Synopsis

Excessive build-up Q/methane and waterpressures was identjftedas the leading cause Q/some Q/the rooffalls ina coal mine in theeastern Transvaal. Inorder to characterize themigration Q/ themethane and water intoand around the minedcoal seam, a laboratoryinvestigation Q/ thepenneability Q/ the roofstrata was undertaken.The aim Q/the investi-gation was to correlatechanges in thefadeswith changes in penne-ability to enable theprediction Q/possiblezones Q/pressure build-up.

The tests werecarried out on samples

from borehole core thathad been obtainedfromthe roQ/ strata as partQ/ the mine explorationprogramme. Penneabilitymeasurements werethen conducted underatmospheric andtrlaxial-stressconditions. For the testsunder atmosphericconditions, a modifiedOhle cell was used and,

for the tests under

. Departmentif Geology

and Applied Geology,University if Natal,

Durban.

<9 The South 4fricanInstitute if Mining andMetallurgy, 1994. SA

ISSN oo38-223X/3.oo

+ 0.00. Paper receivedMf{}'. 1994.

The measurement of gas permeability insediments of the Vryheid Formationby B.J. Venter* and C.A. Jermy*

Introduction

A coal mine in the eastern Transvaal experiencedroof falls as a result of the excessive build-up ofmethane and water pressure in the roof strata ofa mined coal seam. A laboratory-based perme-ability study was therefore initiated to charac-terize the migration of methane and water intothe mine opening. Samples were obtained fromexploration borehole core that had been drilledfrom the roof strata. The permeability of thesamples was tested under atmospheric andtriaxial conditions, with methane as thepermeating fluid, and the liquid equivalentpermeability for each sample was determined byASTM methods!. The liquid equivalent perme-abilities were then used as a basis for comparisonbetween the different types of facies. af thetwenty-four facies in the roof strata2, only elevenwere sampled for the tests in this project. For thetests under atmospheric conditions, use wasmade of a modified ahle cell3,and a modifiedHoek cell4was used for the triaxial tests.

Preparation of the Samples

The core samples were cut into appropriatelengths as required by the different apparatus.The ahle cell required sample lengths of 12 to20 mm irrespective of their diameter. The triaxialtests, to reduce 'end' effects, required samplesconforming to the 2:1Iength-to-diameter ratioproposed by abert and DuvalJ5.As a result, thesamples for these tests were cut into lengths of100 or 120 mm, depending on the diameter ofthe core, TNW (54 mm) or NX (60,3 mm). Theends of the samples were ground with carborun-dum paste until they were parallel and smooth.This was done so that an even load could beapplied perpendicular to the long axis of thesample in the triaxial case, and to assure a betterseal between the a-rings and the sample whentested in the ahle cell. The samples were washedso that their pores could be free of grit and paste,which would affect the permeabilitymeasurements. The samples were then oven-dried at 100°C to remove all the excess moisture,and were placed in desiccators containing silicagel to keep them free of moisture. Drying over aperiod of 12 hours was found to be sufficient inmost cases for the samples to attain a constantweight, Le. to be assumed to be moisture-free.

The Joumal of The South African Institute of Mining and Metallurgy

Methodology

Tests with the Ohle Cell

The ahle cell as described by ahle6 and byChakrabarti and Taylor7 was modified mainly inrelation to the material used for its constructionand the sample holder. Brass replaced the stain-less steel, because it does not react with any ofthe fluids used (e.g. it does not rust in tests withwater) and is strong enough to withstand thefluid pressures used. The sample holder wasmodified to allow it to accept a wider range ofsamples with respect to diameter and length. Theseal between the sample and the sample holder isachieved by two rubber a-rings that are pressedagainst the sample by means of a spacer.

