HAL Id: ineris-00971968https://hal-ineris.archives-ouvertes.fr/ineris-00971968

Submitted on 3 Apr 2014

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Distinct element modelling and mining inducedsubsidence: Influence of the major faults

F. Homand-Etienne, I. Mamane, Mountaka Souley, P. Gaviglio, Marwan AlHeib

To cite this version:F. Homand-Etienne, I. Mamane, Mountaka Souley, P. Gaviglio, Marwan Al Heib. Distinct elementmodelling and mining induced subsidence: Influence of the major faults. International symposium ofthe international society for rock mechanics (EUROCK 1996), Sep 1996, Turin, Italy. pp.1305-1312.�ineris-00971968�

96-42

Distinct element modelling and mining induced subsidence: Influence of the major faultsModelisation des affaissements miniers : Influence des faules

Die verschiedenen Elemente Methode und Bergbaubodensenkung : Einfluss der Verwerfungen

F. Homand-Etienne, I. Mamane, M. SouleyLaboratoire de Geomecanique, ENSG, Vandoeuvre-les-Nancy, France

P. GaviglioLaboratoire de Geologie Structurale et Appliquee, Universite de Besancon, France

M. AI HeibLaboratoire de Mecanique des Terrains - INERIS, Ecole des Mines, Nancy, France

ABSTRACT : This study examines the subsidence due to Underground mining works in a coal mine in France.Coal is mined at a depth reaching 1000 m, according to longwall face method with caving. The seam rangesbetween 2 and 3.4 m in thickness. The dip strata does not exceed 10°. As predicted by the empirical curves, thenumerical and measured maximum subsidence does not exceed 70% of seam thickness. Numerical modellingand empirical methods overestimate the subsidence for extracted longwall width greater than 700 m.

RESUME : Cette etude des affaissements induits par l'exploitation de charbon a ete faite pour une mine situee äune profondeur de 1000 m en France. La puissance de la couche exploitee par longue taille varie de 2 ä 3,4 m etle pendage des strates ne depasse pas 10°. Comme le predisent les courbes empiriques, l'affaissement maximalmesure ou caicule numeriquement n'excede pas les 70% de l'ouverture. Pour des largeurs exploiteessuperieures ä 700 m, les valeurs mesurees sont nettement inferieures ä celles caiculees.

ZUSAMMENFASSUNG : Diese Untersuchung der durch Kohlegewinnung induzierten Bodensenkung ist indem Rahm eine 1000 m tief französische Zeche durchgeführt. Die Mächtigkeit des durch Strebenmethodabgebauten Flöz variiert zwischen 2 und 3,4 m. Die Längsneigung der Schichten ist nicht grösser als 10°. Wievon empirischen Kurven vorausgesehen überschreitet die abgemessene oder berechnete Höchstsenkung 70%der Mächtigkeit nicht. Für über 700 m abgebauten Breiten sind die abgemessenen Werte entschiedend geringerals die berechneten Werte.

l INTRODUCTION

Knowledge of subsidence engineering is ofconsiderable importance to the planning anddevelopment of surface. Subsidence is generallyassociated with the extraction of minerals and naturalresources such äs oil, gas and water. It is vital forpopulation to be aware of the risk of possible surfacecollapses above the Underground openings. In suchsituations, knowledge should be established.Prediction methods exist for determining the likelymagnitude and extent of subsidence development atthe surface. These methods include empirical,numerical and analytical approaehes. Empiricalmethods were essentially based on experience andfield measurements. There are intended to similarother sites. Also, surface subsidence has beenpredicted using a number of numerical modeis basedpredominately on simplified elastic and Isotropieassumptions. Many of these methods appear to makeassumptions that are invalid and, consequently, giveresults which oversimplify subsidence prediction.

The main objective of this study is to compareanalytical, empirical and numerical methods for a

specific site (Provence Colliery) in which thesubsidence data were available, with intent to validthe methods of modelling and to evaluate theirestimated character in relation to new sites. Empiricalor analytical methods can not take into account somelocal phenomena well known asßight ofstairs nearfaults. The numerical modelling, especially the onebased on the Distinct Element concepts, allows forthe presence of discontinuities, therefore permits todisplay prominently the relative movement ofdiscontinuities and to analyse more accurately thismovement. The special feature of the numericalmodelling presented here is that we have taken intoaccount the tectonic history affecting the site duringthe loading sequences. Numerical analyses describedin this paper were conducted using the two-dimensional UDEC code.

