University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Colorado School of Mines
|
1
1.0 INTRODUCTION
Water influenced by mining activities is a major cause of concern for many
communities and watershed organizations throughout the world. Mines currently
operating in the United States must comply with strict environmental regulations.
Environmental cleanup and reclamation plans must be established long before mining
begins. However, this was not always the case for the mining industry. Many past
operations did not treat, contain, or properly dispose of their waste. Over time, water
can flood mines and percolate through mine waste rock piles and mill tailings. Mine
water, combined with air, may result in the production of acid rock drainage (ARD, also
known as acid mine drainage or AMD) and high loads of heavy metals. Water
contaminated with acid rock drainage may not be suitable for municipal use, livestock
watering, wildlife, irrigation, or industrial use (Cohen & Staub 1992).
There are over 25,000 inactive mine sites and exploration prospects in the
Western United States (Drury 2000). Abandoned mine sites can be found throughout
the Rocky Mountains and Front Range of Colorado. These areas lie within the Colorado
Mineral Belt, shown in Figure 1.1. The mineral belt is an area extending north-east from
southern Colorado towards Boulder, CO. The belt is abundant with ore deposits and
while it is most famous for its gold deposits, there are also numerous deposits of
molybdenum, lead and zinc.
Most Colorado historical mining camps and communities were located within the
Colorado Mineral Belt, with the exception of the Cripple Creek Mining District. As a
result of historical mining in the region, abandoned mine sites are scattered across the
Colorado Mineral Belt. Abandoned mine sites are common and hundreds can be seen
along many roads and highways throughout the state.
|
Colorado School of Mines
|
2
COLORADO MINERAL BELT
Adapted from
Twelo and Sims (1963)
Figure 1.1: Location Map of the Colorado Mineral Belt
(Accessed from Wikipedia 2007)
Many environmental problems commonly associated with mining-related
activities result from inactive and often abandoned sites. While these mine sites
represent an important, historical era, water draining from abandoned mines is often
acidic and may contain high metal concentrations. Abandoned mine workings often fill
with water, which can increase the amount of contaminated drainage coming from the
mine. In regions where snowmelt plays a key role in the water balance, spring runoff
can greatly influence the amount of water that drains from abandoned sites. Also, mine
waste piles contain void spaces that permit air and water infiltration, resulting in the
oxidation and weathering of rock fragments and minerals (Diehl et al. 2007). These
impacts can have detrimental effects on water quality for kilometers downstream.
1.1 Research Objectives
For this research project, I tested the applicability of anaerobic passive
bioreactors as a treatment system for mining influenced water draining from an
abandoned uranium mine. A common problem of anaerobic bioreactors is the reduction
of substrate permeability due to the breakdown of organic matter, clogging of pores, and
the possible formation of biofilms. One approach to reduce the affects of pore
obstruction and loss of hydraulic conductivity is to incorporate soil amendments to the
|
Colorado School of Mines
|
3
organic substrate. In this study, I used the Profile brand soil amendment (Manufactured
by Profile Products LLC). Variations of Profile are commonly used under golf courses
and athletic fields to retain moisture for lush green grass as well as to improve the
drainage and avoid pooling. The addition of Profile improves the structural integrity of
the soil or substrate, which can aid in reducing and preventing compaction. Profile soil
amendment is also used in a variety of landscaping applications, including residential
lawns and gardens, due to its ability to retain nutrients within the soil and reduce
leaching. Another advantage of using Profile is its commercial availability and it can be
purchased at most home improvement stores. Therefore, if it proved to effectively
enhance substrate hydraulic conductivity, it can readily be used in full scale passive
bioreactors.
The purpose of this research project was three fold. The first goal of this project
was to fully characterize the water quality at the Fair Day Mine site. The complete
chemical composition of the mine drainage at Fair Day was necessary to choose an
appropriate treatment method for the abandoned mine site. The second purpose was to
limit the reduction of porosity and hydraulic conductivity within the bioreactors over time,
while still achieving a high level of metal removal and acid neutralization. This was done
by amending a base organic substrate with different percentages by volume of a calcite
clay ceramic soil amendment, Profile. Lastly, the layout of the field site permitted the
installation of a series of settling pools along side the anaerobic bioreactors. Thus, I
could compare the treatment efficiency of the anaerobic bioreactor to an aerobic
treatment system as represented by the oxidation, settling pools. Through this
comparison of treatment methods the benefits and pitfalls of each of the systems could
be identified. The knowledge gained from this research will provide additional
assistance in making decisions for treatment options in future mine remediation projects.
|
Colorado School of Mines
|
5
2.0 BACKGROUND INFORMATION
Previous to the installation of any form of mine drainage treatment system the
site-specific characteristics must be evaluated. Each site has its own unique properties
that determine which treatment method would be most effective for that site. Passive
systems offer a variety of treatment methods for mining influenced waters.
2.1 Impacts of Mining Influenced Waters
Mining operations and processes can significantly impact the chemistry of water
that comes in contact with rocks containing sulfide minerals, both in mine workings as
well as the piles of waste rock and tailings. Mining influenced water is often described
as being acidic in pH and containing high concentrations of sulfate and heavy metals.
For decades, water of this composition was referred to as acid mine drainage (AMD) but
this term implies that the mine itself negatively impacted the water. Today, scientists
recognize that the source of acidity and metal loading to the waters is due to the specific
type of rock, and therefore, waters with these chemical characteristics can be described
as acid rock drainage (ARD).
However, not all water draining from mines or mining related processes is acidic.
Waters impacted by mining activities can have near neutral or even basic pH, and still
have high metal concentrations. These types of waters are referred to as mining
influenced water (MIW). This definition of water impacted by mining activities recognizes
the detrimental effects on the effluent and downstream waters without implying that the
waters are acidic (Smith & Ranville 2007). When MIW containing high metal
concentrations is released into the environment, the downstream impact is evident. High
levels of heavy metals can make surface waters uninhabitable for plants and animals.
|
Colorado School of Mines
|
6
Fish and other aquatic life are especially susceptible to the toxic effects of mine drainage
contamination.
2.2 Formation and Chemistry of Acid Rock Drainage
Acid rock drainage (ARD) is unique among industrial contaminants by the fact
that it is self-perpetuating (Kalin et al. 2006). In order for ARD to develop, water, oxygen
and metal sulfides must be present. The most common characteristics of ARD are
elevated concentrations of metals and low pH values. ARD is formed when sulfidic
materials, often pyrite, are exposed to oxygen and water (Stumm & Morgan 1981).
Oxidation of pyrite (FeS2) produces H+ ions that cause the pH of the water to drop. The
oxidation process of pyrite can be summarized by the following reactions (Stumm &
Morgan 1981):
(1) FeS2(s)-To2 + H20 -> Fe2* + 2S042' + 2H+
As pyrite is exposed to water and atmospheric oxygen, it weathers and breaks
down into ferrous iron and sulfate ions. Hydrogen ions are released into the water
during this process causing the pH of the water to drop and the ferrous iron remains in
solution free to react further.
(2) Fe2* + —02 +H+ ->Fe3+ +J-H,0
4 2
The ferrous iron is then oxidized to ferric iron as it is exposed to more oxygen
and hydrogen ions in the water and the pH will continue to drop.
(3) Fe3+ + 3H20 -> Fe(OH)3 + 3H"
The ferric iron continues to be exposed to more water that will lead to iron
hydrolysis and the formation of iron hydroxides. Additional free hydrogen ions are
released in the water and the pH further decreases. Iron hydroxide complexes may
|
Colorado School of Mines
|
precipitate out of solution and can coat and armor rocks, stream beds, or earth which the
water flows across. Fe(OH)3 precipitates are easily identified by the yellow, orange or
red deposits they create. The precipitates are commonly referred to as “yellow boy".
If ferric iron is formed while in contact with pyrite, the following reaction can occur
further dissolving the pyrite into solution increasing the hydrogen ion concentration in the
water:
(4) FeS2 +14Fe34 +8H20->15Fe24 +2S042" +16H4
The overall reaction leading to the formation of acid rock drainage can be
summarized by the following equation (Cohen 2007):
(5) 4FeS2 +1502 +14H20^4Fe(0H)3 4-+8H2S042"
Equation 5 shows that the combination of pyrite, oxygen and water results in the
production of yellow boy and sulfuric acid. Mine waste rock piles and tailings contain
broken and crushed fragments of rock that offer a large surface area for these chemical
reactions to take place and contaminants to form. Precipitation events at mine sites with
exposed waste and tailing piles perpetuate the cycle of ARD formation and
contamination.
In many cases leaching of metals from mine waste rock and tailings piles
containing sulfidic materials does not involve the formation of acidic waters. Waters can
have near neutral pH values, ranging 6.0-8.0, and still have very high concentrations of
metals. Even though these waters are not acidic in nature, they can still be detrimental
to the surrounding environment by supplying significant metal loads to near by bodies of
water. In fact in most cases, the metal load is of greater concern than the acidity of the
water in terms of environmental damage (Sheoran & Sheoran 2006). If mine waste piles
are not remediated and exposed to the air, further weathering of the rocks could cause
perpetual heavy metal contamination.
|
Colorado School of Mines
|
8
2.3 Fair Day Mine
The Fair Day Mine is an abandoned uranium mine located in Boulder County,
Colorado approximately 3.22 kilometers west of Jamestown (Section 26, Township 2
North, Range 72 West). The mine is located on National Forest System land on the
south side of Overland Mountain and north of James Creek at an elevation of 2,400
meters. The mine includes two underground workings. The older working is high on the
south slope of Overland Mountain and is known as the Upper Fair Day Shaft. This site
consists of an inclined shaft that has caved in and was determined not to be an
environmental concern during previous site investigations (Neubert & Wood 2000). The
working with potential for environmental degradation is a lower adit, located
approximately 110 meters below the original inclined shaft. The location of the Lower
Fair Day workings can be seen in Figure 2.1:
Location Map of the Fair Day Mine
A. Fair Day Mine
102J Fores! Road
""*** County Road 94
Figure 2.1: Location Map of the Lower Fair Day Mine
|
Colorado School of Mines
|
9
2.3.1 Lower Fair Day Site History
Previous to reclamation activities, the Lower Fair Day Mine site consisted of one
adit and two waste rock piles located in a narrow valley. The adit and a small waste rock
pile were located on the east side of an unnamed tributary to James Creek. A larger
waste rock pile was located on the west side of the unnamed tributary. The mine portal
was 2.1 meters high and 1.8 meters wide and the adit was open to depth of
approximately 150 meters.
The two waste rock piles contained about 3,000 cubic-meters of material. The
unnamed tributary flowed between the two piles and eroded both piles at their bases.
Erosion rills cut the face of the sparsely vegetated dump, and minor sheet wash erosion
was evident. A concrete foundation, probably the remains of an ore-transfer facility, lied
in the meadow about 150 meters south of the mine. Besides the occasional campers
and forest users that visit the site, local residents that live in a subdivision above the site
use a small path running through the valley and the Fair Day site to access Country
Road 102J into Jamestown.
The main concern and source of environmental pollution at the site was the
contaminated water flowing from the adit. Prior to reclamation, mine water flowed from
the adit and drained south down the access road for about 30.5 meters at which point
approximately half the flow was diverted to the unnamed tributary through a pipe. The
remaining flow continued down the road towards Fair Day Meadow. The effluent
channel was armored in moderate amounts of red precipitates near the portal. The
volume of precipitates diminished downstream of the portal and by the time the mine
effluent reached the meadow south of the mine there were no precipitates.
The two waste rock piles also posed significant threats to the local environment.
Several metals in the waste rock piles were found to exceed Risk Management Criteria
|
Colorado School of Mines
|
10
for metals in soil (Neubert & Wood 2000). These elevated levels of contamination posed
a potential risk to human health and the environment. A paste pH test performed on a
composite sample from the dump produced a pH of 4.27, indicating that the waste rock
pile had the potential to generate acidic water (Neubert & Wood 2000). Scintillation
readings were taken on the waste rock piles to measure the level of radioactivity.
Readings were irregular and varied across the site, but in general most of the waste rock
produced 1,000 to 1,500 counts per second. Background concentrations in the
unmineralized metamorphic rocks surrounding the mine were in the range of 150 to 250
counts per second. While no one lives directly on the Fair Day site, the area is used by
the public for hiking, camping, and target shooting as well as ATV and motorcycle riding
(Neubert & Wood 2000). Intermittent use of the site created the concern of the public
contacting the radioactive material.
2.3.2 Operational History
In 1954, uranium was discovered apporximately 0.8 km north of the Fair Day
Mine. In 1955, uranium was discovered near the upper workings, and claims were
staked in the area. In April 1960, the lower adit was completed and the mine went into
production. From 1956 to 1961, La Salle Mining Company operated the mine under a
lease from Coliowa Uranium Corporation (Neubert & Wood 2000).
In 1961, because of litigation between the leaseholder (Coliowa Uranium
Corporation) and mine operator (La Salle Mining Company), the mine operator removed
most of their equipment with the exception of a vent line and rail. By 1962, timber
throughout the mine was rotting and needed replacement. The mine was rehabilitated
and minor amounts of uranium ore were produced 1964. As of 1967, the raise between
the upper and lower levels was considered hazardous and inaccessible. In August
1976, an attempt was made to re-open the lower adit, but these efforts ceased when it
|
Colorado School of Mines
|
11
was discovered that the adit was caved in about 152 meters from the portal. No
production was reported from these efforts.
During its operation, the Fair Day Mine was the largest uranium producer in
Boulder County. About 95 percent of the ore from the Fair Day Mine was produced
during the period La Salle operated the mine (1958-1961). During this period, La Salle
reportedly removed 3,670 metric tons of ore from the upper level and 14,120 metric tons
from the lower level (SAIC 2002).
The mining claims that included the Fair Day Mine, reported assessment work
until 1985 and were declared void in 1987. Between 1987 and 1991 mapping and
geochemical and geophysical surveys were completed on parts of a new claim block
that included the Fair Day Mine. This block of claims was declared abandoned in
November 1992 and was added to the USDA Forest Service’s Abandoned Mine Lands
Inventory.
2.3.3 Water Quality of the Site
In order to fully characterize the extent of contamination numerous sampling
events took place at the Lower Fair Day before any remediation work was done. The
volume of water flowing from the adit was in the range of 15 to 30 L per minute. Despite
the low flow, the mine water had high metal concentrations. The mine water was
moderately acidic (pH =4.5-5.0) and exceeded Colorado water quality standards for
seven metals when it was sampled in 1999 during the initial site investigation. The adit
effluent exceeded water quality standards for aluminum, cadmium, copper, manganese,
thallium, uranium, and zinc (Neubert & Wood 2000). Uranium and manganese were
considered to be of significant concern, as their concentrations in the mine water were
approximately 100 times the state water quality standards of 30 pg/L and 5 pg/L,
respectively (Neubert & Wood 2000).
|
Colorado School of Mines
|
12
The unnamed tributary running through the site had a pH of 7.6 and 7.4 above
and below the mine site respectively. Conductivity increased slightly downstream of the
mine site. Flow in the tributary stream was approximately 150-230 L per minute. As a
result of mine drainage and contributions from the eroding waste rock piles,
concentrations of manganese, uranium, and zinc exceeded state standards in the
tributary downstream of the mine site (Neubert & Wood 2000).
There was no measurable difference in metal concentrations in James Creek
above and below the confluence with the unnamed tributary (Neubert & Wood 2000).
This is primarily due to dilution, as the flow in James Creek is approximately 200 times
the flow in the tributary stream. Sample results from James Creek downstream of the
Fair Day Mine and upstream of Jamestown show the water is within state standards for
all of the tested constituents. Further downstream towards Jamestown, James Creek
does exceed Colorado State Water Quality Standards for aluminum, cadmium, and
copper and does not meet its Designated Use for aquatic life.
The MIW draining from Fair Day and the waste rock piles presented potential
hazards and concerns for both humans and animals using the area. Both deer and elk
migrate through this area and forage along the stream. James Creek supports a brook
and brown trout fishery that could be impacted by a catastrophic release from the site.
However, the most significant potential hazard associated with the water from Fair Day
is contamination of the drinking water supply for the town of Jamestown. The drinking
water intake for Jamestown is located approximately three kilometers downstream from
the Lower Fair Day site in James Creek.
USDA Forest Service employees determined that if the issues at the Lower Fair
Day Mine were not addressed, on-going and potential releases of hazardous substances
could impact drinking water sources, soils, and wildlife in the surrounding area. The
presence of contaminated soil and water would continue to pose threats to National
|
Colorado School of Mines
|
13
Forest visitors. The Forest Service proposed a Removal Action intended to reduce the
potential threats to human health and the environment by severing existing and potential
soil, water, and air exposure pathways (Neubert & Wood 2000).
A component of the removal action was the construction of a series of aerobic
settling pools. The series of settling pools store the mine drainage and allow many of
the metals to form hydroxide complexes. These complexes precipitate out of solution
and are contained within the treatment pools. The water is aerated slightly as it falls
across a rock cascade while traveling downhill from one step pool to the next.
2.4 Passive Treatment Systems
There are several types of passive treatment systems suitable for the
remediation of MIW. The most common passive treatment systems are aerobic and
anaerobic constructed wetlands. Other common types of passive treatment include
open limestone channels, anoxic limestone drains, permeable reactive barriers, and
anaerobic bioreactors. A major component for success using a passive treatment
system is the complete chemical characterization of the site to determine the application
that best fits the specific site conditions. Each mine site and its effluent drainage are
different and must be fully examined prior to the development or implementation of any
type of treatment system. The decision of which passive treatment method to use
should be based on site-specific data.
2.4.1 Benefits of Passive Treatment
There are many features of passive treatment systems that make them appealing
for the remediation of mine drainage. In comparison to traditional chemical precipitation
methods for treating MIW, passive systems require less operation and maintenance;
important when there is limited access to a site (Ganse 1993, Johnson & Hallberg 2002,
|
Colorado School of Mines
|
14
Neculita et al. 2007). Passive systems are designed to provide effective treatment
without relying on external chemical or electrical inputs (Guertin et al. 1985). Therefore,
if the site can only be reached for part of the year, the system should continue treating
water through the winter or when the site is inaccessible.
Another advantage is that passive systems do not require as many supplies or
facilities as traditional active chemical treatment systems. Pump and treat chemical
precipitation systems require a continuous supply of large amounts of chemical reagents
to function. This would require that chemicals be regularly transported to the site and
chemical storage facilities would be required onsite. Along with onsite chemical storage
comes the increased risk of chemical releases into the environment. Thus, full-time staff
and 24-hour maintenance would be necessary. Therefore, active treatment would not
be a reasonable solution for many abandoned mine sites, especially those with a remote
location or restricted access.
Traditional chemical precipitation treatment methods for MIW also produce large
volumes of oxidized wet sludge (Johnson & Hallberg 2002). The addition of lime,
carbonates, or sulfides to MIW cause metals to precipitate out of solution as metal
hydroxides, carbonates, or sulfide sludges. In order to reduce the volume of sludge,
dewatering is required. The dewatering process demands its own facility, which also
increases the footprint and cost of installing such a treatment system. Due to the
massive volumes of sludge that can accumulate from these kinds of treatment systems,
hauling and disposal costs can be expensive (Jong & Parry 2003, Elliott et al. 1998).
Capital operation costs and continued maintenance costs for chemical precipitation can
be prohibitive at many mine sites. While there are other technologies to treat MIW such
as ion exchange, reverse osmosis, electrodialysis, and electrolytic recovery, these
treatment methods are expensive and, as a result, are not commonly used (Neculita et
|
Colorado School of Mines
|
15
al. 2007). Both of these factors make traditional, active treatment technologies
unsuitable for use at abandoned mine sites.
