Introduction to CMD Prediction

From GARDGuide
C5.1 Introduction
C5.2 Overburden Sampling Considerations
C5.3 Predictive Factors Other than Overburden Analysis
C5.4. Acid Base Accounting (ABA)
C5.5 Other Static Tests
C5.6 Kinetic Tests

C5.1 Introduction

The greatest difference in approach and technology between hard rock and coal mining lies in the area of prediction. The underlying differences (geology, economics, regulations, etc.) between the two all come into play. As discussed in Section 2A, coal deposits are very different from most other mineral deposits. Except in completely new coal districts, the nature of sedimentary rock makes it relatively easy to predict where the coal can be found. Exploration is not generally an important issue; instead, property rights and land access to the coal seam(s) are more likely to determine where a coal mine is placed.

This section starts by discussing overburden sampling, which is critical for prediction of CMD quality at surface mines, less significant at longwall operations, and even less significant at other underground coal mines, where the nature of the coal seam and the immediately adjacent strata dictate eventual water quality. This section only introduces the concepts and some of the challenges that must be considered. However, representative sampling is critical: to quote Block et al. (2000), “precise analyses performed on an unrepresentative sampling plan will, at best, accurately characterize that unrepresentative population.” Failure to accurately predict post-reclamation water quality is probably more often due to an inadequate sampling strategy than to poor analytical techniques.

The following section discusses other important approaches to predicting post-mining water quality; valuable information is provided by water quality at adjacent mine sites and pre-mining water quality. Overburden analysis, otherwise known as static testing, focusing principally on acid-base accounting is examined next. Finally, kinetic tests, which are much more typically used in hard rock mining operations, but are sometimes warranted at coal mines as well, are examined.

C5.2 Overburden Sampling Considerations

The objective of an overburden sampling program is to provide statistically valid estimates of the pyrite and carbonate content of the overburden strata. At surface mines, the entire overburden continuum is important. At underground operations, the coal seam and the strata immediately surrounding the coal seams are most important, while strata further away from the seam being mined may be undisturbed by the mining process and therefore irrelevant. In either case, the amount of rock being sampled compared to the amount of rock being mined is minimal, and therefore there must be a fair degree of consistency or at least a consistent trend that can be generalized across the entire site. With the exception of deltaic deposits, sedimentary rock is usually relatively consistent (laterally pervasive) on the scale of a mine site. However, there are exceptions. Brady et al. (1988) documented five surface mines operating near each other on the same seam in Fayette County, PA, USA, that were generating markedly different water: both acidic and alkaline CMD. He found that that there were abrupt lateral changes in geology which caused some sites (the ones with alkaline CMD) to have calcareous shale strata while the acid generating sites did not. However, this is unusual. Limestone, if present, will be fairly consistent in carbonate content. Sandstone or shale, if calcareous (e.g., calcite cement) at more than one sampling location, will likely be equally calcareous across the site (Caruccio and Geidel, 1982; Tarantino and Schaffer, 1998).

Pyrite content in sedimentary strata is somewhat more variable (Rymer and Stiller, 1989). Channel sandstones are especially variable, due to changes in flow rates over time and area. Ideally, by collecting samples of each geological unit at various random locations across the prospective mine site, consistency in the amount of pyrite and carbonate minerals in that unit across the site and their relative proportions in each unit can be assessed.

Overburden samples are usually collected by air rotary or core drilling; augering is not recommended. When drilling with an air rotary rig, rock chips are collected as opposed to cores, which introduces one additional variable to consider. Samples from one unit can be contaminated with rock chips dislodged from another, depending on the sampling procedures being followed and the type of rotary rig being used. This effect can be minimized (Block et al. 2000; Noll et al. 1988), but in general, cores provide better samples than rock chips.

Because these holes are being drilled almost exclusively for overburden sampling (exploration drilling is usually not required), there may be a tendency to drill as few holes as possible to minimize costs,. For example, in Pennsylvania, the absolute minimum is two sample points, with one additional hole requested for every 40 ha (100 acres) to be mined. However, the standard practice in Pennsylvania is to require 6-7 holes for every 40 ha being mined, though this ratio is sometimes reduced if there is information available from adjacent mine sites. Where predictions are more challenging due to geology (e.g., deltaic coal deposits are notoriously hard to assess, as discussed in Section 2A), operators will typically drill more holes. For example, at one site, the ratio was about one hole per ha (Brady et al. 1994).