2 GEOLOGICAL CHARACTERISTICS OP BASIN

The coalfield under investigation belongs to theGardanne basin, which forms the eastem part of theAre basin, and is an E-W oriented geological unitlocated 10 km norm of Marseiile.

EUROCK'96, Turin , 2-5 septembre 1996, p. 1305-1312

NE

The geological structure is simple : the asymmetricsyncline is strongly affected by a dome-shapedanticiine (Fig. l). In the presently mined area, thestructure is overridden by a major thrust sheetoverthrusting Northward (Etoile unit), due to theNorth-South shortening and the major tectonic eventof the geological history of the region. The sameepisode is responsible for strike-slip faulting all overthe area (Gaviglio et al. 1990).

The lignite seams belong to the Fuvelian (upper partof the Campanian), sequence which is made up oflimestone. This occurrence of coal amid calcareousbeds is major geological particularity of the coalfield,and is responsible for the great difference inmechanical stiffness between the coal seam and thecalcareous beds forming the floor and roof. TheFuvelian limestones are 250 m thick in the miningzone.

SWEtoile overthmsl y^ fgull

E2 Eocenc Q Valdo Fuvelian Bl AptianAlbian0 Rognacian (0 Santonian S JurassicE3 Begudian (33 Urgonian "IB Trias

Fig. l-Geological cross section (Gaviglio etal. 1990).

The investigated zone lies in the autochthonous partof basin, just in front of the tectonic thrust wedge andbounded by the Diote fault and the Safre fault.Numerous detailed geological studies have beenconducted in this region, but little attention has beenpaid to the fracturing pattern. However, due toUnderground workings in the lignite field, detailedstructural data are now available. To illustrate this, ahighest density of faulting lies southward in thevicinity of the Safre fault : this zone has directiyundergone the compressive load exerted by the thrustsheet. Also, the geological history is responsible forthe present state of stress, the Variation of thephysical and mechanical properties of the calcareousbed, and the geometrical arrangement of the majorfaults planes. The principal fault planes may have aninfluence on the behaviour of the strata, through theirgeometrical characteristics, especially if their effectscombine (Gaviglio et al. 1990). This is also,confirmed in a recent study reported by Mamane(1995).

Above the Grande Mine, the overburden is made upof Fuvelian, Begudian, Rognacian and Eocene,below the Fuvelian lie Cretaceous and Jurassic. Thecoal seam is 1000 m deep and dips East-West at anangle of 8°. The thickness of worked seam variesbetween 2 and 3 m. Above and below the coal seam,the rock is stratified and consists of very tough bedsof limestone.

3 EMPIRICAL PREDICTIONMETHODS

In these methods, the maximum subsidence (Wm) iscaiculated based on the extracted seam height (0), thegoaf treatment coefficient (q), the nature of coverrocks and depth ( H ) .

Conceming the influence of longwall width on themaximum subsidence, it seems that the model whichis successfully utilised is based on the use of anempirical curve relating the depth ( H ) to the longwallwidth (L), äs follows :

W. L_H

m (3.1)0

Fig. 2 shows the maximum subsidence expressed interms of extracted seam height and results have beenplotted against the raüos of the extraction width overdepth below surface (L/H). Fig. 2 represents also theProvence Colliery empirical curves. Note that thesecurves are qualitatively similar to those obtained inother countries äs reported by Whittaker & Reddish(1993). However, observations of mining subsidencereported in several countries have shown that themaximum subsidence above the longwall extractioncan be appreciably influenced by the nature of thecover rocks. Another feature illustrated in this figureis : the form "S" of the empirical curve whichcorresponds to three distinct parts (Arcamone 1980):

• The starting of "S" corresponds to the first effectsof subsidence. The maximum subsidence Wm doesnot exceed 10% of the extracted seam height (0), forwidths extraction involving L/H ratios which liewithin of 0-0.4. Subsidence is essentially related tothe settling of cover rocks. This part of curverepresents a highiy effective control parameter of themagnitude of surface subsidence.

• The linear part of curve corresponds to the actualsetting of subsidence due to the collapses of coverrocks. In Provence Colliery, Wm varies between 10and 60% of extracted seam height H, for L/H within0.4-1.2.

