Abandoned mines are a recurring problem for today’s mining industry. In coal, metal, and nonmetal mining they pose a safety hazard if current workings intercept them; and if they are circumvented, large amounts of recoverable resources may be unnecessarily left in the ground.

The problem of delineating the boundaries of these old mines is multifaceted. In most cases, entry is precluded by flooding, caving, or extremely unsafe conditions below ground. In many cases the company that did the past mining has gone out of business, or the old mine maps were incomplete and/or inaccurate, or are missing. Particularly in the case of coal mining, using room and pillar methods, extraction was sometimes less than 50%. Therefore a drilling program is as likely to encounter coal as a void throughout an old mine area.

The resistivity method developed for detecting these old mines has been designed to allow rapid coverage of an area suspected to contain old workings and to delineate the boundaries of these workings where they exist; although it cannot reliably detect individual openings at all depths, it appears to detect boundaries of mined and unmined areas reliably. The method employed was originally developed by Bristow to locate shallow caverns and passages in limestone karst topography. It has been modified by the Bureau to locate air- filled or water-filled voids in mining environments.

This report covers both the results of earlier contract research and 2 yr of in-house research involving six mine sites in four States (Illinois, Wyoming, Colorado, and Kentucky). A total of 16 test lines were run at these sites, and corroborative drilling tests were made at 3 of the sites. The sites encompassed both shallow mines and mines that occurred near the theoretical detection limits.

Resistivity Theory & Void Detection

A mine void in the earth (in an electrical context) may be thought of as a “system” composed of (1) rock and soil, (2) the interstitial free water contained in the soil or rock, and (3) the mine void itself, which may be either air or water filled. The rock materials themselves have very high resistivities (hundreds or thousands of ohm meters) and may be considered as mainly a matrix or framework for the contained free water. It is this water and its dissolved salts that allow conduction of the electrical current for our measurements. Air also has an extremely high resistance and may be considered as an insulator at the voltage levels used for the tests.

While the effect of large discrete voids is readily apparent, small cracks or voids, when occurring in extremely large numbers, may easily show the same overall effect. In a mining environment, these can be microfractures or delamination phenomena above the void which may migrate upward as subsidence phenomena. These cracks may be either air or water filled and will normally be interpreted as void areas.

Electrical resistivity measurements are performed by injecting an electric current into the ground and measuring the resultant voltage potentials generated by this current. Since the method’s inception over 50 yr ago, many different electrode configurations have been used for a variety of purposes. The pole-dipole (Bristow) array is but one of the configurations used. The two other more commonly used arrays are the Wenner and the Schlumberger. For illustrative purposes the Wenner and Bristow arrays will be compared.

The Wenner method uses two current and two potential electrodes in a linear, symmetrical array (fig. 1). The current is injected between the two outer cur-rent electrodes, and the potential difference is measured between the inner pair. As can be seen from the figure, the measurement volume is a function of the electrode separation. To make deeper measurements, the electrode spacings are increased, which results in a larger hemisphere being measured. Theoretically, this electrode expansion can be continued to make measurements to any depth. However, in practice, the volume of the hemisphere becomes so large that a mine void imposed within it would be undetectable.

The equation for calculating apparent resistivity (pa) for the Wenner method is

pa = 2πa E/I………………………………………………………………(1)

where I = current injected between outer electrodes,
E = potential difference measured between inner electrodes,
and a = electrode separation distance (as shown in figure 1).

It has been found that the void detection capability of the method is limited by the ratio of the diameter of the void to the depth of overburden material. The ratio of either the Wenner or Schlumberger method is normally taken to be 1:4 or less.

The Bristow method uses one current electrode near the measurement site and another, the sink electrode, at effective infinity (5 to 10 times the maximum depth of investigation). The two fixed-spaced potential electrodes are moved as a pair away from the current electrode in

order to make deeper measurements (fig. 2). The measurement volume involved in this case is a hemispherical shell whose thickness is equal to the potential-pair separation and whose radius is equal to the current-potential electrode separation.