The gas (either nitrogen or methane) wassupplied from a commercial gas cylinder thatwas capable of delivering gas at a pressure of20 MPa. The pressure was kept constant bymeans of a regulator, and was measured with atransducer with a digital readout. The flowrateswere measured with a stopwatch and a series ofbubble flow meters of 25, 100, and 250 cm3capacity. Figure 1 shows the different compo-nents of the ahle cell for the tests involving gasas the permeating fluid.

After the sample had been installed in thechamber of the sample holder, the apparatus wasassembled, and was sealed with an a-ring com-pressed against the sample. The effectiveness ofthe seal was tested by the application of pressureto a steel disc of the same dimensions as thesample inserted in the cell; fluid reaching thebubble flow meter would indicate a defectiveseal. Once a porous sample had been inserted, apressure of 150 kPa was applied, and the samplewas left to reach equilibrium. Equilibrium isassumed when the measured rate of flowbecomes constant, implying that the sample isfully saturated with the testing fluid. Theflowrate was then recorded and the gas pressureincreased by 50 kPa. The system was again leftto reach equilibrium and the flow recorded. Thisincremental increase in gas pressure wascontinued until the pressure reached 700 kPa.The permeability at each increment was thencalculated from equation [1], which is aderivative of Darcy's Law:

OCTOBER 1994 295 ...

The measurement of gas permeability in sediments of the Vryheid formation

GAS

DIGITAL READ-OUT

SOAP 0 BUBBLE

FLOW METER

OHLECEI..L

SAMPLE

Figure 1- The different components of the Ohle cell used in the tests

triaxial-stressconditions, a modifiedHoek cell was used. Theliquid equivalent perme-abilities qfthe differentsamples were obtainedby use qf ASTM D4525-85, and these were thenused as a basis forcomparison.

Fades type, i.e.grain size, wasJound tohave an iTJlluence on thepermeability but, owingto the variable nature qfthe sediments, a rangeqfpermeability ratherthan a specffic perme-ability value wasassodated with eachtype qfJades. Themeasured permeabilitywasJound to decreasewith an increase in gaspressure. The liquidequivalent permeabilitywasJound to increasewith an increase ingrain sizeJor thecoarser-grainedJades,and to decrease with anincrease in grain sizefortheflner-grained types.The application qf anexternal stress regime tothe sample wasJound toreduce the permeability,but the amount qf thereduction depended onthe type qf stress. Thepermeability was also

found to depend on therecent stress history qfthe sample. The authorsconclude that the

~ 296

2QP J1Lk-

0-

(~2 - ~2)A '

where k = permeability in the direction offlow (Darcy)

Q = volumetric rate of flow at inletpressure (cm3/s)

Po = reference pressure (atm)Jl = viscosityof the fluid (cP)L = length over which the flow is

measured (cm)PI = inlet gas pressure (atm)P2 = outlet gas pressure (atm)

A = cross-sectional area acrosswhich Q is determined (cm2).

The reference pressure, Po, and the outletpressure, P2, are usually taken as being equal toeach other and to the atmospheric pressure,which is assumed to be 1 atm. A gas pressure of700 kPa was chosen as the termination pointbecause the initial tests resulted in erratic plotsof permeability versus reciprocal mean gaspressure after an inlet pressure of 400 kPa hadbeen reached (Figure2). This behaviourwasthought to be caused by either the Klinkenbergeffect9 or by turbulent flow so that Darcy's Lawwas no longer validlO.Subsequent experimentswere continued up to an inlet gas pressure of700 kPa to ensure that the cause could beidentified, as well as providing enough datapoints for the later statistical manipulationsrequired by the ASTM tests I.

[1]

OCTOBER 1994

Generally, for safety reasons, a sample wasfirst tested with nitrogen and then with methaneas the permeating fluid. If the seal were defective,there would be less danger from the use of nitro-gen than from methane. To change over from onegas to another, the gas pressure was reduced tozero and the pressure released from the systemwithout removing the sample from the cell. Theprocedure as outlined here was then repeatedwith the new gas.