• The third part of "S" corresponds to the finalequilibrium. An important feature of this part (inProvence case) is that, for L/H greater man 1.2 ; thevalue of Wm/0 tends towards a constant of 0.6-0.7.In others words, the maximum subsidence does notexceed 70% of extracted seam height in ProvenceColliery case.

In order to predict the shape of subsidence profilestrough for mining dimensions, many authors haveproposed some form of equation or graph. Equationsmay be simple or complex, but generally based onresults obtained from case studies. These profiles areall of the form:

W(x) = Wmßr. x. H, C) (3.2)

- 2 -

where Wm is the subsidence at the centre of trough, yis the critical angle, H is the mining depth, x is thehorizontal distance from the centre of extractedlongwall and C is the profile fünction constant whichis adjusted until a best fit Image with the measuredprofile is obtained.

It is clearly established that geological and hydro-geological factors play a major role on the nature andmagnitude of subsidence induced by mining. Also,the empirical methods used for subsidence predictionare unable to take into account the effect of faults onthe ground surface movements.

.5 l 1.5 2 2.5Width/depth ratio (L/H)

Fig. 2 - Empirical relationship between the magnitudeof subsidence expressed in terms of extracted seamheight and longwall width/depth ratio for coal miningoperations at the Provence Co//;ery(Arcamone 1980).

4 NUMERICALMODELLING

4.1 Models and loading sequence

As reported in § 2, the area is constituted of Jurassic,Fuvelian, Begudian and Rognacian. In the minedarea, the major tectonic events are at the root ofvariety of fractures, and an accurate study performedby Gaviglio (1985) was permitted to define the mainfracture trends. In particular, three major reversefaults : Diote, Etoile and Safre form the structure atthe southem part.

The two-dimensional geometrical model used tosimulate a cross section that strikes North-South, isbased on the geological cross section ofthe basin andillustrated in Fig. 3, with respect to the average coalseam dip. The area of interest for the modelling of the"Grande Mine" site is about 4 km long and the modeldepth chosen is 1.2 km. The deposit is 950 m deep,the height of coal seam is 3 m and the dip strata is 8°.Mining is simulated by instantaneous excavation of100 m of coal at every step and the excavation of theentire longwall cross section corresponds to 1000 m.

In situ stress measurements were carried out in themine workings. These investigations havederiioftsträted thäfthe major principal stress is alwayshorizontal, and is much higher (2 to 3 times) than theminor principal stress (03), whereas the intermediatestress is almost vertical fo'2=0v}. Also, the verticalstress is, in any case, smaller than expected from theweight of overburden (0.53 to 0.87 times). In otherrespects, a recent study reported by Mamane (1995)showed that the in situ stress state results from thecombination of the gravity and the participation oftectonic stresses. For these reasons, we performedtwo modeis : Model A without tectonic stressconsideration and which represents the present-dayconditions, and Model B involving both gravityloading and a North-South tectonic compression tothe value of 5 MPa (Cs=5MPa).

The modelling sequences are performed äs follows :ßrstly, both modeis (A and B) without excavationwere consolidated under in situ stresses (Fig. 4b),secondly, for model B a compressive tectonic stressis applied äs illustrated in Fig. 4b, and thirdly, theexcavation was carried out (roller boundaries wereapplied to the sides and the base of the both modeis)and modeis were cycled to an equilibrium state forstress redistribution. Notice that the boundaryconditions used during tectonic loading relate thegeological history of the region äs reported byGaviglio (1985).

The rock mass is modelled using the fülly deformableblocks assumption. Rock blocks are assumed to beIsotropie and linearly elastic materials. In fact,different materials in Provence Colliery behaveelastically (Gaviglio 1987), even if materialsoverburden the Begudian-Rognacian at the surfaceare constituted to clay materials which can behave inan elastic-plastic manner.

4.2 Choice of materials andjoints characteristics

Concerning the mechanical properties of rockmaterials, Provence site is constituted of three majorsets of materials, though in the lithologic point ofview, the basin is formed of five groups (Jurassic,Fuvelian, Begudian, Rognacian and Eocene).Material number of these sets are illustrated in Tablel. Mechanical characteristics of these materials alsoreported in Table l, are derived from laboratory tests.

Table l - Elastic properties ofrock materials.