Weathering results in the near-surface depletion of pyrite and carbonate minerals, and, if not considered, can lead to erroneous conclusions on the nature of a given geological unit (Figure 1). The depth of this depletion varies with the lithology and the climate, but unless the site and the strata are both flat, rock that is weathered in this manner at one location will be buried deeper and unweathered elsewhere on the same site. Generalizing that the unit is relatively inert based on the weathered sample, or that alkalinity is present in the unit across the site, based on a sample taken from the unweathered unit, could lead to errors.

Figure 1. Depletion of near-surface (at this site, the top 5 m) total sulfur at a prospective coal mine
site. Samples should be collected from both the weathered and unweathered material.


The drill holes should be located to provide as much useful information as possible. So holes should be placed to align with planned mining phases, with at least one hole in the initial mining phase. Also, at least one hole should be located at the maximum highwall height, and at least one hole should be located at an area of much less cover, to account for the effects of weathering (Block et al. 2000)

Underground mine operations must consider the pyrite content of the coal and adjacent strata that will be left behind when mining is completed. Since pyrite varies more than carbonate content, more intensive sampling will likely be necessary to predict post-closure water quality. However, predicting the eventual post-closure water table is even more important since that will determine if there will be a post-mining discharge.

The preceding discussion applies to horizontal variability. Vertical variability is almost a certainty, regardless of the geology. All distinct geologic units should be sampled separately and thicker units should be sampled at every change in appearance or, if the rock appears to be consistent, at every 1-1.5 m. If interbedding makes it impossible to sample individual units, then samples may have to be aggregated and averaged. Weathered strata should be sampled separately from rock that appears to be unweathered. More information on this topic is readily available (e.g., Block et al. 2000; Tarantino and Schaffer 1998).

However, overburden analysis uses only a portion of the samples taken from a given interval. Acid-base accounting (ABA) typically requires only 1 g for sulfur analysis and 2 g for the NP test. Thus, changes in the geochemistry within the vertical extent of the sample introduce another potential source of error, which is typically resolved by crushing the entire sample (or, if from a core, a longitudinal half of the core interval) and then using a riffle or splitter to obtain a representative sample.

C5.3 Predictive Factors Other than Overburden Analysis

Groundwater reflects the mineralogy of rocks and soils that the water has contacted and so can be used to confirm the presence of carbonates in mine site overburden (Perry, 2000). Brady (1998) suggests that alkalinity concentrations should exceed 50 mg/L in deeper groundwater if there are significant amounts of neutralizing minerals present. Further, if the ground water samples are alkaline but overburden samples do not indicate that calcareous rocks are present, the overburden sampling may not adequately represent the site. However, his study also showed that groundwater quality cannot be used to estimate the amount of pyrite, since it only oxidizes upon exposure to the atmosphere. Brady (1998) observed that sulfate concentrations were unrelated to the amount of pyrite present.

Water quality from mining operations at adjacent sites is generally more useful. Large areas can be accurately predicted to be either acid producing or non-acid producing, since their geology is similar. The most obvious examples are most of the coal mines in the western United States, where because the deposits are associated with fresh water paleoenvironments, there is very little pyrite in the coal or overburden. Similar trends can be observed in most coal mining regions where an extensive coal seam is being mined and the overburden is reasonably consistent. However, exceptions do exist, as documented by Brady et al. (1988) and discussed earlier. Nonetheless, the water quality at adjacent mines is often an accurate predictor of water quality at a proposed site. Brady (1998) documents numerous examples that demonstrate the usefulness of this technique and how it has been applied. He states that, “Groundwater quality from previously mined areas, when available and used properly, can be the best mine drainage quality prediction tool in the tool box.” That is because it already represents many of the variables that cannot be assessed in the laboratory, no matter how many samples are analyzed. As a result, when the results of predictive tests seem to contradict each other, water quality at adjacent mines should be given precedence (Brady, 1988).

C5.4. Acid Base Accounting (ABA)

ABA is one of the standard means of overburden analysis characterized as a static test, since it ignores the effect of time-related factors such as weathering. It owes its ancestry to a mine soil classification system developed to determine how much agricultural lime should be added to support plant growth (Skousen et al., 2000). West Virginia University researchers recommended that sulfur profiles and neutralization capacities be determined for all strata down to and immediately underlying the coal seam to be mined (West Virginia University, 1971) and the ABA methodology was subsequently described by Grube et al. (1973). Sobek et al. (1978) presented the field and laboratory procedures in more detail. Subsequent passage of the Surface Mining Control and Reclamation Act in the U.S. virtually required its use since there were no other tools available at the time to characterize overburden material prior to disturbance (Skousen et al., 2000).