Passive treatment systems are lower cost alternatives for treatment of MIW
(Sheoran & Sheoran 2006, Neculita et al. 2007). In fact, the cost of passive treatment
systems are often measured in terms of the amount of land required for the footprint of
the system rather than materials and labor (Gazea et al. 1996). Passive systems
employ slow processes for contaminant treatment that require longer retention times
and, therefore, generally larger footprints than conventional treatment systems. High
equipment mobilization costs can be one of the most expensive aspects of installing a
passive treatment system. However, once equipment and materials are onsite,
construction costs are often comparable to those associated with setting up a traditional
active treatment system. The annual costs required for maintenance and operation of
passive treatment systems are generally much lower than those of traditional active
forms of treatment (Neculita et al. 2007). Another advantage to using passive forms of
treatment for mine remediation is that these systems involve no electrical energy
consumption (Neculita et al. 2007). This is an attractive feature of passive systems
since electricity or other forms of energy supply are usually not available at abandoned
mine sites.
Another benefit to using passive treatment systems is that they often take
advantage of naturally occurring biological processes (Gazea et al. 1996, Johnson &
Hallberg 2005a). Microbial activity within the substrate of anaerobic treatment systems
can aid in reducing metal concentrations in water. Therefore, as long as the microbes
living in the system have the appropriate conditions for survival they can help to treat
mine drainage and usually benefit from the process by obtaining energy. In the past 20
years, research has focused on biological methods for treatment of MIW due to their
numerous advantages (Neculita et al. 2007).
|
Colorado School of Mines
|
16
2.4.2 Pitfalls of Passive Treatment Systems for Mine Remediation
In general, passive treatment systems offer a low-cost, partially self-sustaining
method for metal removal and acidity neutralization. However, there are some pitfalls in
these systems that have been discovered. Two potential problems with passive
treatment systems used to treat mining influenced waters are:
1. The clogging of pores due to the breakdown of organic particles that fill available
pore space and,
2. The accumulation of biofilms. Biofilms are a by-product of the microbial activity
contained within the treatment system, and can lead to clogging and a reduction
in the hydraulic conductivity of the substrate.
The reduction in hydraulic conductivity is a major problem with passive treatment
systems. One reason for this reduction is the accumulation of biomass and biofilms that
may clog pore spaces and reduce the permeability of the substrate (Rockhold et al.
2002, Willow & Cohen 2003, Jong & Parry 2003, Tsukamoto et al. 2004). It has been
reported that microbial activity has the ability to reduce the hydraulic conductivity of the
system by orders of magnitude (Rockhold et a). 2002, Taylor & Jaffe 1990). Taylor and
Jaffe (1990), estimated that the microbial activity could reduce the hydraulic conductivity
of a system by up to 3.5 orders of magnitude. Reduction of hydraulic conductivity
reaches its maximum when a balance is formed between the growth and accumulation
of biomass, with the amount of biomass that is removed due to fluid shearing in the
water phase (Taylor & Jaffe 1990).
There are two ways that bacteria attach and accumulate in saturated systems
(Rockhold et al. 2002). Bacteria can form biomass in discrete colonies through the
aggregation of cells or in larger, continuous biofilms. Growth patterns in porous media
are likely a combination of these two processes, where microbes initially grow in
|
Colorado School of Mines
|
17
colonies and gradually expand to form continuous biofilms (Clement et al. 1996).
Numerous models have been developed that incorporate either one or both of these
forms of bio-clogging including the work done by Clement et al. (1996) and Vandevivere
& Baveye (1992). These models show how biomass in the forms of cell aggregates and
biofilms impact the systems porosity, specific surface area, and permeability.
The characteristics of biofilms and aggregates of biomass are determined by
numerous factors. One dominating factor is the particular strain of bacteria that is living
within the system. The rate at which the bacterial colony grows, accumulates and
decays is dependant on the type of bacteria. Another factor that affects the makeup of
biofilms in passive systems is the type of substrate used. Biofilms are likely to
accumulate more rapidly in systems containing substrate composed primarily of
degradable organic material, due to the activity of heterotrophic bacteria.
Other factors that affect the makeup and properties of biomass aggregates and
biofilms are the chemistry of the water entering the system as well as the rate at which
the water moves through the system. Influent waters containing high levels of iron can
increase the amount of clogging that takes place in the system. Iron precipitates may
form within the system that cause additional clogging or in some cases even an increase
in biomass accumulation. Additional clogging of pore spaces may occur as a result of
metal sulfide solids that precipitate from solution during the treatment of MIW.
The flow rate of influent water into a passive system must be calibrated to the
volume of substrate to achieve a retention time long enough for treatment to occur. If
the flow rate of a system is too high, the water will not be in contact with the substrate
long enough for water treatment. High flow rates can also have a flushing effect;
removing blockages and clogs from the system. However, if flushing occurs at too high
a rate, microbial growth may be inhibited, treatment efficiency reduced, and microbes
may be flushed out of the system (Cunningham et al. 1991).
|
Colorado School of Mines
|
18
Rittmann (1993) determined that no matter what the characteristics of the
attached biomass the net accumulation is controlled by four processes. These four
processes can be defined as four basic life stages and include growth, deposition, decay
or death, and detachment. Biomass growth is assumed to be proportional to the amount
of biomass that is already contained in the system as well as the rate of substrate
utilization. Biomass deposition onto the system depends on the physiochemical
properties of the water as well as the concentration of biomass within the aqueous
phase. The rate at which the biomass decays or dies off is proportional to the amount of
active biomass within the system. Biomass detachment depends on the amount of
active biomass that is attached to the substrate, the growth rate of the biomass, and the
hydrodynamic shear stress applied to the biomass (Rockhold et al. 2002). While these
stages outline the fundamentals of the life cycle of the biomass, all of these processes
vary considerably depending on the particular strains of bacteria living in the system.
In order for a biomass to be considered at steady state, growth of new biomass
through substrate utilization must be balanced by biomass losses, shown in equation 6.
where Smjn is the minimum substrate concentration to support a steady state biofilms; K
is the half maximum rate concentration (ML'3); qmis the maximum specific rate of
substrate utilization (M2T1); Y is the true yield coefficient; and b’ is the overall specific
loss rate for the biofilms (T1) (Rittman 1993).
When the hydraulic conductivity of the substrate is reduced, many changes take
place in the system that can dramatically affect treatment capabilities. The most
common affect of reduced hydraulic conductivity is lower treatment efficiencies. This
occurs when the reduction of hydraulic conductivity results in short-circuiting. Short
circuiting occurs when the substrate in the system functions like an impermeable solid
|
Colorado School of Mines
|
19
rather than a porous media. This forces the water to find alternate route through the
system, the path of least resistance, which can lead to the formation of preferential flow
paths. Water may flow along the walls of a reactor or treatment cell and is not effectively
treated. Therefore, reduction in hydraulic conductivity could reduce the treatment
efficiency of a passive system.
Low hydraulic conductivity can cause the trapping of sulfides within the treatment
system. In systems with water moving effectively through the substrate, excess sulfides
are transported through the substrate and can be released from the system. Increased
sulfide levels in passive treatment systems can have negative impacts on microbial life.
Numerous laboratory studies showed that high concentrations of sulfides can inhibit
sulfate-reducing bacteria (Reis et al. 1992, Utgikar et al. 2002). Reis et al. (1992) found
that inhibition of SRB occurred as a direct consequence of sulfide produced during
sulfate reduction. Without allowing the sulfides to escape from the substrate, the pH of
the system could not increase and microbial activity slowed down impacting the overall
treatment efficiency.
2.5 Anaerobic Passive Treatment
There are several passive treatment options for treating mining influenced waters
that rely on bioremediation. These systems allow for natural processes to be optimized,
improving metal and sulfate reduction. Both anaerobic constructed wetlands and
bioreactors have been studied extensively at the laboratory scale and have been used in
field-scale applications of MIW treatment.
2.5.1 Constructed Wetlands
Over the past several decades, an increased number of passive treatment
wetlands have been built to treat mining influenced waters. Natural wetlands have been
|
Colorado School of Mines
|
20
shown to partially treat metal laden acidic waters and are often sinks for contaminants
within the environment (Ganse 1993). As a result, natural wetlands have been
thoroughly investigated with regards to their potential to treat mine drainage. After many
attempts were made to use natural wetlands systems to treat MIW, the idea of
constructed wetlands began to dominate. The construction of arwetlands system
permitted optimization of specific treatment processes. The dimensions of constructed
wetlands could also be designed to meet specific site characteristics.
Constructed wetlands for remediation of MIW have become a common treatment
method. Much of the pioneering work in the development and engineering of wetlands
to treat mine water discharges was done by the U.S. Bureau of Mines (Johnson and
Hallberg 2005a). Originally, constructed wetlands were developed to treat coal mine
drainages in the Appalachia region of the United States (Johnson and Hallberg 2005a).
The Tennessee Valley Authority also contributed greatly to the development of
constructed wetlands to treat mine drainage (Ganse 1993).
After several successful implementations in treating coal mine drainage,
constructed wetland research expanded into treating drainage from metal mines. While
the basic principles of wetlands treatment can be applied to both coal and metal mines,
there are several differences that require different components of the treatment to be
optimized. Chemical concentrations in metal mine effluents are often greater than those
in coal mine drainage. There are often differences in climate and topography associated
with coal and metal mines. Cooler temperatures at high elevation metal mines will cause
microbial activity in the system to slow down, inhibiting further microbial growth and the
overall effectiveness of metal removal. In order to assist in the proper design of
constructed wetland systems in high altitude regions a number of design guidelines have
been published (Brodie 1990, Cohen & Staub 1992, Ganse 1993, Wildeman & Gusek
1998). One of the first constructed wetlands built in Colorado to treat the effluent from a
|
Colorado School of Mines
|
21
metal mine was the Big Five pilot wetland located in Idaho Springs (Machemer et al.
1993, Cohen and Staub 1992).
2.5.2 Bioreactors
A commonly used type of constructed wetlands is the anaerobic system. These
systems are often referred to as anaerobic treatment cells or bioreactors. Mine drainage
flow can be directed into the system through underground pipes in order to force the
water to flow through the substrate. In upflow anaerobic bioreactors, the contaminated
water flows into a subsurface distribution system, then up through the substrate that
usually consists of organic material. In order for bioreactors to operate efficiently, alkali
generation and removal of metals as sulfides and other insoluble phases are the two key
processes that must occur (Johnson & Hallberg 2005b). As the contaminated water
moves through the substrate, heavy metals in the mine water form complexes with other
available ions such as S2‘ or HS", both of which are very reactive. Ideally, the resulting
effluent water from bioreactors has a reduced concentration of heavy metals as well as a
higher pH compared to the influent.
Anaerobic bioreactors are modeled after anaerobic wetlands, but are more
amendable to design modifications for treatment optimization. Bioreactors are simpler to
simulate in the lab at bench-scale than are constructed wetlands. It has been suggested
that the volume of a system is more important than the surface area in terms of
promoting metal removal and pH neutralization (Cohen & Staub 1992). A larger volume
of anaerobic organic substrate in a bioreactor results in a larger number of SRB and
more effective removal rates (Cohen & Staub 1992). Based on this theory, anaerobic
bioreactors can be designed in tower configurations to minimize the footprint of the
system (Cohen& Staub 1992). Multiple bioreactors can also be used in series or parallel
in order to adapt to the spatial limitations of a site (Dvorak et al. 1992).
|
Colorado School of Mines
|
22
Focusing on the optimization of the SRB sulfate reduction process within
anaerobic systems could cause bioreactors to decrease in size but not treatment
efficiency. Smaller bioreactors will help to expand the use of anaerobic passive
treatment to sites where space is a limiting factor. Using design configurations that
focus on substrate volume rather than surface area, it is possible that anaerobic
bioreactors could be constructed directly within mine workings at sites that do not have
sufficient area to support a large treatment footprint. Installing the system within the
mine workings may provide additional protection from extreme weather conditions,
allowing the system to perform better throughout the seasons (Cohen & Staub 1992).
A potential problem in bioreactors is a shortage of essential nutrients for the
microbial population. If the microbes do not have required nutrients and optimal
conditions for survival, microbial metabolisms may not be high enough to successfully
treat mine drainage. Biological processes can be affected by a wide variety of
environmental conditions. One environmental variable that can affect the microbial
activity and resulting treatment efficiency is temperature (Johnson & Hallberg 2002).
Generally, the activity of microorganisms tends to slow as temperatures decrease and
speed up as temperatures increase. The influence of temperature on the microbial
activity could lead to seasonal and climatic treatment variability.
2.6 Sulfate-Reducing Bacteria
Many passive treatment systems rely on microbes living within the substrate to
reduce acidity and metal concentrations of the influent. More specifically, these metal-
reducing microbes are often sulfate-reducing bacteria (SRB). These obligate anaerobes
have the ability to reduce sulfate in the water to sulfides that complex with heavy metal
ions in the water. Once the metal sulfides are formed they can precipitate out of
solution. This process helps to reduce the concentrations of heavy metals while
|
Colorado School of Mines
|
23
simultaneously reducing sulfate levels. This can be a benefit when applying a passive
biological remediation strategy to mine reclamation, as very high levels of sulfate are
common to mine drainages. As a result, research into SRB and their use for treatment
of mining influenced waters has increased.
Under anaerobic conditions, SRB oxidize simple organic compounds by utilizing
sulfate as an electron acceptor (Reis et al. 1992, Jong & Parry 2003) while the substrate
in the system functions as the electron donor (Neculita et al. 2007). The energy
produced from this oxidation/reduction reaction is used by the SRB for growth and
development (Neculita et al. 2007). The SRB sulfate reduction reaction is generally
expressed by the following equation (Webb et al. 1998):
(7) 2CH20 + SO*- -*H2S + 2HCO3
where CH20 represents a generic organic carbon source. This reduction process
requires two moles of organic carbon for every mole of sulfate reduced. Depending on
the pH of the system, the hydrogen sulfide and bicarbonate ions equilibrate into a
mixture of H2S, HS", S2', and C02, HCQ3", CO2" (Cocos et al. 2002). Excess H2S gas
formed during sulfate reduction is released into the atmosphere, further reducing the
acidity of the water. Waters are especially susceptible to the formation of hydrogen
sulfide gas at pH values £7.0 (Machemer et al. 1993).
The bicarbonate formed through the SRB sulfate reduction reaction causes an
increase in alkalinity. The formation of the bicarbonate is important because it
demonstrates that once SRB are established, they can condition their own
microenvironment by raising the pH (Cohen & Staub 1992).
Soluble sulfides commonly found in natural water systems, such as H2S, HS",
and S2" react with metals in mine water and form insoluble metal sulfide precipitates.
|
Colorado School of Mines
|
24
ARD formation is essentially reversed as dissolved sulfate ions precipitate metals as
metal sulfide complexes. This process can be expressed through the following reaction:
(8) S2' +M2+ ->MS +
where M is representative of a cationic metal such as Cd, Fe, Ni, Ou, Pb and Zn (Webb
et al. 1998, Dvorak et al. 1992, Gibert et al. 2002, Neculita et al., 2007). The
precipitation of metal sulfides improves the water quality by decreasing the mineral
acidity without causing a parallel increase in proton acidity (Sheoran & Sheoran 2006).
In natural waters H2S can dissociate into sulfides and hydrogen ions, shown through
reaction 9.
(9) H2S-»2H+ + S2_
Any protons released into solution from the disassociation of H2S are neutralized by an
equal release of HC03" during sulfate reduction (Sheoran & Sheoran 2006). As the pH
of the water increases towards neutral, conditions favor the precipitation of metal
carbonate complexes that helps to remove more heavy metals from solution (Zagury et
al. 2006).
2.6.1 Limiting Factors for SRB
There are several factors in anaerobic bioreactors that can limit the amount of
mine drainage that can be treated effectively. The available carbon source for the SRB
can be a limiting factor because the sulfate-reducers need a form of readily obtainable
carbon in the form of volatile organic acids in order to grow, fix sulfate and treat water.
The availability of a sufficient carbon source is the most critical limiting factor for the
microbial activity within a bioreactor (Gilbert et al. 2004, Zagury et al. 2006). The chosen
substrate will determine the lifetime of the bioreactor based on how long it can supply
the proper sources of energy for the SRB (Tsukamoto et al. 2004).
|
Colorado School of Mines
|
25
Both low and high pH levels have been found to have negative impacts on SRB.
Low pH levels have been reported as the most significant limiting factor in many sulfate-
reducing systems (Garcia et al. 2001). SRB are highly sensitive to mild acidity as well
as molecular oxygen (Johnson and Hallberg 2005a). Low pH levels will inhibit the
growth of SRB, therefore, fewer microbes will be available to treat metals. Not only does
a low pH inhibit sulfate reduction by SRB but it also increases the chemical instability
and solubility of metal sulfides that may have been formed (Dvorak etal. 1992, Willow &
Cohen 2003). However, if the pH of the system is high or basic, the accumulation of
hydrogen sulfide may inhibit SRB (Reis et al. 1992). Anaerobic systems that receive
near neutral waters, between pH 5.0 and 8.0, provide the best conditions for SRB
survival (Cohen & Staub 1992).
Changes in the rate of sulfate reduction by SRB are often attributed to
fluctuations in temperature (Drury 2000, Benner et al. 2002, Tsukamoto et al. 2004).
SRB activity decreases with low temperatures also decreasing the rate of sulfate
reduction. This in turn results in lower metal treatment efficiency and acid neutralization.
In order to compensate the reduction in individual activity levels, SRB have
demonstrated the ability to increase in their numbers during cold weather (Cohen &
Staub 1992). Sulfate reduction is a rate-limiting step in iron removal and therefore
varying temperatures can also potentially influence Fe concentrations by changing the
solubility of the precipitating Fe sulfide (Benner et al. 2002).
2.6.2 Uranium Contamination and the Environment
Uranium, as an environmental contaminant, can come from a variety of sources,
some natural and some manmade, and can have a significant affect on the quality of
both surface and ground water systems. Ground water contaminated with uranium is of
particular concern because oxidized uranium is toxic, soluble in ground water, and
|
Colorado School of Mines
|
26
mobile in the subsurface (Anderson et al. 2003). Typical sources of uranium
contamination of water include: 1) natural uranium deposits, 2) uranium mining, milling
and processing operations, 3) remediation of uranium contaminated soils that produce
large volumes of uranium bearing leachate, 4) infiltration and snowmelt into mine
workings, waste rock and tailings, 5) some irrigation practices, and 6) uranium
processing (Lovely & Phillips 1992b, Spear et al. 1999, Tucker et al. 1996, Wildeman et
al. 1994). In areas where uranium has been mined or milled, it is common to see
elevated concentrations of soluble uranium that exceed acceptable levels for health for
humans and wildlife (Tucker et al. 1996). Many inactive mines have contributed uranium
loading in surface and ground waters for decades and the prospect of remediation is just
beginning to be considered at some sites.
While current remediation techniques for treating uranium contaminated waters
exist, all have their limitations. The method most commonly used for the treatment of
water contaminated with uranium is the use of ion exchange resins. This method can be
limited however by the cost of materials, interferences with competing ions, and poor
extraction at low uranium concentrations. The use of ion exchange resin for uranium
treatment produces large volumes of waste if the resin is disposed of, or highly corrosive
uranium-containing waste if the uranium is extracted from the resin (Lovely & Phillips
1992a).
Other common uranium treatment techniques include lime softening,
conventional coagulation, and activated alumnia. Uranium treatment usually consists of
using one or a combination of these techniques; the main problem with these treatment
methods is the volume of waste produced has the potential to be significant and costly to
dispose of. Recent studies show that in order to treat low levels of uranium
contamination other unit processes could be involved in the treatment train (Spear et al.
1999). These processes could include rapid mix, flocculation, sedimentation, filtration
|
Colorado School of Mines
|
27
and pH adjustment. Most of these methods require large amounts of chemical doses,
frequent replacement or regeneration of the treatment materials, as well as an increased
footprint of the treatment train, limiting their implementation. These treatment methods
also require frequent maintenance which may not be possible at all sites and would
require the chemicals to be hauled over long distances, thus increasing the cost of the
entire treatment strategy.