ROCK MATERIALSBEGUDIANFUVELIANJURASSICCOAL

v.25.25.25.30

E (MPa)300024000370002000

- 3 -

250mFig. 3 - Geometrical model based on the geological cross section of the basin.

_ 90 o MPfl ^ Constrained displacement

Fig. 4 - Loading and boundary conditions : (a)-models loading using gravity, initial m situ Stresses andboundary Stresses (b)-tectonic loading and boundary conditions during excavation.

Unlike the rock mass properties, it was not possibleto deduct in advance the mechanical properties ofjoints (faults and joints between materials).Therefore, the choice of these Parameters presentedhere is based on some approaches encountered in theliterature. The normal and shear stiffness kn and ksduring the consolidation were caiculated based on arecent study reported by Kulatilake et al. (1992). Inthis approach, kn and ks depend essentially on theelastic properties of rock materials. Computed valuesare reported in Table 2. Conceming kn and ks neededfor joints during tectonic compression andexcavation, it is clearly established that themechanical behaviour ofjoint is generally non-linearand depend on large parameters such äs : stress level,dilatancy, weathering state of joints, joint wallgeometry, strength and deformability of asperities.Based on the relationships suggested by Bandis et al.(1983) and assuming that all joints have no dilatancyand their compressive strength (JCS) is equal to theuniaxial compressive strength of rock materials, theinitial normal stiffness, the maximum normal closureand the shear stiffness for each discontinuity werecaiculated. Therefore it is possible to derive thenormal stiffness (kn) according to the normal stresslevel and the hyperbolic model of Bandis et al.Finally, joints parameters used during tectonicloading and excavation are also, summarised in Table2.We assume that normal and shear behaviour of jointsare linear and the shear stress is limited by Mohr-Coulomb's criterion.

Table 2 - Mechanical properties of joints.

Jointmaterials

l23

* **

ConscV188

2625200

DU• lidation

V75

105080

^RINGEx<„ ,

Kn3000250002500

c:

7.51008

avation. **^s ^(°)

202220

GPa/m; MPa/m; <(> is joint friction angle.1- Begudian/Fuvelian, Begudicm/Begudian, Begudian/Jurassic2- Fwelian/Fuvelian, Fuvelian/Jurassic3- Coal/Fuvelian, CoaVCoal, CoaVJurassic.

4.3 Results

We then analysed the mining induced movements ofground surface with respect to extracted longwallwidth according to model A which represents thepresent-day conditions. A qualitative and quantitativecomparison between modeis A and B was alsoperformed.Fig. 5 shows the magnitude of subsidence expressedin terms of height (Wm/O) and the longwallwidth/depth ratio (L/H), compared to pre-establishedempirical curves. This indicates that values of Wm/0ratio have to fall between the lower and upperempirical curves, except for large values of extractedlongwall width (L>800 m). In spite of the fact thatthe model overestimates (much less than 7% ofextracted seam height: 0) values of Wm for L>SOO,it is interesting to note that the maximum magnitudeof subsidence does not exceed 70% of extracted seam

- 4 -

height; this is agreed with field observations. Also,Fig. 5 illustrates that the caiculated values of Wm arevirtually identical with those obtairied accoi-dmg toprobable curve for L^400, whereas differencebetween simulated and empirical curves does notexceed 8% of coal seam height, for extracted widthranged between 500-700 m. Consequently, based onthese observations, we note a good agreementbetween simulated and empirical curves.

Fig. 6 shows curves of ground surface subsidenceW(x), with respect to L. These subsidence profileslead to the following comments : ( i ) - W(x) increasewith the extracted width and the point of maximunisubsidence (i.e. M in Fig. 6) moves from north tosouth when the face advances. However, äsillustrated in Fig. 6, this point is not located in themiddle of extracted zone (which represents thetheoretical point of maximum subsidence) due to theface advance, behaviour of discontinuities andinclination of coal seam. (u) - No significant effect ofDiote fault is observed at the surface when L<600compared to Etoile and Safre faults influence clearlymarked in Fig. 6.

.5 l 1.5 2 2.5Width/depth ratio (L/H)

Also, interpretation of results when L>700 showsthat while Diote fault increases the magnitudes ofsubsidence, Safre and Etoile faults tend on thecontrary, to reduce the subsidence at the northem sideof Safre fault. In other terms, Diote fault effects takeplace after 700 m of extracted longwall width andtend to reduce the Etoile and Safre faults effects. Thisis clearly marked for L=1000 in Fig. 6. In order toexplain this faults inter-influence with respect to theadvancing face, we analysed the behaviour of faultsaccording to extracted width.