Tests with the Hoek Cell

In the tests simulating in situ stress conditions, amodified Hoek-Franklin triaxial cell was used4.The modifications involved a series of intercon-nected circular grooves cut into the end platens.The grooves are designed to distribute the fluidover the whole cross-sectional area of the sampleat the high-pressure end, and to collect the fluidpassed through the sample over the whole cross-sectional area of the sample at the low-pressureend. In the case of coal, it was found that thesegrooves pressed into the ends of the samplethereby effectively blocking any gas from pene-trating the sample. Sintered discs of the samediameter as the platens were used to counteractthis.

An axial load was supplied by a hydraulictesting machine capable of applying a load of1000 kN.The confiningstress was suppliedby ahydraulic pump that can supply oil pressure upto 50 MPa. The gas was supplied and the flow-rate recorded by the same equipment as describedfor the tests with the Ohle cell.

Figure 3 shows the different components ofthe Hoek-cell tests involving methane as thepermeating fluid.

ColD3

>:0 _

5 ~Q ,0r-.as

~o§ 0,60

~

:>,.....

:s 0,55..casQ)

E:@0,50p..

III

~:400 kPa I

I0,45

0,1 0,2 0,3 0,4 0,5 0,61 / Mean gas pressure (l/atm)

Figure 2-Graph showing the irregular plot for

permeability versus reciprocal mean gas pressure

beyond 400 kPa

The Journal of The South Afncan Institute of Mining and Metallurgy

The measurement of gas permeability in sediments of the Vryheid formation

GAS

AXiAL LOAD

SOAp. BUBBLE

FLOW METER

HOEKCElL

PLATEN

HYDRAULIC SYSTEM ..J

Figure 3- The different components of the Hoek-cell tests

methods described areusifulfor the correlationqffades type withpermeability, but thatthe results are sensitiveto changes in locality.

References

1. AMERICAN SOCIElY FOR

TESTING AND MATERIALS.

Standard test methodjor

permeability if rocks byflowing air. ASTM

D4525-85, vol. 04.08.

1990. pp. 730-733.

2. WARD,J.R., and JERMY,

C.A. Geotechnical

propenies of South

African coal bearing

strata. Symposium: Rock

Mass Characterization.

Randburg, SANGORM,

1985. pp. 57-65.

3. VENTER,B.J. Assessment

of the permeability of

Vryheid Formation

sediments. Durban,

University of Natal, M.Se.

thesis, 1994. 107 pp.

4. DAW, G.P. A modified

Hoek-FrankIin triaxial

cell for rock permeability

measurements.

Geotechnique, vol. 21.1971. pp. 89-91.

5. OBERT,lo, and DUVALL,

W.!. Rock mechanics and

the design if structures

in rock. London, John

WiIey & Sons, Inc., 1967.

337 pp.

Before the sample was placed in theassembled Hoek cell, the annular space betweenthe rubber sleeve and the cell was filled withhydraulic oil. The sample was then placed in thecell, and a hydrostatic stress of 4 MPa wasapplied. (This was found to be the general stressat which no gas flowed between the sample andthe sealing rubber sleeve of any of the samples.)At the outset it was hoped to start the tests atthe measured in situ stress condition in the minebut, since this stress is estimated to be below4 MPa, this could not be done. Only tests usingmethane were carried out: firstly, because of theimportance of the methane tests to this studyand, secondly, because of the time involved inreaching equilibrium. A full range of tests on onesample could take up to four days to complete.The gas pressure was increased in 50 kPa incre-ments up to 700 kpa at a hydrostatic pressure of4 MPa, similar to the Ohle tests. From this point,a series of three tests was carried out.

Firstly, the axial stress was increased inintervals of 2 MPa (usually to 20 MPa) while theconfining stress was kept at 4 MPa and the gaspressure at 200 kPa. The flowrate was measuredafter each load interval. Axial load was thendecreased in 2 MPa steps until hydrostaticconditions at 4 MPa were again reached.