Often these treatment options are too expensive and lack the specificity to treat
uranium against a background of competing metal ions. Many of these treatment
techniques are not applicable to full scale applications; they only perform well under
laboratory conditions. In order to be successful in treating uranium contaminated mine
waters, treatment technologies need to be able to function at a high level of efficiency for
long periods with little maintenance requirements. Microbiological approaches to
uranium treatment offer highly selective removal potentials as well as operational
flexibilities. It is possible that emerging biological treatment schemes for uranium can be
used either in situ or ex situ, allowing these techniques to be implemented in a broad
range of applications (Lloyd & Lovely 2001). It has been proposed that using in situ
methods to immobilize uranium may result in better and more efficient treatment due to
the fact that it takes advantage of the redox character of uranium (Anderson et al. 2003).
Microbial mediated precipitation may be a viable technique for uranium removal from
aqueous solutions.
2.6.2.1 Uranium Treatment using Sulfide Production
Uranium exists in the natural environment as U(IV), U(V), and U(VI). U(VI),
hexavalent uranium, is highly soluble and is the water mobile valence state of uranium.
U(VI) exists in solution as the ion group U022+. This group forms soluble, negatively
charged complexes with C032' such as U(O2(CO3)2^ and UO2(CO3)3^. These different
|
Colorado School of Mines
|
28
uranyl-carbonate complexes are the most common form of U(VI) that occur in natural
surface and ground water systems. In oxic water with elevated C032' alkalinities, these
complexes will dominate the U(VI) spéciation. U(IV), tetravalent uranium, is sparingly
soluble and found in the environment in the form of uraninite, U02. Uraninite is the most
common uranium mineral that occurs naturally, and is found in anoxic sediments and
aquifers. Often U(IV) is referred to as reduced uranium as it is usually the end product in
the reduction of uranium in the environment. U(V) is not a concern for remediation
because it usually does not dominate the uranium spéciation in natural waters.
The most promising microbes to use for the treatment of uranium contaminated
waters are sulfate-reducing bacteria (SRB). There are certain forms of SRB that can
reduce the highly soluble U(VI) to the sparingly soluble U(IV) (Lovely et al. 1991, Gorby
& Lovely 1992, Lovely & Phillips 1992a & 1992b, Lovely et al. 1993, Spear et al. 1999).
By reducing the U(VI) to U(IV) within an aquifer, it is possible to precipitate uranium,
preventing it from moving downstream and spreading uranium contamination. There has
been much research done to determine which species of SRB are the best to use under
different environmental conditions (Lovely & Phillips 199T, Gorby & Lovely 1992, Lovely
& Phillips 1992a & 1992b, Lovely et al. 1993, Spear et al. 1999). Scientists have also
begun to describe and model the kinetics behind the reduction of U(VI). Sulfate-
reducing bacteria have the potential to be of economic, environmental, and biotechnical
importance as more is learned about how to culture the correct species of SRB to treat
uranium.
There are several advantages to using sulfate-reducing bacteria to treat uranium.
Since there is sulfate naturally occurring in the environment, there is usually enough
available to provide energy for the U(VI)-reducing SRB. This ability generally allows for
the SRB to grow in subsurface media. The reduction process occurs rapidly and does
not require many supplements to the medium.
|
Colorado School of Mines
|
29
2.6.2.2 Enzymatic Treatment of Uranium
It was previously known that some anaerobic microorganisms can reduce metals
in the aqueous phase, such as Fe(lll). However, it was not until Lovely et al. (1991)
showed that some of these microorganisms also have the ability to reduce U(VI). Under
further examination of the process involved with the microbial reduction of U(VI), Gorby
and Lovely (1992) found that it is actually an enzymatic process. The primary difference
between enzymatic U(VI) reduction and previously proposed biological mechanisms for
U(VI) removal is that the U(VI) reacts directly with a U(VI)-reducing enzyme within the
U(VI)-reducing microorganism.
Lovely and Phillips concluded that sulfate and U(VI) could be reduced
simultaneously by D. desulfuricans (Lovely & Phillips 1992b). With the discovery that
U(VI) and sulfate could be reduced at the same time, new doors opened to microbial
treatment technologies. This finding suggested that perhaps bioreactors containing D.
desulfuricans could be established to treat uranium contaminated waters if they were
amended with low concentrations of sulfate in order to provide an energy source for the
U(VI)-reducers during the process of sulfate reduction. An additional benefit of this kind
of treatment was that U(IV) is stable in the presence of sulfate. Another key finding of
this work was that the U(VI)-reducing enzyme in D. desulfuricans is not irreversibly
inhibited by exposure to atmospheric oxygen.
Lovely and Phillips (1992b) tested the ability of D. desulfuricans to reduce
uranium concentrations in the effluent from an inactive uranium mine. D. desulfuricans
quickly converted the high concentrations of U(VI) to U(IV). The U(IV) precipitates that
formed during the start up of treatment were able to pass through a 0.2 pm filter,
indicating the initial product of the U(VI) reduction was either soluble or colloidal (Lovely
& Phillips 1992b). Therefore since large, insoluble U(IV) particles do not form
ARTHUR LAKES LIBRARY
goLldenDcoS 8Ho2oiL 0F MINES
|
Colorado School of Mines
|
31
3.0 DESIGN AND OPERATIONAL CONSIDERATIONS FOR ANAEROBIC
BIOREACTORS
Anaerobic passive treatment systems offer an economical way to treat mining
influenced waters. While these systems have numerous benefits there are several
factors involved in the design that must be considered prior to the installation of an
anaerobic passive system.
3.1 Metal Removal Mechanisms
There are a number of ways that metals can be treated and removed from MIW
using bioreactors. The process demonstrated to be the primary removal mechanism is
sulfate reduction performed by sulfate-reducing bacteria (SRB). The microbial
community contained within the substrate of bioreactors contains a diversity of bacteria.
Sulfate-reducers are ubiquitous in anaerobic passive bioreactors. SRB reduce the
sulfate in MIW to sulfide and then the metal ions form insoluble complexes with the
sulfide. Common metal sulfides produced during this process include zinc, cadmium,
iron, lead, and copper sulfide. Previous research shows that the rate of sulfate reduction
can be an important variable in passive treatment systems receiving acidic waters
(Cohen & Staub 1992, Machemer et al. 1993, Willow & Cohen 2003).
In addition to the biologically mediated processes, the quality of the MIW is
improved by filtration of suspended and colloidal materials and sorption of the metals by
the organic matrix (Johnson & Hallberg 2005b). As the pH increases, metal carbonates
and hydroxides form and begin to precipitate out of solution. These precipitates are then
free to sorb with solids, which can help remove more metals (Gibert et al. 2003).
Sorption has been found to dominate as the primary metal removal process during the
initial stages of anaerobic treatment prior to the acclimation of SRB (Zagury et al. 2006).
|
Colorado School of Mines
|
32
During the start up of a bioreactor system, early metal removal can be attributed
to metal sorption to fresh organic material (Amos & Younger 2003). In order to
determine the extent that sorption contributes to metal removal, the mass of metals
removed per day can be compared to the mass of sulfate reduced per day (Willow &
Cohen 2003). If the amount of sulfate reduced per day is a much lower mass than the
amount of metals removed per day, it is likely that sorption played a key role in the
removal of metals from the mine drainage (Willow & Cohen 2003). Once a SRB
community has been established in the bioreactors, metal removal through sorption will
be out performed by the removal of metals through sulfide precipitation.
3.2 Downflow vs. Upflow Systems
The flow path of influent water through an anaerobic bioreactor can greatly affect
the overall performance and treatment of the system. There are three main types of flow
schemes within a bioreactor. Water can flow horizontally, downward driven by gravity or
upward driven by water pressure. The direction of flow affects the amount of time the
contaminated water is in contact with the reactor substrate. Horizontal flow systems
simulate water flow in natural wetlands. One downfall of a horizontal flow regime, in
many cases, is a large reduction of hydraulic conductivity takes place within the first
weeks of operation (Lemke 1989). In one study done at the Big Five Tunnel in
Colorado, the hydraulic conductivity of the system decreased by two to three orders of
magnitude over the course of the first few weeks (Lemke 1989, Cohen & Staub 1992).
Downflow systems have outperformed upflow systems in terms of pH
enhancement and metal treatment efficiency. However, a problem of using a downflow
system is that the weight of the water on the top of the substrate causes a significant
amount of compaction. Compaction of the substrate can decrease the permeability of
the system and reduce the flow rate of the MIW through the rector. Cohen & Staub
|
Colorado School of Mines
|
33
(1992) noted that in downflow systems, flows that are too low tend to follow channels
through the substrate. Low flows following preferential flow paths can leave a large
percentage of the substrate dry and unutilized. Also, pooled water on the surface can
oxidize, negatively impacting the anaerobic nature of the system. Upflow systems have
been shown to retain their hydraulic conductivity for a longer period of time, which in
turn, ensures that the contaminated water stays in contact with the substrate (Lemke
1989, Cohen & Staub 1992). Another advantage of upflow systems is the substrate
tends to stay saturated, which is important for creating an anaerobic environment within
the bioreactor (Cohen & Staub 1992).
3.3 Residence Time
An important factor that can determine the effectiveness of a passive system is
the amount of time that the water is in contact with the substrate. Bioreactors used to
treat MIW with high levels of sulfate are dependent on retention of water within the
substrate of the reactor for the SRB to carryout the desired conversion to sulfide (Lens et
al. 2002.) While many studies focused on trying to achieve long residence times, as
much as 280 hours, column studies reported the formation of metal sulfides and near
100% metal removal within as little as 16 hours (Cohen & Staub 1992, Willow & Cohen
2003). The length of hydraulic detention time is correlated with the amount of time a
substrate can be used before it needs to be replaced (Drury 2000). The longer the
retention time, the longer the substrate can be used. Therefore using a short hydraulic
retention time in a bioreactor requires the commitment to replenish the carbon and
energy source for the bacteria (Drury 2000).
The hydraulic detention time directly affects the rate of sulfate reduction because
more electrons from the substrate degradation will be transferred to the pore water if the
detention time is long (Drury 2000). Substrates which have been well composted before
|
Colorado School of Mines
|
34
use in the system, outperform substrates with lower initial ages. Younger substrates
degrade quickly compared to older substrates, donating many of their electrons in a
short amount of time (Drury 2000).
There are two primary factors that contribute to the hydraulic residence time of
passive treatment systems: mine drainage flow rate and substrate permeability. The
higher the flow rate into the system, the lower the residence time (Willow & Cohen
2003). Short residence times do not allow adequate time for bacterial activity to produce
sufficient amounts of sulfide to precipitate metals and neutralize acidity (Gibert et al.
2005). In fact, the capacity of a bioreactor to generate alkalinity may be suppressed by
very short residence times (Dvorak et al. 1992). As a result of no alkalinity being
produced, the interior of the substrate could become acidified to the point where
bacterial activity is inhibited.
On the other hand, excessive residence times may supply such low levels of
metals and acidity to the reactor that a majority of the alkalinity and H2S formed within
the substrate will exit the bioreactor unused (Dvorak et al. 1992). The amount of time
that water is retained within a bioreactor must be determined because the rate of metal
removal is directly proportional to flow rate below a threshold detention time (Willow &
Cohen 2003).
The permeability of the substrate is another factor that will determine the
detention time in the substrate and overall treatment efficiency of the system. The
substrate used in bioreactors should be evaluated prior to use in a full-scale treatment
system. If the materials used for substrate are too fine they can become packed closely
together and plugging and blockage of pore spaces may occur. When water is initially
introduced to the substrate some compaction will occur (Lemke 1989). Compaction can
significantly reduce the permeability and hydraulic conductivity of the system (Lemke
1989). During compaction, particles are forced closely together until the substrate acts
|
Colorado School of Mines
|
35
as an impermeable layer and short-circuiting may occur, essentially by-passing
treatment in the system.
3.4 Materials Used for Reactor Substrate
The type of substrate used in sulfate-reducing bioreactors is a determining factor
in the magnitude of the microbial processes and resulting treatment taking place within
the system (Chang et al. 2000, Ganse 1993, Gibert et al. 2003). The substrate used in
the reactors will determine the permeability of the system as well as the amount of
organic carbon sources available. Gibert et al. (2002) determined that the composition
of organic matter is a determinant factor of biotreatment efficiency. Many different
materials have been tested as substrates for sulfate-reducing bioreactors. A substrate
must provide an available and useable organic carbon source, reasonable permeability,
and should be able to donate electrons and nutrients for many years before it needs to
be recharged. The chemical composition of the contaminated water dictates which type
of substrate should be used (Ganse 1993).
Some of the materials that have been tested and examined thoroughly for their
applicability to these systems include livestock manure, mushroom compost, peat moss,
limestone, straw, wood chips, leaf mulch, sawdust, vegetal compost, sewage sludge,
whey, and ryegrass (Ganse 1993, Gazea et al. 1996, Gibert et al. 2002, Gibert et al.
2003, Hemsi et al. 2005). Peat moss and mushroom compost were used initially in
many passive treatment systems. However, over time they were ruled out as potential
substrate for treating mining influenced waters. Peat typically has low levels of nutrients
and a low pH. Mushroom compost has a slightly higher pH than peat, but has low
buffering capacity (Cohen & Staub 1992). Livestock manure is often used as a substrate
component because it contains SRB as a result of the bacteria prevalence in the
intestines of livestock (Christensen et al. 1996).
|
Colorado School of Mines
|
36
When these types of substrates are used alone, they do not significantly promote
the activity of SRB (Christensen et al. 1996, Waybrant et al. 1998 & 2002, Gibert et al.
2003, Zagury et al. 2006, Neculita et al. 2007). However, when more than one organic
carbon source is mixed together much higher sulfate reduction rates have been
observed (Waybrant et al. 1998 & 2002, Zagury et al. 2006, Neculita et al. 2007). When
combining materials, one component in the mixture should be relatively biodegradable,
such as livestock manure, and the other substrate should provide a more complex
carbon source, like wood chips or hay (Neculita et al. 2007). Reportedly, the lower the
content of lignin in the organic substrate, the higher is its biodegradability and capacity
for developing bacterial capacity (Gibert et al. 2004). When multiple types of substrate
are used in combination, the efficiency of the system tends to be greater than using one
single carbon source (Gibert et al. 2002). Also, using processes that slowly release
organic substrates can optimize the SRB activity. A number of alternatives, such as
pelletization or encapsulation of organic substrates, are reportedly under development
(Gibert et al. 2002). These new substrate alternatives have the potential to extend the
lifetime and treatment capacity for systems treating MIW.
Characteristics important to the suitability of a substrate are: percent organic
matter, pH, and alkalinity (Ganse 1993, Lemke 1989). The pH of a suitable substrate
should be neutral or basic for use to remediate acidic or slightly acidic waters (Ganse
1993). Wildeman et al. (1994) recommended to use a substrate with an organic content
ranging from 30% to 50%. Substrates with lower organic contents might not have
sufficient biological matter available for the heterotrophic bacteria within the system
(Ganse 1993). While all of these factors should be taken into consideration when
selecting a substrate for bioreactors, the ultimate choice will be dictated by what is
available locally and chemically suitable for the site of interest (Ganse 1993).
|
Colorado School of Mines
|
37
3.5 Substrate Enhancement
Many different materials have been proposed to help improve the permeability of
organic substrates. Some of these materials include silica and gravel, barley husks,
angular gravel, Perlite, and Vermiculite (Benner et al. 2002, Gibert et al. 2003, Lemke
1989). Willow and Cohen (2003) attempted to increase the permeability of the substrate
by using porous ceramic pellets as a bulking agent. Mixing gravel in with the organic
substrate can dramatically change the hydraulic conductivity of the system. Benner et
al. (2002) found that a 5% increase in the fraction of gravel (40-45%) mixed into compost
consisting of partially degraded leafy materials and wood chunks, increased the
hydraulic conductivity by an order of magnitude.
In addition to substrate amendments, substrate materials which also provide
physical support within in the bioreactor can help to extend the lifetime of the system
from years to decades (Tsukamoto et al. 2004). The physical structure of the substrate
is important because bacteria tend to aggregate and thrive in areas that offer some
physical protection (Lyew & Sheppard 1997). A variety of wood, rocks, and plastic
pieces were used as the substrate matrix in a battery of column experiments performed
by Tsukamoto et al. (2004). He found that larger diameter pieces allowed for larger pore
spaces within the system, which provided for a longer residence time as well as
increased treatment efficiency. Tsukamoto et al. (2004) also found that the smaller
diameter rocks provided for better iron treatment than larger pieces. Improved iron
treatment was attributed to the larger surface area provided by the smaller pieces of
rock, resulting in a larger surface area for microbial attachment and growth. Using larger
materials for the physical matrix allows for maintenance of hydraulic conductivity by
flushing the precipitates out of the system (Tsukamoto et al. 2004). These tests show
that when selecting materials for the substrate matrix in bioreactors, the competing
|
Colorado School of Mines
|
39
4.0 MATERIALS AND METHODS
In order to design and install an effective passive treatment system at the Fair
Day Mine site, a column study was conducted prior to field scale application. The
objective of the column study was to determine the best ratio of substrate and bulking
agent materials to achieve hydraulic conductivity within the treatment system that could
be maintained over time. Materials used for the column experiment were exactly the
same as those used at the field scale treatment system.
The only major difference in the inputs to the systems was the type of water
feeding the system due to difficulties involved in transporting large enough quantities of
water from the field site to the lab. For the column experiment, tap water was pumped
through the substrate, whereas the field scale system used contaminated mine drainage.
In addition, using contaminated mine drainage in the laboratory would create disposal
and handling issues because the water from Fair Day contained uranium.
4.1 Materials Used for Experimental Study
The focus of my study was to enhance long-term hydraulic conductivity in
passive anaerobic bioreactors while simultaneously maintaining high levels of metal
treatment efficiency. Composted steer manure was chosen as the base substrate for
the column reactors. Composted livestock manure has been previously shown to be a
viable substrate for anaerobic reactors, especially when used in conjunction with other
bulking materials (Staub 1992, Cohen & Staub 1992, Cheong et al. 1998). For this
study, Earthgro’s composted steer manure blend was purchased in small, prepackaged
bags (volume per bag equal to 0.03 cubic meters). This blend consisted of natural
organic materials including aged steer manure and compost.
|
Colorado School of Mines
|
41
tubing with an inner diameter of 0.318 cm and an outer diameter of 0.475 cm. The head
check tubes are attached horizontally from the columns in 7.6 cm sections and then
extend upwards parallel with the side of the column, forming a right angle connection
joint. The tubes were placed at the same locations on all six columns at the following
heights from the reactor bottom: 15.7 cm, 30.7 cm, 66 cm and 88.9 cm.
Small pieces of PVC screen were glued into each of the columns so that the
screen covered the openings of the head check tubes. These were installed to filter out
pieces of substrate and prevent physical obstructions within the small head check tubes.
Water was pumped into the columns through the base. Once the substrate was
completely saturated, the water could flow out of the head check tubes. By measuring
the heights of the water levels in each tube and using the head drop from tube to tube in
conjunction with Darcy’s Law, the hydraulic conductivity of each column could be
calculated. Then the hydraulic conductivities for each of the substrate mixtures could be
compared.