Fig. 7 shows the direction of block movements andshear displacement through disconünuities at the roofcorresponding to L =2 00. Examination of thesedisplacements with respect to L has demonstratedwhen the extracted width increases, points A, B, Cand D reported in Fig. 7 tend to move from north tosouth, whereas the point E located along Diote faultmoves towards the excavation. In particular, whenL>700, all joints located at the roof and dipped about8° have a dextral movement. Consequently, dextralmovement occurs along Fuvelian-Begudian interface(2/1), whereas Fuvelian-Fuveliqn iomt (2/2)tends toopen due to an increase öFinfIexion of Fuvelian bedat the immediate roof with increase of extractedwidth. Also, the part of Diote fault overburden theimmediate roof Fuvelian has a dextral movement,while Etoile and S a f re faults have a normaldisplacement, but these both faults are reverse indepth. Effectively, behaviour of Safre and Etoilefaults has conditioned the evolution (virtually linear)of subsidence in the Begudian block delimited bythese two faults äs illustrated in Fig. 6.Note that L=700 represents the minimum extractedlongwall width provoking an entire collapse of roof,

Fig. 5 - Caiculated curve for subsidence withrespect to the advancing face compared to empiricalcurves.

Distance x (m)500 1000 1500 2000 2500 3000 3500 4000

Fig. 6 - Subsidence results; illustrating the influence ofthe extracted longwall width.

-5

whereas extraction of 600 m of coal seam induces amaximum roof-floor convergence about 1.9 m.

Evolution of surface horizontal displacement U(x)curves with respect to L is illustrated in Fig. 8. Asobserved earlier in subsidence analysis, magnitudesof horizontal displacements increase with extractedwidth and the point of maximum subsidence(intersection of displacement curve and horizontalaxis) move from north to south. Also, Fig. 8illustrates the great part of faults in groundmovements. Heaves of Diote, Safre and Etoile faultsexpress fault opening at the surface. It is interestingto note that contrary to Safre and Etoile effects on thesubsidence profiles (reducing) observed in Fig. 6 forL^700, Fig. 8 indicates an accumulation of Safre andEtoile effects on the horizontal displacements. Itshould be noted that constant horizontal displacementofJurassic block delimited by Safre and Etoile faultsclearly indicates that this block behave äs a rigidblock, therefore practically govemed by the faultsdeformabüity äs explained earlier.

Differential movements of the ground surfaceproduces ground strains, and damage of surfacestructures is generally correlated to the magnitude of

these strains. Examination of the horizontal andvertical strain curves (i.e. dü/dx and dW/dy) notpresented here, showed that simulated curve fbrmsare similar to those encountered in the literature. Inparticular, horizontal strain curves show two distinctparts : ( i ) compressive strain part, around the point ofmaximum subsidence, ( U ) tensile ground strains onboth sides of the previous compressive strain region.Maximum compressive strain (Emax) obtained at theexcavation of the entire longwall (L=:1000) is about1.69 mm/m which approaches the theoretical valuegiven by the empirical relation : k Wm/H (where k isconstant ranged between 0.8 and 1.5). Aboutmaximum slope of ground surface (vertical strain)and curvature obtained by differentiating thesubsidence profile, we got respectively ~mm/m and Kma^=2.5 10~6 m'1.

lmax~•

Then according to the classification of Baochen(1993) applied to surface structures specifically todetermine the extent of damage caused to the surfacestructures äs a result of ground deförmation, previousvalues of Emax, Tmax äüd Kmax lead to the followingInterpretation : structures at the surface are not verysensitive to surface movements and deförmation inthe case o! Provence Colliery.

S

Fig. 7 - Mechanism of slip through discontinuities overhanging the Grande Mine coal seam (L=200 m).

Fig. 8 - Horizontal displacement results; illustrating the influence ofthe extracted longwall width.

6-

Consideration of the existing N-S tectoniccompression (model B) at the end of modelconsolidation favours the development of shearmovement through discontinuities Överhanging thelongwall. Value of shear displacement along Diotefault decreases from 0.1 m towards the seam level tol mm near the surface. Contrary to model A, analysisof deformability according to model B showed thatfractures have dextral movement for extractedlongwall width less than 500 m. In other words, forL<500, ground surface movements are appreciablyinfluenced by the tectonic compression.