Secondly, the confining stress was increasedin 2 MPa intervals while the gas pressure waskept constant at 200 kPa. This loading was alsocontinued until a load of 20 MPa had beenattained, at which time the hydrostatic load wasdecreased in steps of 4 MPa until the base 4 MPahydrostatic conditions were again reached.

Thirdly, the hydrostatic stress was increasedin 2 MPa intervals while the gas pressure waskept constant at 200 kPa. As in the previousmethod, the stress was increased until 20 MPawas reached. The confining stress was thenrelieved until the starting 4 MPa hydrostaticconditions were again attained.

The Joumal of The South African Institute of Mining and Metallurgy

In each of the three cases the procedure wasrepeated two to three times. The effect of eachcycle was determined from measurements of thespecific permeability at a series of gas pressuresand from the liquid equivalent permeability ineach instance.

Results and Discussion

Ohle-cell Tests

When the permeability measured at each inlet-gas pressure was plotted against the reciprocalmean gas pressure (Pm), it was found that thepermeability decreased with increasing inlet-gaspressure. In the majority of cases, the nitrogenand methane plots for the sample followed thesame pattern (Figure 4). The plot for nitrogenwas found to be higher, with steeper slopes thanthe methane plots. This is as expected sincenitrogen has a smaller molecular size thanmethane. Nearly half of the projected liquidequivalent permeabilities, on the other hand, didnot follow this pattern. Theoretically, thedifference should be zero9, and Ohle6 concludedthat the difference can be taken as the experi-mental error. In the present study, the differenceseldom exceeded 10 per cent, and the sample forFigure 4 has a difference of 5 per cent betweenthe two kif values.

Owing to the extreme variability of the sedi-ments tested, even over short distances(centimetre scale) and in the same type of facies,the liquid equivalent values obtained for eachfacies type is not confined to a single value, butrather to a range of values. This is illustrated inTable I, which shows the ranges obtained fromthe tests involving methane as the permeatingfluid. The relationship between the permeabilityranges for the different types of facies is betterillustrated by Figure 5.

OCTOBER 1994 297 ...

The measurement of gas permeability in sediments of the Vryheid formation

0,35 Nitrogen(k1e = 0,1106 mD)

+

..-~() 0,30J.Icd

"t:I....--.s 0,25--

Methane(kle = 0,1161 mD)

~....:::: 0,20..0cdQ)

SJ.IQ) 0,15

p..x

0,100,0 0,2 0.4 0,6 0,8

1 / Mean gas pressure (l/atm)

Figure 4-Plot of It" versus 1/P m for methane and nitrogen for the same sample

TableI

Ranges of liquid equivalent permeability for the facies testedusing methane as the permeating fluid

Averager2

0,94

0,77

0,81

0,78

0,81

0,92

0,80

0,69

0,80

0,69

6. OHLE, E.L. The influence

of permeability on ore

distribution in limestone

and dolomite: Part I.

Economicgeology, andthe society if economic

geologists, vo!. 46, no. 7.

Nov. 1951. pp. 667-705,

The main objective of this study was todetermine whether any correlation could befound between facies type and permeability. Asillustrated in Figure 5, a relationship, althoughnot a straightforward one, exists between thefacies type and the permeability, In the coarser-grained facies (type 8 and higher) there is anincrease of average liquid equivalent permeability

~ 298 OCTOBER 1994

with an increase in grain size. For the finer-grained facies (those below facies type 7), theaverage liquid equivalent permeability was foundto decrease with an increase in grain size. Thiscould be the result of many factors but, in theauthors' opinion, the factor with the most signi-ficant influence is that of the number of samplestested. The more tests done on a specific type offacies, the more statistically representative therange of liquid equivalent permeabilitiesobtained. The coarser the grain size, the largerthe range of liquid equivalent permeabilitiesbecause of the larger distribution in grain size ofthe facies type.