Q = flowrate
n = porosity (-)
q h -h A = cross - sectional area (m2 )
( 10) Darcy’s Law: — = -KA — where ^hydraulicconductivity^)
h2 = upper height of water (m)
h1 = lower height of water (m)
L = length of column (m)
The columns operated under upflow conditions. Water was pumped from a 1150 L
storage tank to the base of each column reactor. A 2.5 cm ball valve was located at the
base of each column that allowed any column to be turned off and repaired during the
experiment. A landscape fabric filter (water permeable geomembrane) was installed in
the base of each column sandwiched in between two rubber gaskets. This filter was
|
Colorado School of Mines
|
42
installed to prevent substrate particles from clogging up the inlet located at the base of
the column. Next, a layer of pea gravel, averaging 1.3 cm diameter, was placed in the
bottom of each column. The gravel was installed to function as a distribution system
within the base of the column. Ideally, the gravel would help to evenly distribute the
inflowing water throughout the entire cross-section of the column, rather than allowing
the inflow to create preferential flow paths within the substrate. Another small piece of
landscape fabric was placed inside of the column on top of the gravel layer in order to
block the substrate from filling in the void spaces between the pieces of gravel.
The plumbing system for the column experiment was comprised of mostly 1.9 cm
diameter standard garden hose. Some small sections of vinyl and hard plastic tubing
were used to connect peristaltic pumps to the system. Peristaltic pumps were used over
the course of the entire experiment in order to supply a measurable and consistent
amount of water to the columns. The purpose of the column reactors was to determine
the permeability and flow rates through different substrate and bulking agent
combinations.
|
Colorado School of Mines
|
44
In order to vary the substrate to bulking agent ratios, different percentages of
Profile soil amendment were measured by volume and thoroughly combined with
composted steer manure. The mixtures were wet down completely and poured into the
columns. The column tops were placed over the columns and the substrate was allowed
to settle over night before the system was fully connected to the pumps. The following
morning the columns were topped off with the correct mixtures, in order to fill the space
in the columns that developed during the settling process. The remaining headspace at
the top of each column was filled with a bundle of glass wool, commonly used for
filtration in aquariums. The purpose of the glass wool was to keep the substrate in place
as well as to function as a filter for the water before it exits the column at the top: With
this filter in place it decreased the risk of small particles of substrate plugging the outlet
of the column or the outflow tubes.
The top of each column was secured in place with eight 1.6 cm diameter bolts.
Once the column tops were reattached, the system was connected to the water supply.
A 1150 L plastic storage tank was used to hold the feed water for the system. The feed
lines to the pumps ran out the top of the storage tank, the water was then pumped to the
bottom of each column where it entered the system. The column experiments were
always operated under upflow conditions. Flow into each of the columns was
approximately 10 ml_ per minute yielding a residence time of 35 hours. Ideally, after the
water moved through the entire substrate-filled column, it would exit the column through
an outlet valve at the top. The water flowing out of the columns was directed to the
outflow collection tank; another 1150 L storage tank.
4.2.1 Design A
The original configuration of the column system consisted of one pump for every
column, resulting in six pumps. The pumps first used for the experiment were Variable
|
Colorado School of Mines
|
45
Flow Mini-Pumps, style type Pump III - Medium Flow manufactured by Control
Company. For this portion of the experiment, each of the column reactors was filled with
a different ratio of substrate to bulking agent. One of each of the following mixtures was
added to five columns: all manure, 7%, 11%, 15%, and 21% Profile by volume. The
sixth column was filled with steer manure and 15% alfalfa hay by volume. The alfalfa
hay was added in order to test its ability to increase the permeability and resulting
hydraulic conductivity of the manure substrate. The columns were filled and connected
to the pumps during the first week of May 2006.
4.2.2 Design B
The substrate bulking agent mixture in each of the columns from Design A was left
in the columns and used for this portion of the experiment. Two new pumps were
connected to the system as a replacement for the six pumps used in the previous
design. The new pumps were Adjustable Single Head Pumps, style #4NA12,
manufactured by Grainger. Each pump provided water to a set of three columns.
4.2.3 Design C
For the final column configuration, all of the substrate was removed from the
columns and replaced with fresh material. A diagram of Design C for the column
reactors can be seen in Figure 4.2. While the columns were empty the inner walls of
each column were sanded by hand using coarse sand paper. Sanding the inner column
walls provided a textured surface to reduce the chance of preferential flow paths
developing along the sides of the column. The head check tubes as well as the screen
coverings were replaced in order to avoid clogging due to particle build up over the
course of the prior experiments. The right angle joint pieces on the head check tubes
were replaced with flexible tubing that fit on the outside of the tubes to form a connection
|
Colorado School of Mines
|
46
joint. The advantage of this was that the inner diameter of the flexible tubing was larger
than that of the right angle joints used in Design A. This design feature was an attempt
to prevent small pieces of substrate from clogging the head check tubes. The columns
were refilled in duplicates using new steer manure and Profile mixture ratios during the
first week of January 2007. Two of the columns were filled with all composted steer
manure, two columns contained a 10% Profile mixture and two columns contained a
20% Profile mixture. These substrate ratios match those that were used in the field
scale application of the system.
Another major improvement to the column reactors during this trial was the use of
an additional Grainger pump. The new pump was the same make and model as the two
pumps used in the Design B (Adjustable Single Head Pump, style #4NA12). In the new
setup, each pump provided water to duplicate columns. Therefore, one pump supplied
water to the two all manure columns, another to the 10% Profile columns and the third
pump for the 20% Profile columns.
|
Colorado School of Mines
|
48
4.3 Field Reactors
Due to the elevated levels of uranium and metals in the adit flow at the Fair Day
Mine site, along with the remote location of the site, an anaerobic passive treatment
system using sulfate-reducing bacteria was proposed to treat the adit flow. The pH of
the adit flow averaged between pH values of 4.5 and 7.5. This factor also provided an
ideal situation for anaerobic passive treatment under the assumption that the pH of the
influent water would not kill or limit the growth of the microbial community within the
system.
4.3.1 Design and Installation
At the Fair Day site, seven bioreactors were constructed from plastic 120 L trash
cans. Two holes were drilled in each can to allow for inflow and outflow plumbing to be
installed. All of the bioreactors onsite operated under upflow conditions, similar to the
column reactors. Black acrylonitrile butadiene styrene (ABS) tubing 2.5 cm in diameter
was used to construct the plumbing to the bioreactors. A piece of ABS tubing was
perforated with 0.318 cm holes and placed along the bottom of each reactor. This pipe
functioned as the inflow pipe and was covered with a layer of pea gravel, approximately
7.5 cm thick with pieces averaging 1.3 cm in diameter. The inflow pipe and gravel base
served as the water distribution system for each reactor. Water entering the system
flowed into the perforated pipe and spread out within the gravel layer. In theory, this
dispersion would allow for the water flow to remain constant throughout the entire
reactor and help prevent the formation of preferential flow paths within the substrate. A
layer of landscape fabric was placed over the top of the gravel and continued up the
sides of the cans, essentially serving as a liner for each bioreactor. The landscape
fabric liner was installed in order to prevent small particles of substrate from falling down
|
Colorado School of Mines
|
49
into the gravel and clogging pore spaces. A diagram showing the Fair Day bioreactor
design configuration can be seen in Figure 4.3.
LID
69.53cm
0.64cm
PERFORATED PIPE
OUTFLOW
49.21cm
LANDSCAPE FABRIC
SUBSTRATE (ORGANIC OUTFLOW COVER
LANDSCAPE FABRIC
LINER
0.32cm
PERFORATED PIPE
7.50 cm f GRAVEL DISTRIBUTION WITH CAP
? SYSTEM
INFLOW
6.35 cm
0.00 cm
55.88cm
Figure 4.3: Diagram of Individual Field Bioreactor
The substrate of the Fair Day bioreactors consisted of the same types of
materials used in the column experiments, Earthgro’s composted steer manure and
Profile soil amendment. The Profile was measured out by volume and then mixed into
the steer manure in different ratios. The ratios within the substrate were the same as
those used in Design C of the laboratory column study. Two bioreactors contained
solely steer manure, two contained a 10% Profile mixture, and two contained a 20%
Profile mixture. Since there were seven reactors built onsite, rather than the six column
reactors in the lab, one additional substrate mixture was used. The seventh reactor was
filled with steer manure and 15% alfalfa hay, measured by volume. Alfalfa hay was used
to see if it could improve the hydraulics of the system, as well as supplying a source of
nutrients. Nutrients readily available from alfalfa can jumpstart microbial activity, which
could improve the overall treatment efficiency of the system (Madel 1992).
The reactors were filled with substrate on July 6, 2006. The volumes of each
substrate material were measured out separately and then mixed together. The
|
Colorado School of Mines
|
50
substrate blend was then added to appropriate bioreactor. While in the lab experiments
the substrate was wet down prior to filling the columns, the substrate mixtures were
added to the field reactors dry. Due to the severe compaction problems associated with
loading the wet substrate into the columns, it was decided to try and avoid this problem
by loading the reactors dry.
The adit water was collected behind a cofferdam inside of the adit that was
constructed during the 2005 Forest Service reclamation activities. A pipe running
through the cofferdam directed most of the adit flow underground. This pipe resurfaced
near the feed tank to the system and was routed directly into the base of the 1900 L
plastic storage tank. An overflow pipe was installed on the tank at approximately % of
the height of the tank at 158 cm from the base. Once the water in the tank reached this
height, it would overflow through the pipe and be directed into the uppermost settling
pool. This feature was added to the system in case the flow from the adit was much
greater than expected and filled the tank completely between field maintenance visits.
With the overflow pipe installed, the system could run continuously without fear that the
water was overflowing from the top of the tank and spilling down the hillside. It also
provided a constant hydraulic head to drive the water through the reactors. Once the
adit flow was routed into the tank, it began to accumulate and fill up to the overflow level.
The height of the water remained at this level for the course of the experiment.
The Fair Day Mine effluent was fed to the bioreactors using gravity.
Measurements taken at the site prior to construction and installation of the system
indicated that there was approximately 3.05 meters of elevation head difference between
the bottom of the feed tank and the inlet on the reactors. This hydraulic head was added
to the elevation of the water in the feed tank to give the total hydraulic head. This was
determined to be enough to drive the water into and through the system. A sampling
port was installed in the plumbing near the tank in order to be able to sample the adit
|
Colorado School of Mines
|
52
4.3.2 Sodium Chloride Tracer Test
A sodium chloride tracer test was conducted on the bioreactor system at the
Lower Fair Day Mine site. This test was done to measure the amount of time it took the
tracer solution to travel through each substrate bulking agent mixture. The sodium
chloride tracer was injected into the water line feeding the anaerobic bioreactors. The
tracer solution was made by dissolving 36.3 kg of Morton’s Sodium Chloride Pellets into
19 L of tap water. The solution was mixed and then transported to the field site. The
conductivity of the sodium tracer solution was measured to be 247 millisiemens before it
was injected into the system. A high concentration of sodium chloride was used so that
breakthrough could easily be measured in the outflow of the reactors despite dilution
from the water already contained within the system. The tracer test was performed late
in the field season on October 6-8, 2006. This was done so that the high concentrations
of sodium chloride would not negatively impact or stop microbial activity within the
reactors while the treatment efficiency was being closely monitored.
To inject the tracer into the system, the valve controlling flow from the feed tank
was closed and the main feed line running down the hillside from the collection tank to
the system was disconnected part way down the hill. A 20 L carboy containing the
saturated sodium chloride solution was attached the feed line and allowed to drain
completely into the system. It took approximately two hours for the 19 L of tracer
solution to be completely added to the system. Conductivity meters were used to
measure the relative concentration of ions in the effluent of each bioreactor. The
addition of large amounts of sodium chloride to the water increases the amount of ions in
solution and consequently makes the water significantly more conductive. Conductivity
measurements were taken from each reactor outflow on a regular basis for 50
consecutive hours in an attempt to determine the time of breakthrough for the tracer
|
Colorado School of Mines
|
53
solution. The amount of time the tracer took to move through each reactor could be
used in conjunction with physical properties of the system, such as porosity, flow rate
and hydraulic head, to calculate a hydraulic conductivity for each of the bioreactors.
4.4 Water Sampling
A regular water sampling and monitoring program began at the Fair Day site in
mid-June.2006. USDA Forest Service employees conducted all sampling and onsite
analysis on a weekly basis. Filtered and unfiltered samples were collected at the mine
site and the surrounding water bodies to measure total and dissolved metal
concentrations. In order to fully evaluate the extent of contamination and background
levels of contaminants in the water, seven sampling locations were established.
Samples were collected at the following locations: 1) above the site in the unnamed
tributary, 2) the mine effluent, 3) the lowest settling pool, 4) the unnamed tributary below
the intersection point with the mine effluent, 5) the unnamed tributary below the second
drainage confluence, but above the road, 6) James Creek above the confluence with the
unnamed tributary, and 7) James Creek below the confluence with the unnamed
tributary. Samples were collected regularly from the lowest oxidation, settling pool to
directly compare the overall treatment efficiency of the settling pools to that of the
anaerobic bioreactor system.
All water samples collected from the Fair Day site were grab samples. A plastic
sampling bowl with a spout was used to collect each sample. The sampling bowl was
rinsed with sample water three times before the sample was collected. All of the water
samples collected in the field were analyzed immediately for pH, conductivity and
temperature. The pH and temperature were measured using a Thermo Orion portable
pH meter (Model 250Aplus) and conductivity was measured using a Thermo Orion
portable conductivity meter (Model 135A). All of the meters used for collecting field
|
Colorado School of Mines
|
54
measurements were calibrated following the manufacturers instruction manuals. The pH
and conductivity electrodes were calibrated with the appropriate buffers on the morning
of the field work, prior to collecting or analyzing any samples. All of the electrodes were
rinsed with de-ionized water between the collection of each sample to avoid sample
contamination.
Careful handling procedures were established for the collection of the total and
dissolved water samples. Total samples were stored in plastic 250 mL bottles and were
immediately acidified using 5 mL of 70% nitric acid contained in glass vials which could
be easily transported into the field. Dissolved samples were immediately filtered through
a 0.45 micron filter using a 150 mL Nalgene filtration unit. The filter unit was connected
to a hand vacuum pump. The top of the filtration unit was disposed of after each use.
The filtered sample was then acidified using the same type of nitric acid vials used in the
acidification of the total samples. A sterile lid was included with each filtration unit that
could then be screwed directly onto the sample catch container. Both the total and
dissolved samples were placed in coolers and the coolers were kept in the shade.
Samples were kept in cool storage until they could be analyzed in the lab.
Digital photos and global positioning points (GPS) were collected at each of the
sampling locations. A metal stake with neon colored flagging tape was also placed at
each of the sampling locations in order to assure that the samples were taken at the
same place during each sampling event.
|
Colorado School of Mines
|
55
Table 4.1: GPS Coordinates of Fair Day Sampling Locations
Sampling Location: GPS Data:
James Creek Downstream from Site N 40° 06' 653" W 105° 25' 272" 2376 meters
James Creek Upstream from Site N 40° 06' 647" W 105° 25' 346" 2397 meters
Unnamed Tributary Below Site, Below
N 40° 06' 681" W 105° 25' 334" 2286 meters
Drainage & Above Road
Unnamed Tributary Below Site, Above
N 40° 06' 721" W 105° 25' 339" 2424 meters
Drainage
Unnamed Tributary Above Site N 40° 06' 831" W 105° 25' 435" 2441 meters
Lowest Step Pool N 40° 06' 806" W 105° 25' 394" 2380 meters
Overflow from Tank N 40° 06' 830" W 105° 25' 417" 2351 meters
Valve from Tank to System N 40° 06' 810" W 105° 25' 339" 2473 meters
Bioreactor Treatment System N 40° 06' 820" W 105° 25' 413" 2333 meters
This sampling and monitoring schedule continued once the bioreactor system
was built at the site on July 6, 2006. Once the bioreactors were setup and flowing, total
and dissolved water samples of the effluent from each reactor were collected. These
samples were also analyzed in the field for pH, conductivity, and temperature. The flow
rate of each of the bioreactor effluent was also measured during each sampling event.
4.4.1 Quality Assurance (QA)
In order to limit and identify the error associated with the field samples and
analysis, a quality assurance plan was developed. Each of the sampling locations was
kept consistent throughout the project. All of the field sampling personnel were trained
on proper sample collection, preservation and storage processes. Field sampling logs
were filled out during the course of each sampling event and these logs were put on file
at the Forest Service office. Each of the samples collected was labeled with a unique
identification code as well as the date and time of collection. Other information on the
sample labels included the name of the person whom collected the sample, the type of
preservation used and the laboratory analysis desired. After the samples were analyzed
|
Colorado School of Mines
|
56
in the lab, the resulting data was carefully examined and evaluated for validity before the
data was used.
4.4.2 Quality Control (QC)
Contamination is a common source of error in both sampling and analytical
procedures (EPA 1996). Quality control samples can be used to help identify when and
how samples become contaminated. In order to monitor for possible contamination, QC
samples were collected during each sampling event at Fair Day. QC samples consisted
of two field blanks and two field splits, one total and one dissolved for each. The field
blanks were prepared the exact same way as the other water samples, but de-ionized
water was used rather than water from the field site. Each of the blanks was analyzed
for the same constituents as the field samples. During each sampling event, one of the
samples collected at the site was split into two samples. After laboratory analysis, data
from these two samples were compared to make sure they have very similar or exactly
the same concentrations of metals and uranium. The field split samples also helped to
determine if the samples were being contaminated through the sampling and
transportation process.
4.4.3 Analytical Techniques
All of the samples collected from Fair Day were analyzed for metals, uranium,
and sulfur content using inductively coupled plasma absorbance emission spectroscopy
(ICP-AES) (Perkin Elmer Optima 3000) located at the Colorado School of Mines in
Golden, Colorado. During the ICP analysis, one set of standards including a uranium
standard were analyzed after every twenty experimental samples. When using the ICP,
the samples were analyzed for the following constituents: aluminum, cadmium, calcium,
chloride, chromium, copper, fluoride, iron, lead, magnesium, manganese, nickel,
|
Colorado School of Mines
|
57
potassium, silicon, silver, sodium, sulfur, zinc, and uranium. Sulfate concentrations were
determined by converting the measured concentration of sulfur to sulfate. This
conversion assumes that all of the sulfur ions in the water have speciated to sulfate ions
based on the water chemistry.
4.5 Substrate Characterization
Samples of unused substrate, substrate from the column reactors, and substrate
from the field reactors were analyzed in the laboratory to determine changes in physical
properties of the substrate over time. The column substrates had water flowing through
them for 28 days whereas the field reactor substrates had been functioning with water
for 84 days before the samples were collected. The unused substrate had not been
placed in either system had had not been exposed to water. The specific gravities and
bulk densities measured in the laboratory were used to calculate the porosity, or void
space, of each reactive mixture. The calculated porosities from the unused, column and
field substrates could then be compared to see if it had decreased over time during use.
A reduction in porosity could directly affect the throughput of the system and in turn the
overall treatment efficiency. The porosities calculated from these laboratory tests were
also used to calculate a hydraulic conductivity for each substrate mixture.
The procedures for all of the following tests performed on the substrate samples
were similar to those in a battery of tests performed by Lemke (1989) on a variety of
reactor substrates. The methods and techniques were repeated in this test in order to
have comparable substrate characterization data.
4.5.1 Substrate Sampling Methods
Samples of unused, column, and field substrates were collected for laboratory
analysis. The sampling procedure for each kind of substrate was consistent in the
|
Colorado School of Mines
|
58
quantity of samples collected but due to the different system configurations the actual
technique used to collect each sample varied slightly.
4.5.1.1 Unused Substrate
When the columns were refilled with fresh materials for Design C, extra substrate
was made for each of the mixtures. Samples of each mixture were stored in plastic
containers until they were dried and analyzed in the lab.
4.5.1.2 Column Substrate
In order to collect representative samples of the entire length of each column,
substrate samples were collected in a way similar to taking a soil core sample. A 1.22
meter long piece of 4.45 cm diameter PVC pipe was used to collect the sample by
removing the top of each column and then sliding the PVC tube into the column. Once
the tube was nearly fully submerged in the substrate a PVC cap was placed over the
end of the tube, creating a seal. Then as the PVC tube was pulled out of the column a
“core” of substrate was removed. This large sample was split into three samples for the
substrate characterization tests performed in the laboratory.