The equation for calculating apparent resistivity (pa) for the Bristow method (pole-dipole) is

pa = 2π/(1/r1 – 1/r2) E/I………………………………………………………………..(2)

where I = current injected between current and sink electrodes,
E = voltage difference measured between potential electrodes,
r1 = distance between current electrode and first potential electrode,
and r2 = distance between current electrode and second potential electrode.

As is apparent, for a given depth, the volume of material being measured by this method is much smaller than for the Wenner, and the detection capability for a void is increased to a void-overburden ratio of 1:10, or under ideal conditions to 1:15.

The term 2π E/I is the same for both equations 1 and 2. However, the geometry factors “a” for the Wenner and 1/r1 – 1/r2

for the Bristow—are vastly different and become even more pronounced as the depth of investigation increases. As an example, for a 5-m depth, the volume measured by the Wenner is 262 m³ compared with 205 m³ for the Bristow, using a 2-m potential separation. In this case, the volume of the Bristow is 78% that of the Wenner. However, for a 50-m depth the Wenner volume is 261,792 m³ and the Bristow volume is 30,175 m³, or 11.5% that of the Wenner (fig. 3). From the above, it is obvious that the Bristow method is much more sensitive to the presence of voids than the other methods as the void represents a much larger percentage of its measurement volume.

However, even the Bristow method could theoretically detect a 2- by 2-m void at a maximum depth of only 20 to 30 m. From this it is inferred that some other phenomena are at least partly responsible for the deeper apparent voids that have been detected during this research. The concept of microfracturing above an opening in the earth appears to correspond with much of the field data acquired to date. The effect of subsidence features migrating or stoping upward from the void itself would also correspond with much of the data. One or both of these effects may be partly responsible for the fact that known voids are often shown higher in the section than their known depths. Still another possible reason for being able to detect voids at greater than theoretical detection depths might be that the electrical current distortions caused by multiple voids may be larger than those from a single void.

Which of the above phenomena is responsible for the ability to detect old mines is not known. It has been shown, however, that the deeper voids are detectable, although they generally are shown to be above the known or suspected depth of the void. This vertical transposition of the data seems to be a function of depth. That is, deep voids tend to show

more vertical displacement than shallow voids.

In some cases the interpreted void is apparently skewed slightly toward the infinity electrode. In a short series of tests performed by Southwest Research Institute (SwRI), it was found that the potential field around the current electrode is not circular in plan view, but is rather of an elliptical or tear-drop shape with the long axis parallel to the current-sink electrode axis. This shows that the current-sink electrode separations, classically assumed to be effectively infinity, are not sufficient.

One technique that will detect skewness and allow for correction is to run the test line twice—with the sink electrode being placed equal distances from opposite ends of the line. This requires a double effort, but in some cases may be worth the extra work involved in that the skewness can be detected and removed.

Description of Automated Resistivity Method

In explanation of the Bristow method above, the original techniques were described- that is, the potential pair electrodes were moved incrementally away from the current electrode as deeper readings were taken. In effect this results in a vertical resistivity profile of short lateral extent being developed. To cover a greater lateral area, the current electrode would then be moved ahead and the process repeated. While feasible, this process would require several days. Therefore, specialized equipment was developed for the Bureau by SwRI which allows much faster and more accurate data acquisition as well as digital data recording. This equipment, referred to as the automated resistivity system (ARS), is designed as a rugged, automatic system for performing Bristow-type surveys with a minimum of human intervention. The system is described in appendix C.

In operation, 1 to 64 current electrodes are implanted at the desired positions. These electrodes are then connected to individually addressable control modules, which are in turn connected to the current-address cable. This cable is controlled by the system and allows individual, sequential energization of the various electrodes.

Therefore, by setting the pair of potential electrodes once and reading the resultant voltages as each current electrode is automatically energized, a large number of readings may be taken in only sightly more time than would be required for one reading in a manual mode. These readings are recorded on digital magnetic tape for later interpretation. The potential electrodes are then moved ahead one increment, and the process is repeated.

Using this technique, a test line 640 m long with current electrodes implanted every 10 m, sampled to a depth of 100 m in 2-m increments, can be completed in 1 working day using a field crew consisting of four or five persons. It has been found that two such lines may be run in 3 days. The extra day would be spent in laying cables, planting current electrodes, surveying, etc., in preparation for performing the tests themselves. A more complete description of the detailed field techniques and work assignments is given in appendix D.