Table 3 summarises the values of Wm according tomodel with respect to extracted width, and indicatesthat N-S compression reduces the magnitudes ofsubsidence. Value of this reduction is about 60% forL<200m and tends towards a constant of 5-10%when L>600. In the later cases, shear displacementsalong discontinuities are qualitatively similar to thoseobserved according to model A (i.e. Fig. 7).However for extracted width L^200, the behaviourof model B is entirely govemed by the applied o,.This has been again confirmed by the examination ofdisplacement and subsidence profiles correspondingto L=100-200 m.

Table 3 - Comparison ofmaximum subsidence

Wm

Model AModel B

Model AModel B

(cm) ExtractedL=1001.880.7

L=60010093.2

L=2008.73.3

L=700131.5123.6

longwallL=30021.216

L=800154.5145.5

width L (m)L=400

4236.8

L=900171

161.5

L=50067.662

L=1000188.3179

Similarly to W m evolution according to model A, o,tends to reduce the subsidence magnitudes W(x) for agiven extracted width. Recall that Cs induces dextralmovement through faults, whereas mining workingsgenerate a sinistral movement of faults. It isinteresting to note that a combination of the twoprevious phenomena can generate another blockcinematic at the surface in certain cases.

In other respects, the existing of <7s develops a largehorizontal displacement magnitudes. Contrary tomodel A, these horizontal displacements canapproach the values of Wm- To illustrate this, valueof horizontal displacement of the right side of Safrefault according to modeis B and A are respectively,l .24 and 0.76 m. However, horizontal displacementcurves are qualitatively similar to those observed inFig. 8 particularly for L>500 m except in the regioninvolving Fuvelian at the left hand side of model (i.e.where <7$ ls applied).

Finally the existing N-S tectonic compression reducesthe magnitudes of W m and W(x), and increases

significantly the horizontal displacements U(x) withrespect to extracted longwall width.

5 DISCUSSION AND VALIDATIO N

5.7 Comparison with in situ measurements

The main comparisons between empirical andnumerical predicted and measured maximumsubsidence with respect to extracted width areillustrated in Fig. 9. Direct comparison betweenempirical curves and measured Wm leads to thefollowing observations : five of eight measurementswere located between the lower and upper empiricalcurves, whereas three measures have beenoverestimated by both predictions (empirical andnumeric).

It may be seen that numerical maximum subsidencepredicted and the measured data are in fülly doseagreement for three particular extracted longwallwidths. The corresponding longwall width/depthratios are 0.41, 0.43 and 0.65. Also, the measuredmaximum subsidence corresponding to extractedlongwall width/depth ratio higher than 0.71 remainslower than one which was caiculated with modeis Aand B. However, recall that the previous ratio (0.7)corresponds to the minimum extracted widthoccurring an entire collapse of longwall roof,therefore inducing an increase of Diote fault effects äseariier observed in Pigs 6 and 8.

Note that the induced subsidence and horizontaldisplacement at the location of Diote fault (i.e. relativesubsidence and horizontal displacement between theleft and right sides of fault) for L =1000 arerespectively about 0.06 and 0.13 m, and increase atthe point of maximum subsidence location(combination of Diote fault and excavation effects).

Reasons for these discrepancies between modeis andmeasurements for extracted width higher than 700 mare not clear, but possibly their are related to eithervariations of faults behaviour due to immediate roofcollapse, and continuum movement of materialstowards the longwall betöre the roof collapse processwhich depends on the extracted width. To illustratethis, an elasto-plastic analysis performed by Mamane(1995) showed that maximum subsidence is reduceddue to the roof collapse which occurs before 700 mof extracted longwall, therefore continuum movementof roof in present cases will be reduced due to theactual "artißcial" pillar.

Also, making Diote and Safre faults stiffer reducesthe maximum subsidence for L<800 and increases themaximum subsidence for L>900, because in thiscase, roof collapse occurs with respect to 900 m ofextracted width. In other words, maximumsubsidence will be correlated to the critical extractedwidth which provokes the roof collapse.