4.5.1.3 Field Substrate
Samples were collected from each of the field bioreactors on September 30,
2006. These samples were collected prior to the sodium chloride tracer being run
through the system. Each sample was collected from a depth of approximately 30 cm.
Since the field system was used to treat uranium-contaminated mine drainage for nearly
three months, part of each reactor sample was sent to Paragon Analytics Laboratory in
Fort Collins, Colorado for analysis. This analysis was necessary to determine what kind
of waste the substrate from the field reactors would be classified as and how to properly
|
Colorado School of Mines
|
59
dispose of it after the field study was completed. Uranium concentrations from each of
the reactor substrates were also needed to determine whether or not special handling or
disposal would be necessary to work with the samples at CSM.
4.5.2 Specific Gravity Tests
To determine specific gravity, all of the substrate samples were dried in a 100°C
drying oven for 24 hours. Each dry substrate sample was then added to a cylinder of a
known weight. The weight of the substrate was determined by weighing the cylinder
plus the substrate and calculating the difference. Instead of using water for this test,
kerosene was used. Water could not be used because the density of the substrate was
less than that of water and the substrate would float on the surface. However, the
substrate was denser than the kerosene which allowed it to stay at the bottom of the
cylinder and the resulting displacement could be measured. Kerosene was measured to
a known volume using a graduated cylinder and then added to a dry substrate sample.
A vacuum was then applied to the cylinder containing the kerosene and substrate. The
vacuum remained until there were no air bubbles rising from the substrate samples, at
which point it was assumed that the kerosene had filled all of the void spaces within the
substrate.
The total volume of the substrate sample and the kerosene could be directly
measured from the graduated cylinder. The known volume of the kerosene was
subtracted from the total volume of the mixture, yielding the volume of the substrate.
Using the volume and weight of the substrate sample as well as the density of the
kerosene, the specific gravity of each dry substrate sample could be simply calculated
by the following equations:
(11) - Weight 0f Su-b-Strate <8? ■ ■ - Density of Substrate f - U
Volume of Substrate (cm3) vcm J
|
Colorado School of Mines
|
60
Density of Substratef y 3 j
(12) f ^ \J = ^Pec'^c Gravity of Substrate (-)
Density of Kerosene
The density of kerosene was assumed to be 0.81 g/cm3 for the entire battery of
specific gravity tests performed (Sunnyside Corporation 2007).
4.5.3 Bulk Density Tests
The bulk density of each of the substrate mixtures was measured by weighing an
empty beaker and then weighing it again after a sample of wet substrate was added.
The difference in the two measured weights yielded the weight of the substrate sample.
The volume of the substrate sample could be measured by using the cross-sectional
area of the beaker and the height of the sample. Then using the weight and volume of
each sample, the bulk density of the substrate could be calculated as weight per unit
volume of sample.
(13) Weight of the Wet Substrate (g) = Bu|k Densjty of ,he Wet substrate ( V /]
Volume of the Wet Substrate (cm ) v cm J
4.5.4 Organic Content Analysis
To determine the percent of organic matter contained in each substrate mixture
the dry weight was compared to the ash weight. To do this, each sample was dried in a
drying oven for 24 hours at 100°C and then allowed to cool to room temperature in a
desiccator. The dry samples were weighed and placed into ceramic crucibles of known
weights. The crucibles containing the samples were then placed in a muffle oven and
combusted for four hours at 550°C. Once the samples were cool enough to remove
from the muffle oven they were placed into a desiccator to cool to room temperature.
|
Colorado School of Mines
|
61
Once cool, the samples were weighed again. The difference between the dry weight
and the ash weight is indicative of the amount of organic matter in each sample (Lemke
1989). The percent of organic matter contained within each substrate was calculated by
the following equation:
(14) Dry Weight (g)—Ash Weight (g) x1qq _ percent Organic Matter by Weight (%)
Dry Weight (g)
The amount of organic material contained in each substrate mixture was compared
to determine if adding Profile significantly reduced the amount of organic matter and
decreased the treatment efficiency of the system.
4.5.5 Size Particle Distribution
In order to determine if the substrate particle size distribution changed overtime
with use, unused substrate samples as well as column and field reactor substrate
samples were dried thoroughly and sieved. All samples were dried in a drying oven at
100°C for 24 hours. The samples were then moved to a desiccator and allowed to cool.
Each sample was sieved through a series of four sieves. Sieves with mesh sizes #10,
#20, #100, and #200 were used. These size meshes retain particle sizes of 2.0, 0.85,
0.15, and 0.075 mm diameter particles, respectively. After the sample was worked
through the sieves, the material retained on each sieve was collected and weighed. The
materials finer than 0.075 mm in diameter that passed through the #200 mesh sieve
were also collected and weighed. All of the size fractions weights were added together
to determine a total weight for each sample. Then each size fraction weight was divided
by the total weight of the sample in order to calculate the percent of the dry weight by
size fraction. By sieving substrate samples that were unused as well as substrate from
|
Colorado School of Mines
|
65
5.2 Column Reactors
The column reactor experiments were initially designed to test a variety of
substrate bulking agent mixtures prior to use in the field system. The objective of the
experiments was to collect data for a minimum of one month and calculate hydraulic
conductivity values for each substrate bulking agent mixture. The results would then be
used to determine which mixtures would work best in bioreactors at the Fair Day Mine
site. However, many problems were encountered with the design of the system. As a
result, three designs (A, B, & C) were tested, each trying to resolve the issues from the
previous design.
Several technical difficulties occurred while working with the original set of pumps
in Design A. The silicone tubing on the peristaltic pumps sprung leaks on a daily basis.
In addition to the leaking pumps, the columns leaked from the bases and the head check
tubes continuously overflowed. These problems indicated that the system was not
airtight and therefore the system may not have been anaerobic. As a result, changes in
pressure head determined by the difference in water levels of the head check tubes
were not accurate representations of the actual pressure changes taking place within the
system.
Work on the column experiments continued throughout the summer of 2006.
Many adjustments were made to the system and the silicone tubing on the failing pumps
was continually replaced. Despite many improvements, the system continued to leak
and overflow. While the initial intent of the column experiments was to determine which
substrate mixture ratios provide the greatest hydraulic conductivity and should be
applied in the field scale system, I decided to put the column study on hold in order take
advantage of the entire field season. Site access to the Fair Day Mine was limited due
|
Colorado School of Mines
|
66
to the large snowdrifts that accumulate in the area. As soon as the access road was
passable, work began on the construction of the field bioreactors.
After the field scale system was constructed and operating, some attention was
refocused on the column experiment and Design B was created. In an attempt to fix
leaks due to pump failure, two new pumps were connected to the system on Tuesday,
September 19, 2006. This setup resulted in other problems and system failures.
Continuous overflow from the head check tubes continued to occur. The reason for this
was that each pump was connected to three different columns, each consisting of a
different substrate mixture ratio and resulting pressures. For both sets of three columns,
the head check tubes on the columns with the most porous substrate mixtures were the
tubes that continuously overflowed. This behavior lead to the conclusion that within
each set of columns the water was finding the path of least resistance and escaping
from the system.
A final attempt was made to collect pressure data from the columns. For Design
C, all of the columns were completely emptied, thoroughly cleaned, and refilled with
fresh materials. Once the system was turned on it was left to stabilize for one
continuous month. During this time, the head check tubes overflowed nearly constantly.
An overflow collection system was designed to run the pumps continuously without
having to be monitored. It appeared that the system was beginning to stabilize over the
course of the first few weeks as the amount of water that made its way completely
through the system and into the outflow tank increased. By the third and fourth weeks of
operation, the amount of water making it all the way through the column system
decreased. After trying two versions of pumps and three different system configurations
no reliable data was collected from the column experiment, I decided to shut down the
system.
|
Colorado School of Mines
|
67
The underlying problem was that the composted steer manure blend used as the
base substrate was too finely ground. These small particles of organic matter tended to
clog the small diameter head check tubes, especially at the right angle joints. It is
possible that these small particles also clogged some of the pore spaces within the
substrate, causing it to be nearly impermeable. I assumed that the pressure difference
between the top and bottom of the columns was too large to measure using the installed
head check tubes due to the continuous overflow from the columns. From an
operational point of view, the clear columns were easier to work with and made it
possible to observe changes in flow pathways and the color of the substrate over time.
Another process that may have caused the substrate in the column to act as an
impermeable mass is compaction. All of the substrate mixtures were thoroughly wet
down before being placed in the columns. This was done to help prevent significant
compaction of the substrate once water was introduced to the column. However, the
manner in which the wet substrate mixtures were physically added to the column may
have been counter productive in terms of preventing compaction of the materials. The
wet substrate mixtures were poured into the top of each column and allowed to fall
through the length of the column until hitting the base. The force exerted by the wet,
falling substrate on the substrate already resting in the column caused more compaction
to occur within the column. As the column began to fill with substrate, the falling
distance of the substrate being added at the top of the column decreased. Therefore the
resulting force of the falling substrate on that below it also decreased as the column filled
up.
This process may have caused the substrate near the base of each column to be
compacted significantly more than the substrate near the top of the column. As a result
the water entering the column through the base could not push through the severely
compacted substrates and was forced to find a flow path with less resistance. In this
|
Colorado School of Mines
|
68
column scenario the path of least resistance was along the interior wall of each column.
Water flowing along this flow path could easily travel up the column wall until it reached
a head check tube and could escape from the column. The columns could not stabilize
due to these preferential flow paths along the sides of the column and out the head
check tubes, causing the column reactors to continuously overflow. Perhaps reducing
the size of the outflow hose from each column would help. The nipples that connected
the outflow hose to the top of each column were permanently attached, but could be
stepped down in size in order to reduce the inner diameter of the outflow tubes. The use
of smaller diameter outflow tubes may help more water to make it through the entire
column before escaping.
5.3 Field Reactors
The field reactors operated from July 6 to November 3, 2006 for a total of 119
days. During this time, twelve sampling events were conducted at the site that allowed
for the treatment efficiency of each reactor to be monitored on a regular basis. Four
days after the bioreactors were installed the flow rate through each of the reactors was
set to equal 60 mL per minute, yielding a residence time of 35 hours. Once the flow
through each reactor was stabilized and equal, the inflow valves were not changed,
allowing all of the reactors to begin with the same flow rate. The changes in the flow
rate through each reactor could then be observed over time. Throughout the course of
the study, without the reactor inflow valves being adjusted, the outflow from some
reactors would increase significantly while others would stop flowing completely. The
cause of the problem was difficult to determine because the bioreactor system was
designed to be a closed system, where outside factors should not influence quantity of
flow through the system. The water level in the storage tank remained at the same
height throughout the experiment, showing that there were not significant changes in the
|
Colorado School of Mines
|
69
amount of flow entering the tank. Therefore, the tank supplied a constant head to the
bioreactors.
Another problem encountered in the field system was the clogging of the reactor
outlet pipes. The pipes were perforated with 0.635 cm holes and covered with
landscape fabric to filter the water and prevent pieces of substrate from clogging the
outlet pipe. While the filters prevented debris from clogging the outlet pipe, the pores of
the landscape fabric became clogged with fine substrate particles and biofilms. Outflow
was prohibited due to the accumulation of materials and in some cases caused the
reactors to completely fill with water. This problem was realized during a sampling event
at the site when a few of the reactors were overflowing with water. In order to remedy
this problem the landscape fabric filters were changed at least every two weeks, and
more often when the fabric appeared to be clogged and preventing water from exiting
the tank.
5.3.1 Metal Removal Efficiency
All three of the substrate bulking agent mixtures used in the Fair Day bioreactors
showed the ability to reduce the metal concentrations as well as neutralize acidity in the
water. The metal treatment efficiency for all of the reactors was variable. The irregular
treatment capacities can be attributed to the drastic changes in flow rates as well as
clogging of pore spaces within the substrate. When either of these problems was
occurring the water did not contact the substrate for the appropriate amount of time and
therefore was not treated sufficiently. The clogging of pore spaces and reduction of
permeability were contributing factors to the variable treatment. The reduction in
permeability of the substrate forced the water to flow around the substrate rather than
through it. As a result, the water may have seeped up the sides of the bioreactors,
avoiding prolonged contact with the substrate and essentially bypassing any treatment.
|
Colorado School of Mines
|
74
5.3.1.2 Zinc
The Fair Day bioreactor system provided high levels of dissolved zinc treatment
throughout the course of the study. The aerobic settling pool system provided some
dissolved zinc removal, averaging 9.45% removal, but was much less efficient than the
bioreactors. The treatment efficiency results for zinc results are shown in Figure 5.2.
The 10% and 20% Profile mixtures increased in dissolved zinc treatment efficiency over
the first week of the study and then remained constant with slight dips and peaks in
treatment efficiency throughout the rest of the season. The all manure mixtures took
several additional weeks to achieve levels of dissolved zinc removal similar to the 10%
and 20% Profile mixtures. Overall, the 20% Profile mixture provided the best dissolved
zinc removal with an average removal rate of 73.4%. The concentrations of dissolved
zinc in the effluent from the reactors and the lowest oxidation, settling pool were well
below the secondary maximum contaminant level (SMCL) of 5.0 mg/L for zinc.
The anaerobic bioreactors also provided some treatment for total zinc, but results
were more variable. The all manure reactors achieved the highest level of total zinc
treatment, reaching 77.4% efficiency during the ninth week of the study. Based on the
average treatment efficiency of 33.64%, the 10% Profile mixture provided the best total
zinc removal. While the lowest settling pool did not provide high levels of dissolved zinc
treatment, it provided treatment efficiencies for total zinc that were in the same range as
the bioreactors.
|
Colorado School of Mines
|
76
5.3.1.3 Aluminum
For aluminum, the field reactors were outperformed by the oxidation, settling pool
treatment method. All of the bioreactors provided treatment of aluminum, but results
were inconsistent for total aluminum, whereas the settling pool system provided high
levels of removal throughout the study. The average total aluminum treatment efficiency
achieved by the lowest settling pool was 71.05%. The lowest settling pool also provided
the highest level of dissolved aluminum removal, with an average treatment efficiency of
23.39%. During the ninth week of the study period, the 20% Profile mixture began to
provide increased levels of dissolved aluminum removal. The treatment efficiency of the
20% Profile mixture for dissolved aluminum continued to increase for the remainder of
the study period.
There are numerous reasons why the anaerobic bioreactors did not provide
consistently high levels of aluminum removal. Aluminum does not form a stable sulfide
in the presence of water (Dvorak et al. 1992). If aluminum sulfides could not form and
precipitate out like other metals, aluminum would be removed only when the pH of the
system was raised enough for aluminum hydrolysis to take place. Therefore, aluminum
removed from the Fair Day Mine drainage can be considered a result of hydrolysis of
AI3+ to the insoluble form of AI(OH)3(s), which then drops out of solution. This criterion
offers some explanation for the sporadic treatment of aluminum within the field reactor
system. The aluminum treatment efficiencies are shown in Figure 5.3. Despite the low
levels of aluminum treatment efficiency, the concentrations of dissolved aluminum in the
effluent from each of the reactors and the lowest settling pool were below the SMCL of
0.2 mg/L. This concentration was surpassed only in the effluent from the all manure
substrate mixtures during the third and fourth weeks of treatment.
|
Colorado School of Mines
|
78
5.3.1.4 Manganese
The Fair Day bioreactors removed manganese more efficiently than expected for
an anaerobic system. The manganese removal efficiencies are shown in Figure 5.4.
The all manure reactors provided the best treatment of dissolved manganese with an
average efficiency of 43.7%. This removal rate was followed by the 20% Profile reactors
treatment efficiency of 41.3% and then the 10% Profile reactors with 31.7% manganese
removal. The aerobic settling pools did not provide removal of dissolved manganese as
effectively as the anaerobic bioreactors. The average treatment efficiency of dissolved
manganese for the lowest settling pool was 23.3%.
Total manganese was not removed from the Fair Day effluent as well as
dissolved manganese. The highest treatment efficiency of total manganese was 23.6%
provided by the all manure reactors. This removal rate was followed by the lowest
settling pool with 23.2% treatment efficiency for total Mn. The oxidation, settling pools
provided similar levels of removal for both total and dissolved manganese, whereas the
anaerobic bioreactors provided higher removal rates of dissolved Mn than total Mn.
Manganese is known to be unstable as a metal sulfide in waters of pH less than
7.1 and under reducing conditions (Willow & Cohen 2003). In order for manganese to
be removed from water, it usually requires a pH greater than 8.0 so that Mn(ll) can be
oxidized into insoluble Mn(IV) and precipitate out of solution (Hallberg & Johnson 2005,
Sheoran & Sheoran 2006). All of the substrate bulking agent mixtures were able to
provide treatment for both total and dissolved Mn some of the time. It is probable that
this treatment occurred when the pH of the system was raised to pH values greater than
7.1 and manganese carbonate species were able to form. It is possible that Mn
precipitated out of solution in the bioreactors as rhodocrosite (MnC03).
|
Colorado School of Mines
|
80
5.3.1.5 Cadmium
Average treatment efficiencies for total and dissolved cadmium were less than
50% for all of the substrate bulking agent mixtures. The most effective treatment of
dissolved cadmium, 65.13%, was provided by the all manure substrates during the
fourteenth week of the sampling period. The all manure mixture also provided the best
average dissolved treatment over the course of the study with an average of 16.84%.
This was followed by the treatment efficiencies of the lowest settling pool and the 10%
Profile mixture, averaging 12.08% and 11.69%, respectively. While the treatment
efficiency of dissolved cadmium was low in the bioreactors, the dissolved treatment
levels surpassed those for total cadmium treatment. The oxidation, settling pools
provided the best treatment of total cadmium with an average treatment efficiency of
21.64%. Treatment efficiencies for total and dissolved cadmium can be seen in Figure
5.5.
Peaks and dips in the levels of cadmium treatment efficiency occurred on the
same days for total and dissolved cadmium. This trend was observed for all of the
substrate mixtures as well as the lowest settling pool, indicating that the changes in
cadmium treatment efficiencies can be attributed to the rate that cadmium was loaded to
the mine effluent. It is important to note that while the treatment efficiencies of cadmium
were less than desired in the Fair Day system, the dissolved concentrations of cadmium
in the reactor and settling pool effluents were below the 0.005mg/L maximum
contaminant level (MCL). As a result of the small concentrations of cadmium in the
water, a conclusive decision could not be made in regards to which substrate bulking
agent mixture provides the best cadmium removal.
|
Colorado School of Mines
|
83
5.4.1 Specific Gravity Tests
For this set of tests, only the unused substrate and the column reactor substrates
were examined. Three trials were run on both the unused and the column materials.
Results from the three trials were averaged together to determine a specific gravity value
for each mixture that could be used in porosity calculations. The average specific gravity
values for the unused substrate materials can be seen in Table 5.2:
Table 5.2: Specific Gravity of Unused Substrate Mixtures
Substrate Mixture: Average: Std. Deviation:
All Manure 1.77 0.28
10% Profile 1.83 0.26
20% Profile 1.91 0.10
All Profile 2.36 0.26
The average specific gravity values for the used column substrate materials are
shown in Table 5.3:
Table 5.3: Specific Gravity of Column Substrate Mixtures
Substrate Mixture: Average: Std. Deviation:
All Manure 1.94 0.11
10% Profile 2.03 0.13
20% Profile 2.11 0.04
Based on a comparison of the specific gravity data, it did not appear that the
specific gravity of each of the mixtures changed significantly over time. Overall, the all
manure substrate samples had the least amount of change in specific gravity over the 28
day timeframe.
