural clays bond the sand, the mixture is referred to as ‘‘green sand.’’ The term green sand is used because molten metal is poured into the mold when the sand is damp or green (Javed et al. 1994). The sand component of foundry sand is referred to as ‘‘base sand.’’ After the molten metal cools, the casting is removed from the foundry sand mold. The casting is then shaken to break out cores (chemically bonded sands used to form cavities, such as cylinders in an engine block) and to remove any foundry sand adhering to the casting. The used sand is reclaimed and mixed with crushed cores. Base sand, bentonite, and additives are introduced to the reclaimed sand to maintain the desired properties of the foundry sand. This results in a gradual accumulation of foundry sand, a portion of which is discarded when the storage capacity of the foundry is reached. The discarded sand byproduct primarily is composed of excess foundry sand, but may also contain chemically bonded sand as well as core and mold butts. Previous Studies on Foundry Sand Reuse in Landfill Covers Kunes and Smith (1983) investigated the use of foundry sands as a construction material for landfill cover systems. Grain size distribution, compaction curves, and hydraulic conductivity were measured on several samples of foundry sand. Kunes and Smith (1983) provide little detail about their testing methods, but reported that their compacted foundry sands had hydraulic conductivities in the range of 1.3 ⫻ 10⫺6 to 6.1 ⫻ 10⫺6 cm/s. Grede Foundries Corporation of Reedsburg, Wis., conducted a field study regarding the suitability of their foundry sand as a material for landfill liners and covers. Freber (1996), Vierbicher Associates (1995), and Abichou et al. (1998c) provide summaries of the study. Five full-scale test sections were con-
1174 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / DECEMBER 2000
structed that simulated earthen final covers. Each test section was underlain by a lysimeter. Two of the test sections were constructed with compacted clay barrier layers, whereas the other three test sections were constructed with barrier layers composed of foundry sand. Clay barrier layers were compacted to at least 90% of modified Proctor maximum dry unit weight, whereas foundry sand barrier layers were compacted to at least 95% of maximum standard Proctor dry unit weight. Control on compaction water content was not reported. Percolation data were collected from the test sections for four years. Percolation rates for the test sections constructed with foundry sand were approximately two orders of magnitude lower than percolation rates for the test sections constructed with clay. Test pits excavated during decommissioning of the test sections showed that the clay barriers were severely cracked due to frost action and desiccation, whereas the barrier layers constructed with foundry sand were still intact (Albrecht 1996). Iroquois Foundry Company is conducting a cover study using foundry sand at the Monroe, Wis., airport (Environmental Compliance Consultants, Inc. 1997). The objective of the present study is to determine if foundry sand from Iroquois Foundry can be used instead of clay as a hydraulic barrier layer in the cap of a containment system. Three test sections were constructed during this study. Sections 1 and 2 contained a barrier layer of foundry sand, whereas the barrier layer in section 3 was compacted clay. Lysimeters were installed beneath each test section. All barrier layers were compacted to at least 90% of maximum dry unit weight based on modified Proctor and wet of optimum water content. To date, percolation from the test sections constructed with foundry sand has been at least four times lower than percolation from the test section constructed with a clay barrier layer. Foundry Sands as Sand-Bentonite Mixtures Since foundry sand is primarily sand and bentonite, its hydraulic properties should be similar to those of sand-bentonite (SB) mixtures. The hydraulic conductivity of SB mixtures typically ranges from 1 ⫻ 10⫺6 to 1 ⫻ 10⫺9 cm/s, with lower hydraulic conductivities generally associated with bentonite contents of 5–10% (Chapuis 1981; Lundgren 1981; Van Veen 1983; Abeele 1986; Garlanger et al. 1987; Chapuis 1990; Kenney et al. 1992; O’Sadnick et al. 1995; Gleason et al. 1997; Howell and Shakelford 1997; Kraus et al. 1997). Kenney et al. (1992) report that the hydraulic conductivity decreases at a rapid rate as the bentonite content increases from zero to 8%; thereafter, the hydraulic conductivity becomes comparable to the hydraulic conductivity of bentonite. Hitoshi et al. (1995), Howell and Shackelford (1997), and Abeele (1986) report that this break point occurs at different bentonite contents, depending on the type of bentonite and the gradation of the sand. Several investigators have studied how water content and compactive effort affect the hydraulic conductivity of SB mixtures. Haug and Wong (1992) indicate that molding water content is not a critical factor in design and construction of SB barrier layers because water is not required to break down clods and facilitate remolding, as is required for clayey soils. They report a maximum change in hydraulic conductivity of 4.5 times from dry to wet of optimum for their SB mixture. Kraus et al. (1997) report similar findings. However, at low water contents, Kenney et al. (1992) explain that molding water content can affect the hydraulic conductivity of SB mixtures because it affects the distribution of bentonite. Kraus et al. (1997) also suggest that the hydraulic conductivity of SB mixtures is not particularly sensitive to compactive effort. The insensitivity to compactive effort is consistent with the lack of correlation between porosity and hydraulic conductivity of SB mixtures reported by Chapuis (1990).