- 7 -

5.2 Comparison with the model ofBaochen (1993)

Analytical model of Baochen for subsidenceprediction is used to valid the numerical modelling.Validation was made for 1000 m of extractedlongwall width and we present the essential results.Parameters of subsidence profiles (maximumhorizontal displacement, maximum subsidence andcritical angle) needed for the theoretical model, havebeen derived from the empirical relationshipssuggested by Baochen (1995).Evolution of subsidence profile according toBaochen's theory has globally the same trend thanthat observed using modeis A or B (i.e. Fig. 6).Also, the part of analytical and numerical profilescorresponding to the right side of the longwall matchqualitatively and quantitatively. Note that empiricalrelationship of W m given by Baochen underestimatesthe maximum subsidence compared to simulations.Consequently, Baochen's model tends tounderestimate the magnitudes of subsidence,particularly over the longwall. However, both points(analytical and numerical) of maximum subsidencefall in the same ränge.

If we had used the value of Wm obtained from oursimulations in Baochen's model, results of thisvalidation will be certainly ameliorated.

upper curveprobable curvelower curveModel AModel Bin situ measures

Fig. 9 - Maximum subsidence : comparison withmeasured data.

6 CONCLUSION

Ground surface movements due to mining workingswere examined for a specific French 'Coal Mine.Results involve numerical, empirical and analyticalapproaches, and in situ measurements. Based on theanalysis reported here, the main conclusions can besummarised äs follows :• We note a good ägreement between simulated andempirical curves of maximum subsidence withrespect to extracted width. As predicted by theempirical curves, the numerical and measuredmaximum subsidence does not reach 70% of seamheight.• Comparison with the in situ measurements hasshown that five ofeight measures are located between

2 L/H

the lower and upper curves and consequently agreewith the values predicted by caiculations. However,significant discrepancies between numerical andempirical predictions on the one hand, and observedmeasurements on the other hand were noticed forextracted width greater than 700 m. Attempt of anexplanation is made.• The existing N-S tectonic compression reduces themagnitudes of subsidence and increases thehorizontal displacements.• Comparison between the numerical modelling andthe analytical model of Baochen for subsidenceprediction indicates that the curve forms matchqualitatively.

ACKNOWLEDGEMENTS - The authors wish toacknowledge the HBCM Unite d'Exploitation deProvence for the permission to use the subsidencedata.

REFERENCES

Arcamone, J. 1980. Methodologie d'etude desaffaissements miniers en exploitation totale etpartielle. These de Doc. INPL. 99p.

Bandis, S.C. Lumsden, A.C. & Barton, N.R. 1983.Fundamentals of rock joint deformation. Int. J.Rock Mech. Min. Sei & Geomech. Abstr. Vol.20, ?6, pp. 249-268.

Baochen, L. 1993. Ground surface movements dueto Underground excavation in the People'sRepublic of China, Compressive Rock Engng.Hudson (ed.). Vol. 4, pp.781-817.

Gaviglio, P. 1985. A fault and stress field analysis ina coal mine (Gardanne, Bouches du Rhone,France). 113: 349-366, Tectonophysics.

Gaviglio, P. 1987. Consequences de l'evolutionstructurale sur les caracteristiques physiques descalcaires füveliens du bassin de l'Arc (Bouches duRhone, France): fracturation, proprietesmecaniques, etats de contrainte. Revue deGeologie Mediterraneenne, XIV, 3: pp. 221-232.

Gaviglio, P. Revalor, R. Piguet, J.P. & Dejean, M.1990. Tectonic structures, strata properties androckburst occurrence in French coal mine.Rockburst and Seismicity in Mines, Fairhurst(ed.), 289-293. Rotterdam: Balkema.

Kulatilake, P.H.S.W. Ucpirti, H. Wang, S.Radberg, G. & Stephansson, 0. 1992. Use of thedistinct element method to perform stress analysisin rock with non-persistentjoints and to study theeffect of joint geometry parameters on strength anddeformability of rock masses. Rock Mech. andRock Engng. Vol. 25, ?4. pp. 253-284.

Mamane, I. 1995. Modelisation ä grande echelle del'influence de la fracturation sur les contraintes insitu et les affaissements miniers. These de Doc.INPL. 202p.

Whittaker, N.B. & Reddish, D.J. 1993. Subsidencebehavior of rocks structures, Compressive RockEngng. Hudson (ed.). Vol. 4, pp.751-780.

- 8 -