5.4.2 Bulk Density Tests
The bulk density of each of the unused substrate and column substrate mixtures
was tested. The results from this analysis were used in conjunction with the specific
|
Colorado School of Mines
|
84
gravity data to calculate porosity estimates for each of the substrate mixtures. Three
trials were performed on each substrate mixture. The averages from the analysis on the
unused substrate are shown in Table 5.4:
Table 5.4: Bulk Density of Unused Substrate Mixtures
Substrate Mixture: Average (g/cm3): Std. Deviation:
All Manure 1.17 0.02
10% Profile 1.20 0.00
20% Profile 1.20 0.02
All Profile 1.03 0.02
The average bulk densities from the analysis of the column substrate are shown
in Table 5.5:
Table 5.5: Bulk Density of Column Substrate Mixtures
Substrate Mixture: Average (g/cm3): Std. Deviation:
All Manure 1.26 0.01
10% Profile 1.27 0.00
20% Profile 1.24 0.02
There did not appear to be any major changes in the bulk density of the different
substrate mixtures over time. Since there was no significant change in the specific
gravity or bulk density of the substrate mixtures over time, and these two factors are
used to determine the porosity of the material, it can be assumed that there was not a
major reduction in porosity over time.
5.4.3 Porosity Calculations
The results from the specific gravity and bulk density tests were used to calculate
average porosity values for each mixture of unused and column substrates. Porosities
were calculated for each trial and mixture based on the following equation:
Average Bulk Density [ y 3 |
(16) 1— ------------------------------Porosity
f 9 /^3 J
Average Specific Gravity
|
Colorado School of Mines
|
85
Individual porosity values from each trial were averaged and the results are shown
in the Tables 5.6 and 5.7:
Table 5.6: Calculated Porosities for Unused Substrate Mixtures
Substrate Mixture: Average:
All Manure 0.339
10% Profile 0.334
20% Profile 0.372
All Profile 0.564
Table 5.7: Calculated Porosities for Column Substrate Mixtures
Substrate Mixture: Average:
All Manure 0.350
10% Profile 0.371
20% Profile 0.411
By comparing the porosities of each substrate mixture of unused and used
column materials, a slight change in porosity over time was observed. The resulting
porosities show that there is actually an increase in the effective porosity of each of the
mixture types of substrate over time. The 20% Profile mixture had the highest increase
in porosity over the 28 day period. This was followed the 10% Profile mixture. The all
manure porosity increased the least over the measured period of time.
5.4.4 Organic Content Analysis
Unused substrate mixtures as well as the column substrates were run in
triplicates for this part of the analysis. The results from each of the trials were averaged
together to obtain an average value for each unused and column mixture. Table 5.8
presents the average results from the unused materials:
|
Colorado School of Mines
|
87
content similar to that of the steer manure alone. These results can be compared to the
average values for the used column materials listed in Table 5.9:
Table 5.9: Organic Content of Column Substrate Mixtures
Substrate Mixture: Average Percent by Weight (%): Std. Deviation:
All Manure 25.55 5.07
10% Profile 23.25 0.91
20% Profile 17.56 1.41
The percent organic matter in the column substrates is less than the organic
content in the unused substrate mixtures, showing the depletion of organic materials
over time. This trend is expected because the organic matter in the substrates provide
nutrients for the microbial consortium in the system. Therefore, these results show that
there was a significant level of microbial activity taking place within the columns. The
percent of organic matter in the all manure samples decreased by approximately 5.64%
from the unused to the column substrates, resulting in the greatest depletion of organic
matter seen between the mixtures.
5.4.5 Size Particle Distribution
Several factors may have caused changes in the particle size distribution of each
substrate mixture over time. One reason could be the microbial activity taking place
within the system. Microbial activity can lead to the decomposition of organic matter,
which will result in smaller size particles of organic materials breaking away from the
bulk substrate. The small organic particles produced by microbial activity as well as
amorphous metal-sulfide slimes could accumulate within the pore spaces of the
substrate, clogging the system and leading to a reduction of porosity and resulting
z hydraulic conductivity.
|
Colorado School of Mines
|
88
5.4.5.1 Unused Substrate
The all manure, 10% Profile, and 20% Profile mixtures were tested as well as an
all Profile sample. The unused substrate mixtures were dried sieved three times. The
average fraction sizes for the three trials are shown in Table 5.10 by percent of dry
weight.
Table 5.10: Size Fraction Distribution of Unused Substrate Mixtures
% Dry Weight
Sieve Size Retains Particles All Manure 10 % Profile 20% Profile All Profile
10 2 mm 0.42 0.47 0.45 0.56
20 0.85 mm 0.29 0.30 0.31 0.44
100 0.15 mm 0.27 0.21 0.22 0.00
200 0.075 mm 0.02 0.01 0.02 0.00
Pan < 0.075 mm 0.00 0.00 0.00 0.00
5.4.5.2 Column Substrate
Triplicate trials were performed on each of the substrate samples from the
column reactors. The results from duplicate columns were combined to calculate an
average size fraction distribution for each substrate bulking agent mixture. Average
particle size fractions are shown in Table 5.11 below by percent of dry weight.
Table 5.11: Size Fraction Distribution of Column Substrate Mixtures
% Dry Weight
Sieve Size Retains Particles All Manure 10% Profile 20% Profile
10 2 mm 0.37 0.49 0.56
20 0.85 mm 0.22 0.24 0.27
100 0.15 mm 0.30 0.20 0.14
200 0.075 mm 0.09 0.06 0.03
Pan < 0.075 mm 0.02 0.03 0.02
In comparison with the size fraction distribution of the unused substrate mixtures,
the column reactor materials have significantly more fine particles. The percent by
weight of each size fraction shifted making the particles in size range 0.075 mm-0.15
|
Colorado School of Mines
|
mm as well as the particles finer than 0.075 mm more dominant. The all manure
mixtures had the greatest increase in the amount of fine particles. The 10% Profile
mixtures had the second largest increase in the quantity of fine particle sizes followed by
the changes seen in the 20% Profile mixtures. This trend implies that the addition of
Profile soil amendment to the composted steer manure helped to reduce the amount of
the organic substrate that was broken down into finer particle sizes in this timeframe.
S.4.5.3 Field Reactor Substrate
Substrate from each of the field reactors was analyzed through two separate
trials. The results from these two trials were averaged together and the results are
presented in Table 5.12 in terms of percent dry weight for each of the size fractions.
Table 5.12: Size Fraction Distribution of Field Substrate Mixtures
% Dry Weight
Sieve Size Retains Particles All Manure 10 % Profile 20% Profile
10 2 mm 0.21 0.35 0.37
20 0.85 mm 0.26 0.29 0.31
100 0.15 mm 0.38 0.25 0.22
200 0.075 mm 0.12 0.09 0.08
Pan < 0.075 mm 0.05 0.04 0.04
The analysis of the field reactor mixtures further shows that fine particles within
the substrate become more abundant over time. One reason for this is the microbial
activity within the system. Another possible reason for the increase in fine particles over
time and use is that initially the fine particles are attached and strongly bound to larger
particles within the substrate. When water is run through reactors containing this
substrate, the fine particles may be released from the larger particles. The fines then
agglomerate in the voids spaces within the substrate and cause an overall reduction in
porosity and hydraulic conductivity. This same phenomenon was thought to occur in
|
Colorado School of Mines
|
93
Conductivity measurements were taken from each of the bioreactors for two
subsequent weeks during normal sampling events. This data can be seen in Table 5.13.
Table 5.13: Conductivity of the Effluent from Bioreactors After Tracer Test
Date of Sampling Event 10/8/06 10/11/06 10/20/06
Reactor # Conductivity (mS)
1 21.400 12.630 6.950
2 2.580 1.211 0.646
3 0.374 0.375 0.328
4 1.112 0.956 0.676
5 2.570 0.733 0.394
6 1.602 0.872 0.522
7 0.598 0.716 0.646
The conductivity measurements shown in Table 5.13 for the October 8, 2006
sampling event represent the last measurement taken during the 50-hour tracer test.
The conductivity the effluent from each bioreactor continued to decrease over the course
of the following weeks. The conductivities of all of the effluents were back to
background levels by the October 20, 2006 sampling event, with the exception of the
effluent from Reactor #1. This reactor also had the highest measured conductivity
values over the course of the tracer test.
5.5.1 Hydraulic Conductivity Calculations
Hydraulic conductivity was calculated for each of the field reactors and then the
values for the duplicate reactors were averaged together. A hydraulic conductivity was
also calculated for Reactor #4, which contained a substrate mixture of 15% alfalfa hay.
This value was not averaged with another reactor since there was not a duplicate;
however this value is presented along with the other averaged values in Table 5.14:
|
Colorado School of Mines
|
97
6.0 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE RESEARCH
This experiment resulted in numerous findings which can be applied to the
design and installation of future anaerobic passive treatment systems. There are a
number of conclusions that can be made based on these findings. As a result, several
design components could be altered and optimized in order to improve overall treatment
efficiency of mining influenced waters.
6.1 Design Configuration
The addition of Profile soil amendment to organic substrates can improve the
permeability of the system as well as reduce the amount of substrate materials that are
broken down into smaller size particles. The addition of Profile in percentages by
volume of 10% or 20% did not appear to decrease the metal removal efficiency of the
bioreactors. One major improvement that could be made to the system is to use a
variety of organic substrates in conjunction with the Profile rather than solely composted
steer manure. Numerous studies have found evidence that suggests sulfate reduction
efficiency can be augmented by using mixtures of a number of organic materials (Cocos
et al. 2002, Zagury et al. 2006).
Several aspects of this experiment could be improved upon and redesigned for
future applications. First of all, the base substrate for bioreactors should be fully
analyzed and evaluated in the laboratory prior to use in lab or field scale projects. The
most important test to perform prior to using a substrate material is the particle size
distribution analysis. If the initial particle size distribution shows a majority of fine
particles, the substrate should not be used in a reactor system. Overtime organic
substrate materials will break down into smaller particles, increasing the amount of fines
in the system. Therefore, if the initial particle size distribution contains too many fines
|
Colorado School of Mines
|
98
from the start, it can be assumed the problem will only intensify over time and a different
substrate should be used.
Another aspect of the experiment that should be redesigned for future use is the
configuration of the column experiment. The laboratory columns used for this study
were adapted from a previous experiment; therefore the size of the columns was
predetermined. In order to make the column experiments easier to work with and collect
data from, test columns should be developed on a smaller scale. Columns that can sit
on a bench top would provide a more ideal experimental configuration, allowing for leaks
to be dealt with quickly and efficiently. The initial reasoning for using large scale
columns for the laboratory portion of this experiment was to simulate the processes
expected to occur in the field reactors on a similar scale. While using columns similar in
size to the field reactors might provide more representative data, the large scale of the
columns proved to be a challenge.
Numerous forms of SRB have been demonstrated throughout the years to
successfully transform hexavalent uranium to its less mobile tetravalent form. The many
forms of (J(VI)-reducing SRB vary greatly metabolically and produce different treatment
times and metal removal efficiencies. As a result, the decision of which species to use
for remediation processes has become more important. In many cases, in order to
achieve high levels of SRB sulfate reduction inoculums of particular strains of bacteria
can be added to the substrate prior to the addition of the mining influenced waters. This
process allows the SRB to acclimate in the bioreactors prior to exposure with the metal
laden mine waters. Using specific strains of bacteria can allow the treatment of specific
metals to be optimized, resulting in better overall treatment. Therefore, it is beneficial to
determine which SRB are appropriate to use for specific applications rather than relying
on SRB naturally contained within the substrates.
|
Colorado School of Mines
|
99
6.2 Metal Removal Efficiency
While the treatment efficiencies of the bioreactors were variable, overall the
anaerobic reactors provided better treatment for the Fair Day Mine effluent than the
aerobic settling pools. Since the oxidation, settling pools will function as the long-term
treatment system at Fair Day, perhaps the top settling pool could be redesigned as an
anaerobic wetland containing organic substrate. There is an underground supply line of
adit drainage, currently supporting the bioreactor system, which could be easily
reconfigured to direct the flow to the top settling pool. This pool would then be able to
function in an upflow manner, allowing anaerobic conditions to exist in the pool. After
the water moved through the substrate it would flow out of the wetland, across a rock
cascade, and into the next pool, which would remain aerobic. Allowing the water to go
through both anaerobic and aerobic treatment systems would help to remove some
metals that would not be removed by the bioreactors or settling pools alone.
As a final polishing step before the water is released back into the environment, a
limestone cascade could be installed after the lowest settling pool. The cascade could
consist of a series of steps coated in limestone aggregate that the water exiting the
lowest pool would flow across. Contact with limestone aggregate would help to raise the
alkalinity of the water and help the metals that could not be removed during anaerobic
treatment, such as aluminum, to precipitate out of solution. The aeration of the water
provided by falling across the steps would also help to decrease the amount of ammonia
and the biological oxygen demand being released into the environment.
At many sites treating MIW, the use of limestone is not an option due to the high
concentrations of iron contained in the water. When iron-contaminated water contacts
limestone the iron precipitates out of solution. While this is a suitable removal method,
the iron precipitates tend to armor the limestone aggregate, rending the limestone
|
Colorado School of Mines
|
100
treatment useless. However, the concentrations of iron in the Fair Day effluent are low
enough that the limestone cascade would be able to function for a long period of time.
The use of limestone neutralization in conjunction with anaerobic treatment may help to
precipitate metals contained in the Fair Day effluent that were not removed during
anaerobic treatment, thus increasing the overall treatment efficiency of the system.
An addition form of treatment that could be applied to the end of the Fair Day
treatment process is filtration. In many cases when treating mining influenced waters,
the effluent from anaerobic bioreactors can still contain a significant concentration of
metal precipitates which have not yet fallen out of solution. Passing the effluent from
anaerobic bioreactors through a filtration system will help to further remove the metal
precipitates from the water that formed during the treatment process. The result of this
finishing filtration will be cleaner effluent ready to be released back into the environment.
6.3 Hydraulic Conductivity
Profile soil amendment provided increased hydraulic conductivities to the
substrate when mixed at ratios of 10% and 20% by volume. If Profile was to be used as
a bulking agent in future studies, it would be beneficial to use larger amounts based on
volume. For example, mixing in 20%, 40% and 60% Profile by volume to composted
steer manure. Using mixture ratios in a wider range might better show differences in
treatment efficiency and long-term performance for each of the substrate bulking agent
mixtures. The mixtures used in the present study were close in ratios of Profile added to
the steer manure and in terms of metal treatment efficiency performed similarly.
These factors made it difficult to determine the benefits or limitations associated
with each substrate mixture and therefore the best substrate mixture could not be
identified. While the mixtures using 10% and 20% Profile by volume did not seem to
negatively impact the treatment capabilities of the system, mixtures using higher
|
Colorado School of Mines
|
101
percentages by volume may inhibit the treatment of metals due to the reduction of
organic matter initially provided to the system. Further work needs to be done to
determine the maximum amount of Profile that can be used as a bulking agent without
limiting treatment efficiencies.
Several preventative measures can be taken in order avoid complete clogging of
pore space due to metal precipitates and biofilms. First, increasing the velocity of the
influent water intermittently will have a flushing effect on the system, pushing the metal
precipitates and biofilms out of the pore spaces (Jong & Parry 2003). Ideally after the
flushing phase, water can flow through the substrate uninterrupted which will also help to
prevent the development of preferential flow paths. Flushing out bioreactors in this
manner on a regular basis may help to increase the lifespan of the substrate and long
term treatment efficiency of the system.
While anaerobic bioreactors provide effective treatment of MIW, they are limited
by pore space clogging and short circuiting. In order to reduce the amount of clogging,
porous bulking agents need to be mixed with organic substrates to provide more
structural support as well as to yield larger pore volumes for water to move through. The
use of Profile soil amendment as a substrate bulking agent may reduce the amount of
organic content, but this is outweighed by the beneficial properties provided to retain
hydraulic conductivities over time and reduce the formation of fine particles. Bulking
agents can help to extend the operational lifetime of anaerobic passive treatment
systems, which may allow for more full scale MIW treatment systems to be installed.
|
Colorado School of Mines
|
ABSTRACT
Unstablefailureinundergroundcoalminingisthesuddenandviolentejectionofcoalfrom
mine walls and pillars into the mine opening. This thesis demonstrates the use of the discrete
element method to simulate stable and unstable modes of compressive failure of a western
U.S. coal. Two discrete element models are evaluated for their ability to simulate unstable
and stable compressive failure using the discrete element program Particle Flow Code in
TwoDimensions(PFC2D):thebondedparticlemodelandthedisplacementsofteningmodel.
Compressive strength tests show that the displacement softening model is better suited for
unstable failure studies based on consistent behavior in stable and unstable modes of failure
and a post-peak softening characteristic that is independent of the loading rate.
Asetofmodelbehaviors, calledindicators, areanalyzedontheirabilitytodistinguishthe
stability of failure in a series of unconfined compression tests and then a series slender pillar
compressivestrengthtests. Generally, theindicatorsshowconsistentvaluesforstablefailures
and increasing magnitude with increasing levels of instability. A grid based measurement
technique is used to observe indicator behavior and model damage spatially.
The work by the damping mechanism, kinetic energy, and the mean unbalanced force are
used to analyze pillar edge failure in a model with excavation induced loading conditions.