Wallace (1987), Wong and Haug (1991), Haug and Wong (1993), Kraus et al. (1997), and Zimmie et al. (1997) have investigated how freeze-thaw cycling affects the hydraulic conductivity of SB mixtures. Wallace (1987) reports that a 2% SB mixture exhibited no change in hydraulic conductivity when subjected to ten freeze-thaw cycles. Wong and Haug (1991) report that freeze-thaw cycling causes a decrease in hydraulic conductivity of SB mixtures, and that the decrease in hydraulic conductivity is greater for SB mixtures with lower bentonite content. They postulate that freeze-thaw cycling helps hydrate the bentonite, and that thawing redistributes bentonite into spaces between sand grains. Haug and Wong (1993) tested an 8% SB mixture subjected to three freeze-thaw cycles and two wet-dry cycles. They reported that the freeze-thaw cycling caused an increase in hydraulic conductivity of less than one order of magnitude; however, the increase in hydraulic conductivity disappeared after wet-dry cycling. Kraus et al. (1997) conducted laboratory and field studies to investigate how freeze-thaw cycling affected the hydraulic conductivity of a 12% SB mixture. In the laboratory and field, Kraus et al. (1997) found that the hydraulic conductivity of the SB mixture was unaffected by freeze-thaw cycling. Kraus et al. (1997) show that ice segregation does not occur in SB mixtures, which prevents cracking and subsequent increases in hydraulic conductivity. Zimmie et al. (1997) reported that the hydraulic conductivity of a mixture of a poorly graded medium sand with 10% bentonite was not affected by freeze-thaw cycling. The hydraulic conductivity of their mixture was 4.8 ⫻ 10⫺9 cm/s before freeze-thaw cycling, and then decreased to 3.4 ⫻ 10⫺9 cm/s after 15 freeze-thaw cycles. Zimmie et al. (1997) also report that thin sections and x-ray photography showed that ice-lenses and cracking did not occur in their SB mixtures during freezing. Albrecht (1996) prepared specimens of a SB mixture of well-graded crushed mine rock and 10% sodium bentonite. The hydraulic conductivity of the SB mixtures ranged from 2 ⫻ 10⫺6 cm/s (compacted dry of optimum) to 3.5 ⫻ 10⫺8 cm/s (compacted wet of optimum). After three desiccation cycles, the hydraulic conductivity of the SB mixture was essentially unchanged for all specimens, suggesting that SB mixtures can be resistant to damage caused by wetting and drying. Gipson (1985) permeated SB mixtures containing 7.5%, 10%, and 15% bentonite with acid liquor (pH = 2.2) that had high concentrations of calcium, calcium oxides, sodium, chloride, and very high concentrations of fluoride and sulphate. The acid liquor was obtained from an operating phosphogypsum field. During the hydraulic conductivity testing, the acid liquor was first allowed to flow through a gypsum layer before coming in contact with the SB mixtures. The hydraulic conductivity of mixtures prepared with 7.5, 10, and 15% bentonite content increased by 7, 13, and 41 times, respectively, relative to the initial hydraulic conductivity after one year of permeation. Shackelford (1994) permeated SB mixtures containing 16% bentonite with a saturated calcium tailings solution. The SB mixtures were compacted at water contents above optimum water content, and permeated at a constant gradient of 48 and an average effective stress of 41 kPa. Specimens that were permeated with the tailings solution first had a hydraulic conductivity of 2 ⫻ 10⫺5 cm/s. Specimens that were permeated with water first and then with tailings solution had a hydraulic conductivity of 2 ⫻ 10⫺7 cm/s. Alston et al. (1997) permeated silty SB mixtures containing 5–6% bentonite with tap water and a paper mill effluent having 260 mg/L dissolved sodium, 30 mg/L dissolved calcium, and pH = 7. The paper mill effluent was also used to perform Atterberg limits and free swell index on the bentonite. The liquid limit decreased from 450 to 395 and the plastic limit