Data Interpretation

A large amount of data is recovered on each test line, and a computer is required for their interpretation. To date, two separate interpretation methods have been used. The first, developed by SwRI, constructs a computer model and inserts a void at some discrete location within the section. It then compares the field data with this model and assigns a correlation number in the range of 0 to 1, depending on how well the field data correspond to the model. It then iteratively changes the location of the void in the model and compares its correlation with the data. Upon completion, a profile of the section is drawn based upon these correlation numbers. This results in contours where the higher numbers represent the higher probability of the void’s location. The computer model may be used to look for both air-filled and water-filled voids. This method is referred to in this paper as the probability contour model.

The second method, used by the Bureau, simply calculates the resistivities of the field data and plots the information as a simple pseudosection beneath the current electrode positions. Based upon this profile section of resistivities, the interpreter makes decisions as to the location of voids. The second method uses much less computer time and allows the interpreter to interject personal knowledge of the area into the final interpretation. A description of these computer programs and their logic appears in appendix B.

Discussion of Method

The technique that has been described is a rapid, inexpensive method for investigating a large area in order to define the boundaries of old workings. However, it must be realized that phenomena other than old workings can cause some of the anomalies noted. Therefore, when interpreting the results, judgments must be made as to what the anomalies shown by the data actually represent.

For example, at Sheridan, WY, the Monarch Mine was defined clearly by resistivity lows. These corresponded with the known edge of the old mine as well as with surface subsidence features. However, our drill tests at the nearby Acme Mine showed that one particular low anomaly drilled was caused by a wet silt pocket. In both cases the data were correct—there was an accumulation of moisture that caused a resistivity low. In one case it had collected in the silt, while in the other it had collected in the subsidence-caused loose material.

At the other extreme was the case at the Busick Mine, in Kentucky, where a known air-filled working was masked by an artificially created, perched water table located nearer to the surface. Surface water leaching through the spoil material and perching near the original ground surface created a highly conductive layer that tended to act as an electrical shunt and prevented a majority of the current from reaching the old workings below. While this was an isolated incident, it illustrates that judgment must be exercised in interpreting the data.

In general, the method may be used in most terrains. However, buried electrical conductors may adversely affect the results and should be avoided.

The current flow paths are assumed to approximate those shown in figure 2 when the data are calculated and displayed for interpretation. If a nonearth flow path exists between the current and sink electrodes or between the current and potential electrodes, the model and associated equations for apparent resistivity are not valid and the calculated values will be in error. For this reason, it is not advisable to perform resistivity tests where any metallic conductors, such as pipelines or fences, are present.

In coal mining, it must be considered that the old mines in this country were of the room and pillar type, and extraction, in some cases, was only about 50%. From this it can be inferred that 50% of the coal remains in these areas. In ordinary drilling tests, the drill may not encounter the old mine voids and so they will remain undetected. However, in the case of a resistivity survey of a mine area, the workings will probably be detectable. This is because resistivity is an areal technique, whereas drilling provides a single-point measurement.

Also, as pertains to drilling, it must be closely monitored and even subtle changes in the drilling noted. Loss of water or air, even for a short distance, should be logged. Softer, more easily drilled material at or above the old mine level should be noted. Unusual moisture conditions (in the case of pneumatic drilling), either within a hole or between adjacent holes, may be important in making the interpretation. Furthermore, to assure that complete observations are made, a trained observer should be at the drill throughout the drilling program, instead of having the driller make the observations. Drillers are trained to do the most rapid and efficient drilling in given situations, rather than as observers. Even the most conscientious driller may have stepped away to do minor maintenance work on the drill when some transient feature is encountered; thus it passes undetected. In other cases, the drillers’ background does not show them the importance of what they see. For example, on one project it was noted by the driller that the hole did not encounter any evidence of mine workings. However, air blew out of a nearby hole, which had also bottomed at the old mine level. In many cases this would not have been entered in the driller’s log. To the driller it was only a curiosity, while to a trained observer it could represent important evidence.