Another very significant factor that couldinfluence the quoted liquid equivalent perme-ability ranges is the choice of data points used inthe extrapolation of the liquid equivalent perme-ability, The ASTM1recommends that a straightline be drawn through at least three data pointsat the lower values of reciprocal mean gas pres-sure. This best-fit straight line is then extrapo-lated to they axis to give the liquid equivalentpermeability of the sample. The points chosenfor this best-fit straight line may influence thevalue obtained in the extrapolation and theeventual liquid equivalent permeability rangesobtained. In the present study, all twelve datapoints were used when the straight line wasfitted to give the liquid equivalent permeability.

In some cases (16 per cent of the testsinvolving nitrogen and 23 per cent of the testsinvolving methane), the plots of permeabilityversus reciprocal mean gas pressure hadnegative slopes for either the methane testsalone (Figures 6 and 7) or for both the methaneand the nitrogen tests (Figure 8). Such resultswere unexpected.

6,0 -,~

;;;0

'"os-c::::4,0

's~

t';:::::8~2,0

S'"Q)

Po.

l-0,0

0 2 4 6 8 10 12Facies type number

14

Figure 5-Plot showing the relationship between the klopermeability ranges and the different types of facies in

the tests involving methane as the permeating fluid

The Journal of The South African Institute of Mining and Metallurgy

The measurement of gas permeability in sediments of the Vryheid formation

0,0070 -::;-j1->.

Q

~0,0065'1:1....--Os->.°,0060+';.:::;Ba!QJ

E0,0055

QJIl.,

0,00500,0

Metha ne

0,1 0,1 0,2 0,21 / Mean gas pressure (l/atm)

Figure 6-Plot of k. versus 1/Pm, showing a negative slope for a test involving methane as

the permeating fluid

0,0095

~Q

'"a!'1:1;.:::0,0085-Os-t> 0,0075-°'"..ca!QJ

S~0,0065

Il.,

Nitrogen(Kle = 0,0059 mD)

0,00550,00 0,10 0,20 0,30 0,40

1 / Mean gas pressure (l/atm)

Figure 7-Plot of k. versus 1/Pm, showing a positive slope for a test involving nitrogen as

the permeating fluid for the same sample as in Figure 4

->.Q

~ 2,00

'1:1....--Os->.1,50+'....:;:::..c

a!QJ

E1,00

QJIl.,

2,50

~

0,500,1

Methane

0,2 0,3 0,4 0,5 0,6 0,71 / Mean gas pressure (l/atm)

Figure 8---Plot of k. versus 1/Pm, showing negative slopes for both the methane and the

nitrogen tests on the same sample

The Joumal of The South African Institute of Mining and Metallurgy

The two main phenomena that could beresponsible are the Klinkenberg effect-andmethane adsorption!!. Klinkenberg's method9 ofanalysis (or that of the ASTM!) removes theeffects of experimental error. If the methane plothas a negative slope but the nitrogen plot for thesame sample has a positive slope (Figures 6 and7), it can be assumed that adsorption is thedominant active phenomenon because methane,and not nitrogen. is reactive towards the samplestested. If both plots result in negative slopes(Figure 8), gas slippage or the Klinkenberg effectmust be the dominant mechanism present. In asmall number of instances. the extrapolatedliquid equivalent permeabilities were found tohave negative values, which is a physicalimpossibility. Scheidegger12 stated that Darcy'sLaw represents a straight line between thequantity of flow and the pressure differentialand. as such, should result in a line through theorigin. Experimental error could thus force theline to intersect they axis at a negative value.