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / DECEMBER 2000 / 1175
remained constant when performed with the mill effluent instead of water. The free swell index increased from 28 mL to 38 mL when performed with the mill effluent instead of water. Hydraulic conductivities of all specimens were below 1 ⫻ 10⫺8 cm/s regardless of the permeant used and remained constant even after 2.5 pore volumes of flow. Stern and Shackelford (1998) prepared SB mixtures and permeated them with CaCl2 solutions. Stern and Shackelford (1998) report that the hydraulic conductivity of the SB mixtures increased by up to two orders of magnitude when permeated with CaCl2 having a concentration 0.5 M. Evans et al. (1985) investigated how organic fluids affect the hydraulic conductivity of SB mixtures. The hydraulic conductivity of a SB mixture increased by four orders of magnitude when permeated with concentrated aniline (an organic base with a dielectric constant of 6.89), whereas permeation with an aqueous solution of aniline near the solubility limit (0.41 M) did not affect the hydraulic conductivity. Nirmala et al. (1995) also investigated how aniline affected SB mixtures. Hydraulic conductivity of a specimen containing 10% bentonite increased by more than four orders of magnitude, whereas the hydraulic conductivity of a specimen containing 20% bentonite was slightly lower when permeated with aniline relative to the hydraulic conductivity to water. MATERIALS AND METHODS Bulk samples of foundry sand were shipped to the University of Wisconsin-Madison from participating foundries in 210 L drums. Index properties were measured for each foundry sand. The liquid and plastic limits were measured in accordance with ASTM D 4318, except for the time of hydration. ASTM D 4318 specifies that a hydration time of 16 h is sufficient to conduct Atterberg limits test on clayey soils. Experience with the foundry sands suggested that a week of hydration is required to reactivate the ‘‘dead’’ bentonite in foundry sands (Kleven 1998). Thus, a one-week hydration period was used for the Atterberg limits as well as all other tests requiring hydration. Particle size distribution was determined in accordance with ASTM D 422. All foundry sand specimens were soaked in water for a minimum of one week prior to particle size distribution analysis. Methylene blue titration (ASTM C 837) was used to determine the bentonite content of all foundry sands. The foundry industry relies on ASTM C 837 to determine the clay content in their systems. Participating foundries performed the methylene blue tests on their foundry sand. Specific gravity was measured following ASTM D 854. Particle size distribution curves for all foundry sands are shown in Fig. 1. The particle size distributions are similar and resemble that of the base sand used at these foundries, except for the fines fraction. All of the foundry sands are comprised primarily of uniform fine sand. The fines content (percent passing No. 200 sieve) ranges from 10.0% to 16.4% (Table 1) and the bentonite content varies between 4.7% and 16.0%. Atterberg limits, specific gravities, bentonite contents, and classifications are summarized in Table 2. The foundry sands classify as SP-SM, SM, or SC. The liquid limit (LL) ranges from nonplastic to 29, and the plasticity index (PI) ranges from nonplastic to 7. Most of the foundry sands fall below the hatched zone and all fall above, but near the A-line on the plasticity chart. In contrast, clays normally used for compacted clay liners nearly always fall above the hatched zone and between the A- and U-lines (Benson et al. 1999). Plasticity of the foundry sands varies with bentonite content, as shown in Fig. 2. Foundry sands are plastic only when the bentonite content is >6%. As with SB mixtures, a bentonite content exists at which foundry sand changes from nonplastic to plastic.