Field Test Results

The results below briefly summarize each of the six test sites investigated to date. Data analysis and other pertinent information on each site appear in appendix A.

Central City, Ky—Busick Mine

Four lines were run at this site by SwRI under a Bureau research contract. This was an active mine which contained abandoned, flooded sections, as well as air-filled workings. For three of the lines surveyed, resistivity high and lows were indicated in close proximity to known air-filled and water-filled mine workings. The anomalies were consistently shown shallower than the known depths of the mine. The fourth line was run across a large, flat strip mine spoil area which had been emplaced over an air-filled section of the mine. The natural ground beneath the spoil area was a heavy clay material. In this case there was no indication of the mine in the data; however, an almost solid layer of resistivity lows showed at about the interface between spoil and natural ground. It was theorized that surface water was percolating through the spoil and leaching out soluble minerals, thus forming a highly conductive layer perched on the old natural ground surface. This, in effect, shunted the electric current from penetrating deeper and prevented us from seeing the underlying resistivity highs from the mine workings.

Marshall, Co—U.S. Geological Survey Test Area

At this site the same physical line was run twice, with the current sink electrode at opposite ends of the line. In this manner it was possible to estimate the amount of skew the data would exhibit toward the sink electrode. The mine workings are 70 to 100 yr old.

The test line was laid out to cover both an abandoned mine and an equal area of unmined coal. It was later determined, however, that the supposedly unmined area had Indeed been mined by another company. The original overburden thickness in the area ranged from 14 to 26 m. Based upon resistivity data, three holes were drilled. Two in high anomalies encountered voids, the third was in a low anomaly and encountered coal with free water at the base of the coal.

Danville, IL—VJ-Day Mine

This line was run in the same manner as the Marshall test with reversed current sink electrodes. The mine had been closed for 25 to 30 yr. Complete mine maps of the section were available, and surface control was excellent. However, during the field tests it was discovered that after the mine officially closed during the 1950’s, several small groups of miners periodically mined for the next 6 to 7 yr. The depth to coal was approximately 45 m.

The data show persistent low anomalies at about the expected mine elevation, corresponding in a general way to the mined-out areas. An east-west set of main entries near the center of the line shows much wider anomalies than the entries themselves, and it is speculated that this might have been where the illegal mining took place.

Sheridan, WY—Acme Mine

This site was selected because it offered a deeper target, 65 to 75 m, and had good location control between the surface and the old mine workings. This test line was run in a reverse sink electrode configuration similar to that used in the Marshall and VJ-Day Mine tests. At this site the terrain was rolling, sage-covered, open land without habitation. For this reason the infinity electrode separation was extended to approximately 1,830 m without interference. There may be some lessening of data skewness on these tests, but it is not significant.

Most of the test area is underlain by sandstone at shallow depth. Near the center of the line many cracks and open fissures to 2.5 m or more in depth were observed. As other, shallower overburden sections of the same mine were showing spectacular massive subsidence, it was speculated that this broken sandstone might also be subsidence related. The test line was established to center it over the old mine boundary, as shown on the mine map, as the company presently operating in this area had indicated there was some uncertainty as to the exact boundary. In addition, the line was positioned well south of a short series of “stub entries” shown on the mine map.

When trying to reconcile the data with the mine plan, it was learned that, while the working mine map was dated 1-3-38, the mine had continued operation until April 1941—a period of 27 months. It was not determined precisely where mining had taken place during this time, but it was somewhere in the vicinity of the test area. Corroborative drilling was attempted the next summer to verify the anomalies detected. At this time, active strip mining of the overlying coalbeds was in progress. In order not to interfere with the mining operations, drilling was accomplished in only one small area to the west of the mapped mine workings. On the basis of these holes it was determined that the low anomaly in this particular spot had probably been caused by a moist silt pocket. We were unable to drill any of the known mine area anomalies.

Sheridan, WY—Monarch Mine

This site offered a shallower mine than the Acme (approximately 20 m) in the same geologic setting. The edge of a large panel had been precisely located during previous work by the Bureau and Office of Surface Mining on a mine fire at this location. The area near the test line showed extensive subsidence, and several craters and fissures were very close to the line.