Both the measured permeability (ks) and theextrapolatedliquidequivalentpermeability (kIe)were found to vary over very short distances.The plots in Figure 9 represent four samplestaken over a distance of 10 cm. They all belongto the same type of facies, but the grain sizeincreases visibly from the medium-grainedsample A to the coarser-grained sample D. It canbe seen in Figure 9 that ks and kle increase withincreasing grain size. This shows that there mustbe some relationship between grain size andpermeability. The distribution of the data pointsalso becomes less linear as the grain sizeincreases. and the slopes become less steep asthe grain size decreases. This means that the gashas a path of larger diameter to flow through,and some turbulent flow (resulting in a non-linear plot) can occur. Figure 9 also illustrateswhy a range of liquid equivalent permeabilities,rather than single values, were obtained for aspecific type of facies.

OCTOBER 1994 299 ....

The measurement of gas permeability in sediments of the Vryheid formation

2,00 Methane

~0

~ 1,50

't:f.............'s--->,1,00

.....

..............,Q(!jQ)

E0,50

Q)

i:l..

B (0,520 mD)

A (0,345 mD)

0,000,10 0,20 0,30 0,40 0,50 ')60 0,70

1 / Mean gas pressure \l/atm)

Figure 9-Plot of It" versus 1/Pm, showing the variation of permeability over a 10 cm

interval. k1e is shown in brackets

7. CHAKRABARTI,A.K., andTAYLOR, R.K. The porosityand permeability of the

Zawar dolomites.lnt. /.

Rock Mech. Min. Sci.,

voI. 5.1968. pp.261-273.

8. STORMONT,I.C., and

DAEMEN,I.I.K. Laboratory

study of gas permeability

changes in rocksaIt

during deformation.

/. Into Rock. Mech. Min.

Sci. & Geomech. Abstr.,

voI. 29, no. 4.1992. pp.325-342.

Hoek-cell Tests

The time needed for the tests involving methane(or nitrogen) as the permeating fluid to reachequilibrium was approximately 1 hour in thecase of the Ohle-cell tests. When a hydrostaticstress of 4 MPa (Le. °1 = °2 = °3) was applied in

the Hoek-cell tests, equilibrium was reachedonly after about 5 hours. The higher the fadesnumber, Le. the coarser the grain size, theshorter the time needed to reach equilibrium. Inorder to obtain a basis for comparison with thetests under atmospheric conditions (Ohle cell), aseries of permeability measurements atincreasing gas pressures were made at differenthydrostatic loads. Table II shows the result at

Table11Liquid equivalent permeability ranges for the facies tested withmethane as permeating fluid under a hydrostatic pressure of4MPa

Averager2

0,92

0,72

0,87

0,69

0,99

0,57

0,67

~ 300 OCTOBER 1994

4 MPa. Of the tests, 16 per cent producedplotsindicating a deviation from Darcy's Law. Thesamples in these tests all contained some carbon,indicating that adsorption of methane was thereason for the anomalies.

It was found that an increase in any of theapplied stress regimes (axial stress, °1; confiningstress, °3; hydrostatic stress, °1 = °3) leads to areduction in permeability. An increase inhydrostatic stress has the largest influence, withconfining stress the second largest and axialstress the smallest.Figure 10 shows the influenceof a continued increase in confining stress on thepermeability. This decrease can be attributed tothe closing of the pores under external pressure.The initial decrease, Le. from 4 to 8 MPa, wasfound to be the largest in all cases. In somecases, this decrease in permeability was as highas 50 per cent. Subsequent increases producedprogressively smaller and smaller decreases inpermeability, as can be seen in Figure 10.

Gawugall found that, at some point, there isno further decrease in permeability irrespectiveof how much the load is increased. The authorscould not duplicate this result, but the samplesin this study were not subjected to the samenumber of load cycles as in Gawuga's tests. Thedecrease was, however, not permanent, and theinitial permeability could again be attained afterthe load had been relieved for a few days. Thismeans that the permeability depends on thestress history of the sample, at least for a while.