FIG. 1. Sand
Particle Size Distribution for Foundry Sands and Base
TABLE 1. Chemical Analysis of MSW Leachate Used as Permeant Liquid Parameter (1) pH BOD COD Total dissolved solids Chloride Sulfide Sulfate Sodium Manganese Barium Beryllium Cadium Chromium Cobalt Copper Lead Zinc Nickel Silver Kjeldahl nitrogen Nitratre and nitrate nitrogen Total phosphorus Total sulfate Total sulfide Total suspended solids Hardness Benzene 1,4-Dichlorobenzene 1,1-Dichloroethane Cis-1,2-Dichloroethane Ethylbenzene Methylene chloride Naphthalene Sytrene Toluene Trichloroethene 1,3,5-Trimethylbenzene M&P-Xylene O-xylene
Typical rangea (4)
U.S. EPA method (5)
— 6.33 mg/L 11,700 mg/L 20,900 mg/L 393 mg/L 2,030 mg/L 3 mg/L 199 mg/L 1,380 g/L 5,420 g/L 1,140 g/L 7%. When the bentonite content is greater than 6%, the hydraulic conductivity is less than 1 ⫻ 10⫺7 cm/s for all but one data point (Sand 3, BC = 7%, standard Proctor). These findings are consistent with the behavior of SB mixtures. Kenney et al. (1992) report that compacted SB mixtures with bentonite content ⱕ8% have much higher hydraulic conductivity than mixtures with higher bentonite content. Between 8% and 12%, the hydraulic conductivity of their
FIG. 6. Hydraulic Conductivity at Optimum Water Content versus Bentonite Content
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / DECEMBER 2000 / 1179
SB mixtures decreased gradually with increasing bentonite content. Kenney et al. (1992) report that at bentonite contents >12%, all open voids between the sand particles were filled, and the mixture behaved like a matrix of bentonite filled with sand grains. Abeele (1986) also reports that the hydraulic conductivity of SB mixtures decreases dramatically between bentonite contents of 0% and 5%, and that at bentonite contents >10% the hydraulic conductivity of a SB mixture equals that of the bentonite. The trend in Fig. 6 is similar to the trend reported by Abeele (1986). Haug and Wong (1992), Hitoshi et al. (1995), Howell and Shakelford (1997) report that the hydraulic conductivity of their mixtures decreases sharply with increasing bentonite content until it reaches a lower limit of hydraulic conductivity associated with that of bentonite. This break point occurs at different bentonite contents, depending on the type of bentonite and the gradation of the sand. The data in Fig. 6 suggest that bentonite content can be used to screen out foundry sands that are unlikely to achieve low hydraulic conductivity. If a foundry sand has bentonite content ⱖ6%, additional hydraulic conductivity testing should be performed to define compaction conditions that yield acceptable low hydraulic conductivity (e.g., as in Daniel and Benson 1990). In contrast, if the bentonite content is 10⫺7 cm/s and these three specimens have hydraulic conductivity near 10⫺7 cm/s. In contrast, all specimens with bentonite content 10⫺7 cm/s. Also, porosity of the foundry sands with bentonite content >6% varies through a broad range (0.23– 0.47), whereas, porosities of the foundry sand with bentonite content 6%. Thus, when the bentonite content is >6%, hydraulic conductivity 2, or bentonite content >6% can be compacted to achieve hydraulic conductivities