The indicators reveal unstable failure events, and a comparison between stable and unstable
mining steps show that the indicators can be used to detect local instabilities on, such as
pillar rib failure. Grid based measurements show that the unstable failure is initiated due to
a single mining step and that failure occurred along a diagonal failure plane originating from
the mine face similar to that seen in practice. Unstable failures show highly localized planes
of failure while stable pillar failure is more dispersed. Future application of the techniques
developed in this thesis include more in depth study of factors influencing unstable failures
in coal mines including the mine/coal seam contact condition and depth.
iii
|
Colorado School of Mines
|
Figure 3.16 DSM coupled simulation stress-strain curves . . . . . . . . . . . . . . . . 47
Figure 3.17 Loading system and specimen post-peak moduli in EPC tests . . . . . . 49
Figure 3.18 BPM variable loading rate unconfined compression stress-strain curves . 53
Figure 3.19 DSM variable loading rate unconfined compression stress-strain curves . 53
Figure 4.1 Kinetic energy indicator results for EPC test with 5 GPa platens . . . . 62
Figure 4.2 Accumulated damping work during the failure interval in EPC tests . . . 63
Figure 4.3 Maximum instantaneous kinetic energy in EPC tests . . . . . . . . . . . 64
Figure 4.4 Maximum instantaneous mean unbalanced force in EPC tests . . . . . . 64
Figure 4.5 Maximum instantaneous maximum unbalanced force in EPC tests . . . . 65
Figure 4.6 Cumulative kinetic energy during failure in EPC tests . . . . . . . . . . . 66
Figure 4.7 Cumulative mean unbalanced force in EPC tests . . . . . . . . . . . . . . 66
Figure 4.8 Cumulative maximum unbalanced force in EPC tests . . . . . . . . . . . 67
Figure 4.9 Contact softening in EPC tests . . . . . . . . . . . . . . . . . . . . . . . 68
Figure 4.10 Number of broken contacts in EPC tests . . . . . . . . . . . . . . . . . . 68
Figure 4.11 Slender pillar test geometry and boundary conditions . . . . . . . . . . . 71
Figure 4.12 Illustration of typical pillar simulation behaviors . . . . . . . . . . . . . . 72
Figure 4.13 Grid based measurement algorithm flow chart . . . . . . . . . . . . . . . 74
Figure 4.14 Stress-strain curves for width to height ratio one pillar tests . . . . . . . 75
Figure 4.15 Stress-strain curves for width to height ratio two pillar tests . . . . . . . 76
Figure 4.16 Stress-strain curves for width to height ratio three pillar tests . . . . . . 76
Figure 4.17 Loading system displacements for width to height ratio one pillar tests . 77
Figure 4.18 Loading system displacements for width to height ratio two pillar tests . 78
Figure 4.19 Loading system displacements for width to height ratio three pillar tests . 78
ix
|
Colorado School of Mines
|
Figure B.4 Stability test stress-strain curves for two DSM specimens with different
post-peak behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Figure D.1 Damping work, EPC Test with 1 GPa loading system . . . . . . . . . . 191
Figure D.2 Kinetic energy, EPC Test with 1 GPa loading system . . . . . . . . . . 191
Figure D.3 Mean unbalanced force, EPC Test with 1 GPa loading system . . . . . 192
Figure D.4 Maximum unbalanced force, EPC Test with 1 GPa loading system . . . 192
Figure D.5 Contact softening, EPC Test with 1 GPa loading system . . . . . . . . 193
Figure D.6 Broken contacts, EPC Test with 1 GPa loading system . . . . . . . . . 193
Figure D.7 Damping work, EPC Test with 1.5 GPa loading system . . . . . . . . . 194
Figure D.8 Kinetic energy, EPC Test with 1.5 GPa loading system . . . . . . . . . 194
Figure D.9 Mean unbalanced force, EPC Test with 1.5 GPa loading system . . . . 195
Figure D.10 Maximum unbalanced force, EPC Test with 1.5 GPa loading system . . 195
Figure D.11 Contact softening, EPC Test with 1.5 GPa loading system . . . . . . . 196
Figure D.12 Broken contacts, EPC Test with 1.5 GPa loading system . . . . . . . . 196
Figure D.13 Damping work, EPC Test with 2.5 GPa loading system . . . . . . . . . 197
Figure D.14 Kinetic energy, EPC Test with 2.5 GPa loading system . . . . . . . . . 197
Figure D.15 Mean unbalanced force, EPC Test with 2.5 GPa loading system . . . . 198
Figure D.16 Maximum unbalanced force, EPC Test with 2.5 GPa loading system . . 198
Figure D.17 Contact softening, EPC Test with 2.5 GPa loading system . . . . . . . 199
Figure D.18 Broken contacts, EPC Test with 2.5 GPa loading system . . . . . . . . 199
Figure D.19 Damping work, EPC Test with 5 GPa loading system . . . . . . . . . . 200
Figure D.20 Kinetic energy, EPC Test with 5 GPa loading system . . . . . . . . . . 200
Figure D.21 Mean unbalanced force, EPC Test with 5 GPa loading system . . . . . 201
xii
|
Colorado School of Mines
|
Figure D.22 Maximum unbalanced force, EPC Test with 5 GPa loading system . . . 201
Figure D.23 Contact softening, EPC Test with 5 GPa loading system . . . . . . . . 202
Figure D.24 Broken contacts, EPC Test with 5 GPa loading system . . . . . . . . . 202
Figure D.25 Damping work, EPC Test with 10 GPa loading system . . . . . . . . . 203
Figure D.26 Kinetic energy, EPC Test with 10 GPa loading system . . . . . . . . . 203
Figure D.27 Mean unbalanced force, EPC Test with 10 GPa loading system . . . . . 204
Figure D.28 Maximum unbalanced force, EPC Test with 10 GPa loading system . . 204
Figure D.29 Contact softening, EPC Test with 10 GPa loading system . . . . . . . . 205
Figure D.30 Broken contacts, EPC Test with 10 GPa loading system . . . . . . . . 205
Figure D.31 Damping work, EPC Test with 20 GPa loading system . . . . . . . . . 206
Figure D.32 Kinetic energy, EPC Test with 20 GPa loading system . . . . . . . . . 206
Figure D.33 Mean unbalanced force, EPC Test with 20 GPa loading system . . . . . 207
Figure D.34 Maximum unbalanced force, EPC Test with 20 GPa loading system . . 207
Figure D.35 Contact softening, EPC Test with 20 GPa loading system . . . . . . . . 208
Figure D.36 Broken contacts, EPC Test with 20 GPa loading system . . . . . . . . 208
Figure D.37 Damping work, EPC Test with 35 GPa loading system . . . . . . . . . 209
Figure D.38 Kinetic energy, EPC Test with 35 GPa loading system . . . . . . . . . 209
Figure D.39 Mean unbalanced force, EPC Test with 35 GPa loading system . . . . . 210
Figure D.40 Maximum unbalanced force, EPC Test with 35 GPa loading system . . 210
Figure D.41 Contact softening, EPC Test with 35 GPa loading system . . . . . . . . 211
Figure D.42 Broken contacts, EPC Test with 35 GPa loading system . . . . . . . . 211
Figure D.43 Damping work, EPC Test with 50 GPa loading system . . . . . . . . . 212
Figure D.44 Kinetic energy, EPC Test with 50 GPa loading system . . . . . . . . . 212
xiii
|
Colorado School of Mines
|
Figure D.45 Mean unbalanced force, EPC Test with 50 GPa loading system . . . . . 213
Figure D.46 Maximum unbalanced force, EPC Test with 50 GPa loading system . . 213
Figure D.47 Contact softening, EPC Test with 50 GPa loading system . . . . . . . . 214
Figure D.48 Broken contacts, EPC Test with 50 GPa loading system . . . . . . . . 214
Figure E.1 Damping work, width to height one pillar 5 GPa loading system . . . . 216
Figure E.2 Kinetic energy, width to height one pillar 5 GPa loading system . . . . 216
Figure E.3 Mean unbalanced force, width to height one pillar 5 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Figure E.4 Max unbalanced force, width to height one pillar 5 GPa loading system 217
Figure E.5 Contact softening, width to height one pillar 5 GPa loading system . . 218
Figure E.6 Broken contacts, width to height one pillar 5 GPa loading system . . . 218
Figure E.7 Damping work, width to height one pillar 20 GPa loading system . . . 219
Figure E.8 Kinetic energy, width to height one pillar 20 GPa loading system . . . 219
Figure E.9 Mean unbalanced force, width to height one pillar 20 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Figure E.10 Max unbalanced force, width to height one pillar 20 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Figure E.11 Contact softening, width to height one pillar 20 GPa loading system . . 221
Figure E.12 Broken contacts, width to height one pillar 20 GPa loading system . . . 221
Figure E.13 Damping work, width to height one pillar 35 GPa loading system . . . 222
Figure E.14 Kinetic energy, width to height one pillar 35 GPa loading system . . . 222
Figure E.15 Mean unbalanced force, width to height one pillar 35 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Figure E.16 Max unbalanced force, width to height one pillar 35 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
xiv
|
Colorado School of Mines
|
Figure E.17 Contact softening, width to height one pillar 35 GPa loading system . . 224
Figure E.18 Broken contacts, width to height one pillar 35 GPa loading system . . . 224
Figure E.19 Damping work, width to height two pillar 5 GPa loading system . . . . 225
Figure E.20 Kinetic energy, width to height two pillar 5 GPa loading system . . . . 225
Figure E.21 Mean unbalanced force, width to height two pillar 5 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Figure E.22 Max unbalanced force, width to height two pillar 5 GPa loading system 226
Figure E.23 Contact softening, width to height two pillar 5 GPa loading system . . 227
Figure E.24 Broken contacts, width to height two pillar 5 GPa loading system . . . 227
Figure E.25 Damping work, width to height two pillar 20 GPa loading system . . . 228
Figure E.26 Kinetic energy, width to height two pillar 20 GPa loading system . . . 228
Figure E.27 Mean unbalanced force, width to height two pillar 20 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Figure E.28 Max unbalanced force, width to height two pillar 20 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Figure E.29 Contact softening, width to height two pillar 20 GPa loading system . . 230
Figure E.30 Broken contacts, width to height two pillar 20 GPa loading system . . 230
Figure E.31 Damping work, width to height two pillar 35 GPa loading system . . . 231
Figure E.32 Kinetic energy, width to height two pillar 35 GPa loading system . . . 231
Figure E.33 Mean unbalanced force, width to height two pillar 35 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Figure E.34 Max unbalanced force, width to height two pillar 35 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Figure E.35 Contact softening, width to height two pillar 35 GPa loading system . . 233
Figure E.36 Broken contacts, width to height two pillar 35 GPa loading system . . 233
xv
|
Colorado School of Mines
|
Figure E.37 Damping work, width to height three pillar 5 GPa loading system . . . 234
Figure E.38 Kinetic energy, width to height three pillar 5 GPa loading system . . . 234
Figure E.39 Mean unbalanced force, width to height three pillar 5 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Figure E.40 Max unbalanced force, width to height three pillar 5 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Figure E.41 Contact softening, width to height three pillar 5 GPa loading system . 236
Figure E.42 Broken contacts, width to height three pillar 5 GPa loading system . . 236
Figure E.43 Damping work, width to height three pillar 20 GPa loading system . . 237
Figure E.44 Kinetic energy, width to height three pillar 20 GPa loading system . . . 237
Figure E.45 Mean unbalanced force, width to height three pillar 20 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Figure E.46 Max unbalanced force, width to height three pillar 20 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Figure E.47 Contact softening, width to height three pillar 20 GPa loading system . 239
Figure E.48 Broken contacts, width to height three pillar 20 GPa loading system . . 239
Figure E.49 Damping work, width to height three pillar 35 GPa loading system . . 240
Figure E.50 Kinetic energy, width to height three pillar 35 GPa loading system . . . 240
Figure E.51 Mean unbalanced force, width to height three pillar 35 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Figure E.52 Max unbalanced force, width to height three pillar 35 GPa loading
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Figure E.53 Contact softening, width to height three pillar 35 GPa loading system . 242
Figure E.54 Broken contacts, width to height three pillar 35 GPa loading system . . 242
Figure G.1 Contact softening in the deep simulation . . . . . . . . . . . . . . . . . 279
Figure G.2 Maximum unbalanced force in the deep simulation . . . . . . . . . . . 280
xvi
|
Colorado School of Mines
|
ACKNOWLEDGMENTS
I would like to first acknowledge my thesis advisor, Ugur Ozbay. Guided by your kind
wisdom I have learned much about myself and life in general. Each time I arrived at new
stages of realization throughout my studies, I also gained a greater appreciation for the
efforts you afforded me. To put it mildly, I owe you one, and thank you.
My peers Ray Gu and Ryan Garvey spent many hours with me discussing topics of
research. These conversations, which inevitably made their way towards more esoteric sub-
jects, helped keep a flame of inspriation flickering in my mind when my way forward was
murky. Thanks gents!
Graham Mustoe, who helped to provide much needed support with questions regarding
nuancesoftheDEM.You’vealwayssentmeonmywaywithnewideasandpossiblepathways
to explore. Cheers!
To my friends in Golden and Albuquerque. It’s been a blessing having such friends
that have always made me feel loved and at home. And a special shout out for my canine
compainion, Roofis. That little guy has been through it all with me. Love ya buddy!
To my family back in the midwest. Barbara and Kevin, I owe you debt of gratitude for
your advice to turn my sights on the Ph.D. at CSM. Without you I would have never taken
this route. Also to the Rowlands, Dale, Charlie, and Grandma. I can’t say enough about
having a loving family to come back to.
To the esteemed faculty at the Colorado School of Mines who have gone great lengths to
make the last four years a life changing experience. In particular Juan Lucena, John Berger,
and Christian Frenzel. I strive everyday to emulate your confidence and professionalism.
xxv
|
Colorado School of Mines
|
CHAPTER 1
INTRODUCTION
An underground mine is constructed as a system of hallways, or entries, that are kept
open by pillars and abutments comprised of the in-situ material that is left behind. This
primary support system is accompanied by a secondary support system composed of wood,
steel, and hydraulic props that provide the additional support necessary to ensure stability
of the working areas. A greater burden is put onto the support system as activity in the
mine advances and more material is removed. Ideally, the support system will gradually fail
under the increasing load and then these areas will be sealed permanently. However, the rock
doesn’t always fail in a controlled manner. In some cases, large amounts of rock are suddenly
ejected with great velocity from mine walls resulting in injury or death of mine workers and
suspension of operations. This sudden, violent failure of rock is called rockburst, or is more
generally referred to as unstable failure.
Unstable failure is common in underground coal mining operations. The magnitude of
unstable failures in underground coal mining can range from audible readjustment of mine
stress to ejection of material from mine walls in a localized area to collapse of entire panels
of coal pillars. In coal mining terminology, localized unstable failures are typically referred
to not as rockburst but bumping or bouncing. While much effort has been dedicated to
understandingthephysicalmechanismofunstablefailureofrockingeneral, andtheconcepts
have been applied to the coal mining situation, mining operations still are unable to predict
the time and intensity of bumps.
Recent advances in numerical modeling have allowed for research into the physical mech-
anism of unstable failure that has potential to aid existing theoretical and experimental
methods. An increase in computer processing power has allowed for models with increased
complexity and size to be practical. More specifically, the discrete element method (DEM)
1
|
Colorado School of Mines
|
shows promise in capturing the micro-mechanical behavior of rock during failure that may
be a crucial part in understanding the unstable failure of underground coal. An attractive
feature of the DEM is the property of having emergent rock like behavior despite no explicit
assignment of specific rock properties. For example, the ability of DEM to allow for crack
propagation, a realistic Poisson effect, and increased strength with confinement. This thesis
describes the development of improved numerical tools in DEM for analysis of the problem
of unstable failure in an underground coal mining situation.
1.1 Problem Statement
In underground mining conditions, it is currently impossible to reliably predict when,
where and with what intensity an unstable failure will occur. By studying the failure mech-
anism and factors that affect unstable failures, improvements can be made towards assessing
the probability of intense, unstable failure. However, studying the mechanism of unstable
failures in underground coal mines is a challenging task for two reasons. The unpredictable
nature of unstable failures makes observation of the events problematic. Aside from a few
case studies, anecdotes from the surviving mine workers are the only data available to de-
scribe the failure. And, due to the nature of the failure, evidence of the failure mechanism
is lost because it is unsafe or impossible to access the failed area of the mine.
A recent occurrence of a series of unstable failures at the Crandall Canyon Mine, Utah
in 2007 illustrates the devastation potentially associated with unstable failures in coal min-
ing. The Mine Safety and Health Administration (MSHA) coordinated an investigation of
the incident that included the participation of the mine operator, MSHA investigators and
consultants Stricklin [80]. The initial collapse failed highly stressed pillars throughout a
distance of approximately one half mine and registered as a magnitude 3.9 seismic event.
This failure entombed six miners, and three were subsequently killed during a second failure
while performing a rescue excavation operation. Figure 1.1 is a picture of the entry in which
rescue miners were working to rescue the trapped miners after the second failure. The coal
ejected from the entry walls in this area of the mine rendered this entry impassable and in
2
|
Colorado School of Mines
|
1.2 Research Objectives and Methodology
The objectives of this research are to implement a DEM model appropriate for studying
unstable compressive failure that utilizes the state of the art techniques in DEM modeling
and knowledge of the mechanism of unstable failure. This goal is achieved by accomplishing
a series of tasks listed below.
Calibrate candidate contact models in PFC2D to approximate an in situ western U.S.
•
coal.
ImplementadvancedtechniquesinDEMmodelingtofacilitateappropriatemodelchar-
•
acteristics, e.g. apply appropriate confinement via boundary forces, utilize mechanical
coupling algorithm to apply realistic and computationally efficient loading via con-
tinuum model, and construction of large DEM assemblies using a periodic material
generation procedure.
Test candidate models in a variety of compressive tests to determine poignant model
•
behaviors and assess their applicability to studying compressive unstable failure.
ValidatetheabilityofthecalibratedDEMmodeltosimulateunstableandstablefailure
•
modes in compression.
Identify a series of potential numerical indicators for distinguishing between stable and
•
unstable failure.
Analyze performance of indicators in cases of known failure stability.
•
Develop a method for calculating and displaying indicator values so that spatial and
•
magnitude attributes can be observed.
Test identifier performance in a mine model with realistic loading applied by in situ
•
stress and excavation.
4
|
Colorado School of Mines
|
1.3 Thesis Organization
Chapter 2 is a review of previous research found in the literature. Topics included are the
rock mechanics of unstable failure, coal minig methods, and unstable failure in underground
coal mining. The subject of numerical modeling in coal mining is discussed thoroughly by
introducing numerical methods used in analyzing underground mining, some applications
to study unstable failure, and special attention is given to the discrete element method to
establish context for the DEM models used in this research.
Chapter 3 describes the calibration and comparison of two discrete element models, a
widelyusedbondedparticlemodelandamodelthatusesthesocalleddisplacementsoftening
contactmodel. Thespecimensarefailedincompressioninfourseparatetesttypestoevaluate
the characteristic material behavior under rigid loading, the effect of confinement on stress
to determine the Mohr-Coulomb friction angle, a test designed to reveal model behavior in
stable and unstable failure modes, and a test to determine the effect of loading velocity on
material behavior. Based on the results of these tests, the more appropriate DEM is chosen
for further use in the thesis.
Chapter 4 introduces several failure stability indicators and applies them to two sets of
compressive strength tests. One test set is the failure stability test from chapter three and
the other is a compressive strength test for a series of slender coal pillars. Indicator behavior
is evaluated in the the context of failure intensity and also the size of the model. A grid
based indicator measurement technique that was developed for this study is explained and
a selection of indicators are used to analyze the stable and unstable slender pillar failures.
Chapter 5 describes the implementation of a complex hybrid model designed to simulate
a realistic mining situation. In situ stresses are installed and the coal material, as modeled in
PFC, is incrementally mined and the stress distribution in the model is allowed to readjust
after each mining increment. Failure stability is difficult to detect when failure is local,
so the successful indicators from chapter two are used to closely analyze a situation of
suspected instability. The analysis is then supported with grid grid based measurements of
5
|
Colorado School of Mines
|
CHAPTER 2
BACKGROUND INFORMATION ON UNSTABLE FAILURE IN UNDERGROUND
COAL MINING
Since the inception of underground mining, miners have dealt with unstable failures.
However, before the nineteen sixties, mitigation for this danger has depended upon miner’s
intuitionandrulesofthumb. Improvementshavebeenmadeindealingwithunstablefailures
by increasing the understanding of the physical mechanisms associated with unstable failure
andbyapplyingsophisticatedtechnologiestomitigatedangerousminingsituations, augment
mining practices, and further study the problem. This chapter presents a background of the
progression of research on unstable failure in underground mining in the context of rock
mechanics and numerical modeling.
First, the rock mechanics of unstable failure are introduced. Then a background of un-
derground coal mining is given and the geological conditions that are widely understood to
influence unstable failure are presented. A review of noteworthy modeling tools in under-
ground mining is also presented, with special attention paid to those focused on modeling
unstable failure. Finally, a brief history of application of the discrete element method for
rock mechanics is presented, with special attention given to applications in underground
mining.
2.1 The Rock Mechanics of Unstable Failure
Due to the nature of underground mining methodology, rock structures in mines are
often subjected to stresses high enough to cause failure. In order to formulate a theory to
explain the mode of rock failure it is necessary to describe the behavior of rock after failure.
Although, up until the 1960’s, no theoretical basis was available to describe the behavior of
rock after failure. The two prevailing theories in solid mechanics, linear elasticity and Mohr-
Coulomb yield with perfect plasticity were insufficient to describe the state of equilibrium
7
|
Colorado School of Mines
|
within a fractured rock mass [54]. In 1965, N. G. W. Cook provided a lasting contribution
which improved our ability to describe rock behavior after failure. He theorized that rock
behaviorafterfailureisgovernedbytheformationofcracksurfacesfromstoredstrainenergy.
The process of forming cracks from the available strain energy results in a non linear path
from peak stress to residual stress [17].