Axial Stress = 4 MPa0,70 ~

~O,60

>.<:)kaj

a0,50

],0,40

*

~27 mD)

8 MPa (0,2070 mD)

~12 MPa (0,1140 mD)

~'--'

- -

t';:::: 0,30;c;et!Q)

S0,20

kQ)

IJ.. 0,1016 MPa (0,0886 mD)

0,000,1 02 0,3 0,4 0,5 0,6

1 7 Mean gas pressure (l/atrn)

Figure 1o-Plot of It" versus 1/Pm' showing the decrease inpermeability with an increase in confining stress from4 to 16 MPa for the same sample. k.. is shown in brackets

The Joumal of The South African Institute of Mining and Metallurgy

The measurement of gas permeability in sediments of the Vryheid formation

9. KLINKENBERG,L.J. The

permeability of porous

media to liquids and

gases. Drilling and

production practice.

Dallas, 1941. pp.

200-213.10. DESAI,C.S. Finite element

methods for flow in

porous media. Finite

elements injluids-

Volume 1: Viscousjlow

and hydrodynamics.

Galagher, R.H.,

Oden, J.T., Taylor, C., and

Zienkiewicz, O.C. (eds.).

New York, John Wiley

and Sons, 1975. pp.

157-182.11. GAWUGA,J.K. Flow of gas

through stressed

carboniferous strata,

Parts 1 and 2. University

of Nottingham, Ph.D.

thesis, 1979. 362 pp.

12. SCHEIDEGGER,A.E. The

physics qfjlow in porous

media. 3rd edn.

University of Toronto

Press, 1974.353 pp.

Conclusions

It can be concluded that permeability is a physicalproperty of rock that varies over short distancesand is dependent on many internal and externalfactors. The permeability of a sediment for a gassuch as methane or nitrogen decreases with anincrease in gas pressure. If carbon in one form oranother is present in a sample being tested, itmay adsorb methane and yield anomalous plotsof permeability versus reciprocal mean gaspressure. This mechanism can, however, beshown to be active rather than due to gas slip-page (the Klinkenberg effect) when the sample istested with both nitrogen and methane, and theresulting plots of permeability versus reciprocalmean gas pressure are compared. Application ofan external stress results in a decrease in specificand liquid equivalent permeability. The amountby which the permeability decreases is controlledby the type of external stress applied. The largestdecrease occurs when the applied stress is hydro-static, and the smallest decrease is obtained whenthe applied stress is axial, with an intermediatedecrease for confining stress. Permeability isdependent on the stress history of the sample,but the influence of the stress history decreaseswith each application of a load cycle. The defini-tion of the lower and upper limits for each rangewere found to depend on the number of testscarried out on a particular type of facies. None-theless, a relationship of increasing permeabilitywith increasing grain size was found in thecoarser-grain facies (types 8 and higher). For the

finer-grained facies, the permeability was foundto decrease with an increase in grain size. This isnot detrimental if the results are used as an indi-cator of a roof-fall hazard or to characterize theflow character for the extraction of coal-bedmethane. A graph of permeability versus faciestype can be constructed as shown in Figure 4.When a well is to be sunk for the extraction ofcoal-bed methane or holes are to be drilled inadvance of the mining face, one can obtain anidea of the permeability, and therefore the flowcharacter, by identifying the facies type andreading the permeability from the graph.

In general, it can be concluded that themethods discussed can be useful in charac-terizing the flow in one specific location, butmore work is needed to more fully define therelationship between facies type and permeability,especially with reference to the influence ofmacro-structures such as joints, fractures, anddiscontinuities.A more detailedmicroscopicstudy might delineate the borders between thedifferent types of facies more conclusively, andso enable a better relationship between faciestype and permeability to be determined.

Acknowledgements

The authors thank Rand Coal(Pty)Ltd formaking the coreavailableand the UniversityofNatal,Durban, for financialsupport throughoutthe project. .

The Journal of The South African Institute of Mining and Metallurgy OCTOBER 1994 301 <Ill