The data show the edge of the mine quite clearly with a slight skew toward the infinity electrode. Several subsidence features seen on the surface also show in the data. It was not deemed safe to drill this site owing to the subsidence and rough surface topography.

Sparta, IL—Holiday Mine

This test site was used for two reasons. First, it offered a target 80 to 85 m deep, which is near the theoretical maximum depth of the ARS techique, and second, it was an area of extreme interest to a mining company planning underground development near the suspected limits of the old mine.

There was no mine-map control at this site. The mining company had performed some drilling in the area but had encountered a void in only one of four holes drilled.

Field tests were conducted during 1981 and 1982. During both field seasons rainy weather was encountered, which caused the surface materials to be moist for several days at a time. While this helped in making electrode contact with the soil, it may have caused some surface “shunting” of the current, similar to what happened on line 4 at the Busick Mine in Kentucky. The 1981 tests consisted of a north-south line starting near the drill holes and progressing southward to what was assumed to be well beyond the old mine. A second line was started about midway along the first line and ran westward 700 m. It appeared that the old workings were visible in the data and extended farther than suspected by the mining company. For this reason, additional tests were performed in 1982. Line 2 was extended westward and line 1 southward. In addition, two more long north-south lines were run to the west of line 1. On the basis of these tests, an assumed “dangerous” perimeter was drawn around the old mine, and a zone where mining should only be approached cautiously was drawn to the south of the danger zone. While the old mine was not completely delineated at this site, enough information was obtained for the mining company to make a judgment as to the boundaries of the old workings.

Summary and Discussion

The six mining areas investigated to date have ranged from shallow to deep in both wet and dry environments. While this number of mines cannot represent all possible conditions that might be encountered, it does offer a broad range of test conditions.

In general, the tests have been successful. In some cases the tests answered the question as to the location of the old mines but raised other questions that were not always satisfactorily answered. Examples of such questions follow: why the anomalies tend to be translated in the data toward the current sink electrode, why the anomalies tend to show higher in the section than the mine’s known depth, and why only portions of the old mines seem to show, when a more general occurrence of anomalies would be expected.

Of 16 test lines run, only 1 failed to detect a known void. This was line 4 over the Busick Mine in Kentucky, where a low-resistivity layer between the spoil and the natural ground prevented current penetration. The other extreme, prediction of a void where none exists, has not occurred. However, the western end of the test line at the Acme Mine, supposedly unmined, showed anomalies similar to those of the mined portion. The one small area drilled showed that a wet silt pocket had probably caused this particular anomaly. Other causes remain unknown because it was impossible to drill other areas along the line.

As mentioned earlier in this report, a 2- by 2-m void should only be detectable under ideal conditions, assuming a 15:1 overburden-void ratio, to a maximum depth of 30 m. It has been noted in the data analyzed to date that anomalies deeper than 25 to 30 m are, in general, less well defined than shallower features. If microfracturing or a similar phenomenon is responsible for creating a larger target than the actual void, this loss of definition may be attributable to a “halo” effect of cracks around the void. While this is speculation, it would account for some of the features noted in the data from many of the test sites.

Of the 16 test lines, 15 (91%) tended to show mine voids near their known or suspected locations. However, at the present state of the technology, the human interpretation of the computer out¬put is the most critical element in the system. Knowledge of the geology, lithology, and general characteristics of the mine involved such as depth, approximate location, and whether air or water filled is a prerequisite for accurate interpretation of the data. Where corroborative information is missing, ambiguous interpretations can occur. Therefore, all available project information such as drill-hole information, geologic or lithologic logs, mine maps, etc., should be used during data interpretation.

Conclusions

The automated resistivity method offers a rapid, inexpensive way to detect old mine workings and to delineate their boundaries if the workings are within 100 m of the surface. Individual features within these workings such as entries or small panels are difficult or impossible to define. Owing to the uncertainties of drilling, the method can locate old workings more easily and quickly than is possible with drilling programs. However, drilling tests or other corroboration should be attempted whenever possible.

Followup research is being conducted to increase the depth of detection to at least 200 to 300 m based on the concept of “focused resistivity,” which promises to increase both the investigative depths and the resolution of targets related to old mines.