An set of laboratory compressive strength test tests are also presented in [17]. The results
demonstrate the implications of a non linear post peak curve. The theory implies that if
there is additional energy supplied by the loading system, the crack formation will not be
capable of absorbing the additional energy and failure will occur unstably, along with a
considerable release of excess energy. In the tests, similar rock specimens were failed with
two different compression machines. One machine was very stiff and one had lower stiffness,
capable of storing a larger amount of energy and hence failing the specimen unstably. The
non linear post peak theory was confirmed when the soft testing machine failed the specimen
unstably, as evinced by a loud shock, and the stiff machine failed the specimen stably with no
noticeable shock. Cook’s theoretical work and laboratory study together provided a tenant
of failure stability that will be echoed throughout this thesis. That is, when the stiffness of
the loading system is lower than the post peak stiffness of the failing material, there will be
excess energy available that cannot be absorbed during the failure process, and failure will
be unstable. Figure 2.1 is a stress strain plot that illustrates the concept of failure stability
due to the effect of loading system stiffness. The solid line represents a UCS specimen’s
characteristic behavior that is only obtainable under perfectly rigid loading conditions. The
dotted lines represent the load lines of a soft and a stiff loading system. When the loading
system is stiff as compared to the post-peak stiffness of the specimen, the material is capable
of absorbing the energy stored in the loading system through the failure process. When
the loading system is soft, there is an excess of energy stored in the loading system, with
magnitude dependent upon the angle theta, which the specimen is unable to absorb during
failure process. For unstable failure two conditions must be met. The material must fail and
8
|
Colorado School of Mines
|
the loading system stiffness must be less than the post-peak stiffness of the material.
Figure 2.1: UCS failure stability stiffness criteria, after Kias et al. [49]
Other researchers followed in Cook’s footsteps by performing laboratory testing exploring
thepostpeakbehaviorofrockusingstifftestingmachines. Theseexperimentsbothimproved
our understanding of the post peak behavior of rock and made advances in the technology
required to investigate it. In 1967, Z. T. Bieniawski conducted a broad set of experiments to
study various stages of rock failure, including post peak crack growth [9]. He obtained the
first complete stress-strain curves of hard rock and verified the dependence of failure stability
on loading system stiffness. W. Wawersik and C. Fairhurst in 1970 used a uniquely designed
machine to study the post peak behavior of rock failure [88]. Six rock types were tested:
two types of granite, marble, slate, basalt, and sandstone. They presented results for two
types of post peak behavior, Class I and Class II. Figure 2.2 shows examples of stress strain
curves of Class I and Class II behavior. Class I behavior is post peak behavior exhibiting
increasing vertical compressive strain with decreasing load, a negative slope, while Class II
post peak curves have a positive slope.
WawersikandFairhurstrelatedtheslopeofthepostpeakcurvetotheamountofavailable
strain energy in the rock specimen itself to cause failure. Class II failure indicates that
additional energy must be supplied by the system to create additional crack surfaces and
9
|
Colorado School of Mines
|
progress failure. The negative slope of the Class I specimen indicates that an excess of strain
energy is available in the specimen and the loading platens must be retracted to sustain a
stable failure process. The vertical line between Class I and Class II, as shown in Figure 2.2
denotes the boundary in which precisely the exact amount of energy required for failure is
supplied by the loading system and the specimen. The lettered regions of the Class I curve
indicate different stages of failure discussed in [88].
Figure 2.2: Class I and Class II post peak behavior, after Wawersik & Fairhurst [88]
Another careful study of post peak behavior using a stiff testing machine was published
in 1971 by Wawersik and Brace [89]. The authors inspected crack patterns in specimens at
various levels of confinement with Class I and Class II behavior. They concluded that the
fracture mechanisms in rock were highly dependent upon confinement pressure and therefore
a single failure criterion could not be used to describe failure of rock. Concerning unstable
failure, they observed that small distributed cracking was more prominent in stable failures
while unstable failures where coincident with the formation of longer and spatially focused
fractures. Different fracture patterns suggest different failure mechanisms are in effect for
stable versus unstable failure.
Additional, notable contributions to the study of rock post peak behavior were made
in the following years. They include the introduction of a testing machine with a servo
10
|
Colorado School of Mines
|
controlled loading mechanism that was based on a a lateral strain gauge feedback measure-
ment system [36]. This machine paved the way for modern testing machines with elaborate
electronic systems designed to provide stress controlled loading schemes. A review of the
progress in rock testing technology to date was provided in [35]. Then, the time dependent
nature of rock behavior using advanced servo controlled testing equipment was presented by
Peng in 1973 [63]
Elastic theory predicts a release of seismic energy simultaneous to the enlargement of an
underground excavation, a theory for the mechanism of rockburst. Elastic theory was even
used to propose a mechanism for collapse of room and pillar mines [76]. However, the energy
release predicted by elastic theory is far greater than that measured during rockburst events
[16]. Withabetterdescriptionofthepostpeakbehaviorofrockandatheoreticalexplanation
of unstable failure in place, progress was made towards understanding this discrepancy by
means of a possible mechanism for rockburst. Immediately following his seminal paper in
1965 Cook published a paper, aptly titled, A Note on Rockburst Considered as a Problem
of Instability. In this paper, he explains that as a mining face advances a region of failed
rock precedes it. This failed region is created by transfer of energy from the loading system,
releasing available elastic energy in a less violent fashion than supposed by elastic theory.
Furthermore, due to the mechanism of instability proposed in his previous paper a ‘crudely
periodic’ series of instabilities can result as the relative stiffnesses of the loading system and
failing rock change during the mining process.
It was later pointed out that unstable failure could generally be grouped as either com-
pressive or shear failures. Cook’s failure stability stiffness criteria was applied to disconti-
nuities by Salamon [77]. Ortlepp claimed that compressive and shear unstable failure could
manifest in a variety of forms including pillar crushing, buckling, shear rupture in intact
rock, and fault slip [61]. The work in this thesis deals only with the compressive type of
unstable failure. Therefore, the term unstable failure is used in this thesis to refer to general
compressive unstable failure, and not as distinguishing between types of compressive failure
11
|
Colorado School of Mines
|
as defined by Ortlepp.
Additional theories for the basic mechanism of rockburst have been proposed [74][86][83].
However Cook’s mechanism is widely accepted as the leading theory for unstable failure
in both rock and coal. So, the advances made by Cook on the understanding of failure
stability and post peak behavior form the foundation for the physical mechanism of failure
this research is based upon. The following section presents context for the application of
these concepts to a coal mining situation.
2.2 Unstable Failure in Underground Coal Mining
The study of unstable failure of rock originated in response to the violent and often
deadly failures that occur in underground mining situations. Throughout the history of
underground mining practices, unstable failures have remained a pervasive danger that are
not well understood, still to this day. While it is generally understood what conditions make
unstable failures likely, predicting the precise moment and intensity of these failures is not
possible. The mechanics of rock failure provide us with an insight to how rock fails, but
the conditions in a particular mining situation add a layer of complexity and uniqueness.
The characteristics of underground coal mining that influence unstable failures are discussed
here.
Unstable failure in coal mining is often referred to as a coal bump or bounce in reference
to the deep sound of shifting Earth reported during these failures. These failures range in
intensity and volume depending on the location and nature of the failure. For example, a
bump can result in nothing more than an audible sound in the roof or floor in the coal mine
due to slip along joints or bedding planes. It can also cause the expulsion of tons of coal
from mining faces or entry walls with fragment velocities up to the order of 10 m/s, entirely
filling the mined out area with coal and burying mine employees and equipment. It can
even result in a series of pillar failures, resulting in the complete destruction of entire mine
panels. In each of these cases, the basic mechanism for unstable failure discussed above is
in effect, but it is important to note that unique conditions for each case trigger instability
12
|
Colorado School of Mines
|
or determine the extent or intensity of failure.
The unpredictable and potentially fatal nature of coal bumps renders their in situ study
not only dangerous but also impractical from a logistical standpoint. Coal bump researchers
are forced to rely on mine incident reports for data and case studies that contain first hand
accounts of bump incidents. A limited amount of seismic, load and displacement data is also
available. Together, these sources help to provide information in establishing factors that
influence the frequency and intensity of coal bumps.
Two basic mining methods are used in mining coal. They are the room and pillar min-
ing method and longwall mining. The type of mining method can have an effect on the
occurrence of coal bumps. In both types of mining, coal pillars are developed by mining
away surrounding coal to create transport hallways, or entries. The remaining coal pillars
and secondary support provide the ground control necessary for safe mining conditions. The
sizing of pillars is important depending on the function they are designed to serve. Abut-
ment pillars are large pillars that are capable of withstanding the total overburden stress.
Abutment pillars typically have a pillar width to pillar height ratio greater than 10. Yield
pillars are designed with failure in mind, so they fail gradually during the mining process and
providethe loadingsystem a stablemeans ofreleasing of strain energy. Yield pillars typically
have a width to height ratio less than 5. Pillars sized in between abutment pillars and yield
pillars are potential coal bump hazards as they are too large to yield stably and too small to
withstand the total overburden stress [50]. These pillars are called critical pillars. Figure 2.3
is a diagram of the performance of gate road pillars for longwall mining that illustrates the
effect of pillar size on stability. Note that the yield pillars fail while the abutment pillars do
not, but they are both considered stable in their ground control performance.
2.2.1 Coal Mining Methods
In room and pillar mining, a panel of coal pillars is developed by first removing coal
in a grid of hallways, also called entries. The pillar sizes in these developed panels can
range widely in side and shape about 100 feet in width on each side. Upon completion of
13
|
Colorado School of Mines
|
Figure 2.3: Pillar size versus stability
development, the pillars are removed in a process referred to as retreat mining. Starting
from edge of the developed panel furthest from the mine’s main transport area, the pillars
are then mined out moving towards the opening of the mine. The roof of the mine is allowed
to cave in after the pillars are removed. Due to safety, pillars are not always retreated in
room and pillar mines, but it is often desired in order to maximize production.
Unstable failures are more likely to occur during retreat mining in room and pillar mines
than during development [33]. This is due to the stress concentrations at pillar lines that can
provide the load necessary to fail the pillar and then facilitate unstable failure. Pillars that
are highly stressed yet not failed can also fail unstably during the retreat process. During
the retreat mining process, pillars should yield as the pillar line approaches. The yielding
process allows stress in the roof and floor to redistribute to larger load bearing structures or
to facilitate caving inby the pillar line. If a pillar has yet to yield, a considerable amount of
energy can be stored at the core of the pillar, acting as a critical pillar.
A method of pillar removal, called pillar splitting, is used to section a pillar into smaller,
mineable parts to allow stress to redistribute safely. A bump can occur if an attempt is
made to split a critical pillar [40]. The bump can occur when the highly stressed core of the
pillar is suddenly deconfined. Sudden deconfinement results in a reduction in strength of the
pillar core, and if the loading condition is unstable, a bump can result [12]. High stresses in
14
|
Colorado School of Mines
|
critical pillars can be due to inadequate caving. In any retreat mining operation continuous
caving of the roof is desired. If roof rock remains intact, the cantilevered rock can impart
additional load onto the pillar line. Suggestions for mitigating high occurrences of bumps
near pillar extraction lines include taking preventative bump cuts ahead of the pillar line,
pillar sequencing, and complete removal of pillars to promote adequate caving [38][39][94].
In extreme cases, entire panels of pillars can fail in a chain reaction as load is quickly
shed from one failing pillar to the other and the potential energy in the loading system is
abundantenoughtopropagatefailure. AnotoriousexampleofsuchafailureistheCoalbrook
North Colliery in South Africa 1960. Thousands of 12 by 12 by 4.2 meter pillars failed in
a matter of minutes, destroying a mining area around 750 acres in size, killing 437 people
[10]. Numerous other examples have been documented and studied to determine key factors
of failure and possible mitigations procedures [13]. The effect of loading system stiffness in
these failures in discussed separately by Salamon and Zipf, stating similarly that if the local
mine stiffness is lower than the post peak stiffness of the failing pillars, the failure will be
unstable [76][93].
The longwall mining method involves the use of a technologically advanced machine that
moves mechanically back and forth along a wide panel of coal. A rotating drum attached to
the machine scrapes coal from the mining face. The coal then drops onto a moving conveyor
which transports the coal to a nearby entry and then to the mine opening. In the United
States, the panel of coal being mined is typically 300 to 400 meters in width, 1.5 to 3 meters
in height, and 3 to 4 kilometers in length or even longer. The coal shearer machine advances
throughout the length of the panel under the protection of mechanized roof supports that
advance with the mining face. The unsupported ground behind the supports is allowed to
cave. In order to mine an entire coal seam, a series of longwall panels are developed alongside
one another. These panels are mined in succession, and the entry on the trailing edge of the
mined panels is named the tailgate while the entry on leading edge of the current panel is
named the headgate.
15
|
Colorado School of Mines
|
In comparison to the room and pillar method, longwall mining is seen as a safe under-
ground mining method due to the highly mechanized nature of the process. However, severe
coal bump incidents are encountered. Similar to room and pillar mining, high stress areas
are located along boundaries where mining has recently occurred and the overburden is un-
supported. If material is not caving properly, cantilevered overburden leads to high stress on
these mining boundaries. This increase in stress, called abutment stress, occurs in two loca-
tions in longwall mining, along the mining face and in the tailgate entries. So, the abutment
stress increase is magnified at the tailgate side of the mining face. The likelihood of bump-
ing occurs when highly stressed material is unconfined by the mining process. Therefore,
increased bumping activity occurs when the coal shearer approaches and changes direction
at the tailgate side of the panel. The pillars in the gate road entries serve an important roll
in providing safe travel for mine workers, conveyance of materials, and proper ventilation.
Much importance is placed on the design of gate road pillar systems [39]. Bumping in tail-
gate pillars is a common problem that often leads modern mines to discontinue mining on
problematic panels [1][7]. In addition to the mining method, the frequency of coal bumps
can be attributed to geological conditions in the mine, discussed in detail in the following
section.
2.2.2 Geological Conditions Contributing to Coal Bumps
The geological conditions in a coal mine contribute to the frequency of bump incidents.
For example, the presence of thick competent rock in the immediate roof and floor is a
prominent feature of bump prone mines. The thick competent roof, usually sandstone, can
contribute to the likelihood of bump incidents in multiple ways. First, the sandstone roof can
resist fracture when underlying material is mined out and create cantilevered mass thereby
increasing the load on critical structures such as a pillar line or the longwall face [70][29][30].
For this reason, sand stone members in the roof of a coal mine contribute to the inadequate
caving seen in problematic mines. The competent sand stone also prevents an alternative
mode of failure in the mine called punching. Punching is the failure of the immediate roof
16
|
Colorado School of Mines
|
and floor cause by the penetration of the still intact pillar into the roof or floor material
[47]. The presence of the competent roof and floor ensure that, given a sufficient load, pillar
failure will occur rather than the floor and could then possibly facilitate a bump by means
of Cook’s unstable failure mechanism.
An additional consideration associated with competent roof and floor rock is the contact
condition. Iannacchione has suggested that a sudden slip along the coal rock interface would
result in sudden deconfinement of the stressed coal [37]. Deconfinement would lead to a
simultaneous drop in strength and lead to failure and a possible bump. This mechanism
was earlier demonstrated in the laboratory by [3]. This experiment showed that similar coal
specimens would fail stably or unstably depending on the confinement condition.
More traditional geological features such as faults or dykes contribute to bumping by
augmenting the stress field when mining approaches these features. Dykes are channels of
rock formed within the crack of another rock formation. Sandstone can be one such dyke
material that when overlaying a coal seam can lead to increased stress as a result of the high
stiffness [1]. Faults are preexisting zones of shear failure that allow shear movement more
freely than intact or fractured rock. Depending on the orientation of a fault, approaching a
fault by mining can lead to unstable slip along the fault. If the fault is dipping downward
from the approach perspective then mining activity can unload the fault in the roof and lead
to unstable roof failure. While heaving floor is possible when approaching upward dipping
faults, the risk is not as severe.
One of the foremost factors in coal bumps is the depth of the mining activity. It is widely
understood that deeper mines are at greater risk for bumping. The simple reason for this
is that for coal bumps to occur, failure must occur, and failure is pervasive in deep mines
as opposed to shallow ones. In U.S. coal mines ‘deep cover’ is defined as being greater than
1500 ft (457 m), but mines at depths greater than 1000 ft (305 m) are at risk [57].
Coal is a brittle rock, which means that it fragments upon failure and has little ductile
deformation capacity. The brittleness of coal makes it susceptible to bumping in contrast
17
|
Colorado School of Mines
|
to more ductile materials such as salt. Complete stress strain curves for in situ coal pillars
have been obtained of by Wagner and by Van Heerden [85]Van Heerden [84]. These tests
results show an in situ coal behavior that has a defined post peak softening characteristic
that decreases in slope as the width to height ratio of the specimen increases. The brittleness
of rock is key to fulfillment of the Cook failure stability mechanism, in so far as for unstable
failure the loading system stiffness must be less than the coal’s post peak stiffness. It has
also been proposed that pre-peak stiffness of coal is relevant to the intensity of failure [64]
2.3 Numerical Models and Unstable Failure
Due to the cost and logistical issues associated with experimentation, numerical models
have become a popular means for investigating the failure of rock. The behavior of rock in
miningsituationsrequiresanunderstandingofhowtherockmasswillbehavepriortofailure.
As explained above, the mining process necessarily leads to the failure of surrounding rock.
The surrounding rock is an agglomeration of layered, jointed material that behaves according
to the properties of the joints and rock itself at all stages of stress and strain. Numerical
models provide a tool to understand the key mechanisms at work on the process in question.
The methodology by which one investigates such processes using numerical models is an area
of contention. For example, it may seem desirable to build a complex model that contains
all of the geological features of a rock mass. Although, Starfield and Cundall argue that
simplification is needed in order to allow for comprehension of model results and thoughtful
iteration of the experimentation process [79].
In this section, I will discuss a range of numerical modeling tools that have been used
for modeling stress and strain in underground mining and discuss their capability to capture
unstablefailure. Then, Iwillpresentasetofimportantstudiesthatarefocusedonsimulating
unstable failure. Using these examples as a basis, I will demonstrate that there is a well
defined need for improving the ability of numerical models to simulate stable and unstable
failuremodes. Thediscreteelementmethodisaprovenmethodformodelingrealisticfeatures
of rock failure and as this thesis will show, it is capable of capturing stable and unstable
18
|
Colorado School of Mines
|
modes of failure. So, a thorough review of rock failure simulated using the discrete element
method and advanced techniques and capabilities are presented.
2.3.1 Underground Mine Models
Modeling stress and displacements in underground mining is a difficult task due to un-
known geological features and properties. Through simplifications of the model such as
idealized geometry and assumptions on material behavior, key features of mine behavior
can be satisfactorily to allow for study. It is important to be aware of assumptions and
limitations of each model in order to prevent erroneous interpretation of results.
Avariationontheboundaryelementmethodcalledthedisplacementdiscontinuitymethod
has been used in various programs. The program MULSIM/NL uses this method to model
up to four parallel seams with optional non-linear behavior and elastic non seam material
[92]. Another displacement discontinuity method program similar in structure, LaModel, is
seen as having surpassed MULSIM/NL [32]. LaModel uses a lamination formulation that al-
lows for more closure in excavated areas and matches subsidence observations better. These
programs do not allow failure in any of the non seam regions.
The company Rocscience offers a series of programs for simulating stress and strain
around underground openings. The boundary element programs Examine2D and Exam-
ine3D are boundary element programs that use the elasto-plastic material models and the
modified Hoek-Brown failure criterion to simulate rock behavior [71][72]. The finite element
program Phase2 utilizes the same material models but allows for polygonal material zones
for customized geometries and multi stage calculations in order to simulate excavation pro-
cesses [73]. Due to their explicit nature, these programs are incapable of converging on a
solution if a physical instability occurs.
Including accurate simulation of failure stability in a mine model is not always a priority.
However, such a model could aid in the preventative design of less bump prone mines.
Mark provided an alternative to complex mechanical models with the programs Analysis of
Retreat Mining Pillar Stability (ARMPS) and Analysis of Longwall Pillar Stability (ALPS)
19
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.