If pyrite is not the dominant form of sulfur present, then the actual amount of pyrite can and should be determined. For example, coal mining sites in the western U.S. sometimes generated relatively high acid potential values in the laboratory even though they contained virtually no pyrite. Organic sulfur (which is complexed within the plant material that makes up the coal and does not produce significant amounts of acidity) in the coal, and non-acid forming sulfate salts such as gypsum, are measured in a total sulphur assay. Therefore pyritic sulfur, rather than total sulfur should be measured at sites where the nature of the total sulfur present is unknown.

Initial simple use of ABA in the eastern and mid-western U.S was not as accurate as hoped. A U.S. Bureau of Mines study (Erickson and Hedin, 1988) of sites that were not easy to predict (sites that were clearly dominated by high NP or AP values were intentionally excluded) generated the data shown in Figure 2.

Figure 2. Sites where actual water quality exactly matched predicted water quality should fall on or near the blue
line; any square in the upper left or lower right quadrants represented completely erroneous predictions. Thus, if
accuracy is defined as merely predicting whether the CMD would be acidic or not, predictions were correct only about
2/3 of the time for these sites.


Although comparison of results for individual rock units worked well in the laboratory, water quality predictions done by multiplying the NNP values by the thickness of the units and the area represented by each core sample being analyzed were often inaccurate. Brady and Hornberger (1990) suggested that ABA only be used qualitatively (acid vs. non-acid) and that NP would have to significantly exceed MPA to produce alkaline water. The Pennsylvania Dept. of Environmental Protection showed that an NNP value of 12 tons per 1,000 tons effectively separated sites that produced non-acidic CMD from those that did not. NNP values between 0 and 12 produced both acid and alkaline sites; these were interpreted as sites that can only be permitted if special procedures are used to prevent acid generation. Sites with NNP values less than 0 were likely to produce acidic CMD (Brady et al. 1994). Alternatively, a ratio of NP:AP (NPR) can be used. A ratio less than 1 represented a likely acid producer while a ratio greater than 2 would be alkaline; ratios between the two are variable and will likely require some alkaline amendments or special handling.

Figure 3. A graph of some mine sites operating in Pennsylvania in the early 1990s, revealing how an NNP cutoff
value of 12 tons per 1000 tons can be used to predict sites where alkaline drainage will be produced.


Siderite (iron carbonate), if present, affects the NP determinations at some sites since it is measured in NP determinations even though it produces neither acidity nor alkalinity. The alkalinity initially generated as the carbonate dissolves is neutralized by the acidity generated as the Fe3+ hydrolyzes. Leavitt et al. (1995) proposed a modified method to determine NP. This modified ABA methodology was subsequently assessed by Skousen et al. (1997) and found to effectively compensate for the errors introduced by siderite while dramatically reducing variability in NP determinations among analytical laboratories. In general, this modified version of ABA is now the accepted standard method for CMD.

C5.5 Other Static Tests

Other methods to predict CMD water quality have been suggested, though none have yet displaced the modified ABA method. These include ABA down-hole prompt gamma ray spectroscopy wireline logging (Skousen et al., 2000), which is a field technique, and evolved gas analysis (Hammack, 1987; Skousen et al., 2000). Both of these approaches are relatively expensive compared to ABA.

However, a modified version of the NAG test is becoming common in the southeastern US and may someday displace ABA analysis. As discussed in Section of this Guide, the NAGtest is an Australian approach to static testing of non-organic rock strata that is useful in assessing the ARD potential of prospective hard rock mining operations. It can differentiate the effects of forms of sulfur and siderite. However, the presence of organic material (e.g. coal) in sedimentary strata made it inappropriate for predicting CMD quality. Recently, Stewart et al. (2006) modified the method to extend it to coal mining operations and have recently tested it, using coal washery reject material (coal refuse) (Stewart et al., 2009).

C5.6 Kinetic Tests

Kinetic tests (e.g., leaching columns, humidity cells) are discussed in more detail in Section of this Guide. As stated there, the fundamental concept is to simulate the cyclic wetting, drying, and flushing that occurs over time at every mine site. Their chief advantage is that they can be used more quantitatively than static tests, to simulate the relative rates of acid generation and alkaline production, the relative concentrations of net acidity, metals and sulfate concentrations, and the potential value that would be gained by alkaline amendments (Perry, 2000). However, kinetic tests are much more expensive and time consuming than static tests.

Before making a decision regarding kinetic testing, readers are encouraged to consult Geidel et al. (2000) and Hornberger and Brady (1998). Several researchers, as part of the U.S. Acid Drainage Technology Initiative (ADTI), have been developing a standard kinetic test procedure (Hornberger et al., 2005: The U.S. Environmental Protection Agency and the ASTM may designate this procedure as an official standard method in 2009 or 2010.