You are on page 1of 5

Structural analysis of mine pillars using finite element method -A case study

D.H.Y. Tang and S.S. Peng


Abstract - Three-dimensional finite element modeling ( 3 - 0 FEM) was performed to analyze the safety factors of stump pillars, 5.5 m x 5.5 m, under a cover of 63 m using the Modified Drucker-Prager Theory. The results showed that t h ~ were y safe. This was substantiated by underground observations. But the safety factors predicted by three commonly used pillar design formulae showed otherwise. This paper discusses the modeling details and analyzes the difference between 3 - 0 FEM method and traditional pillar design formulae. Introduction
The major objective of this study was to evaluate the pillaring plan for preventing pillar failure in a northern West Virginia coal mine. The study was carried out by both mine visits and finite element analysis. The mine visit consisted of underground investigation of the pillaring area and the collection of rock samples for rock property determination. The finite element method was employed to model the configurations of mine workings in the same pillaring sequence as that practiced underground. By applying appropriate failure criteria to the finite element results, the stability conditions of all members of underground structure and surface zones were evaluated. The results of finite element analyses confirmed the field observations while the traditional strength formulae were too conservative. The finite element results also indicated that the horizontal stress (confinement) in the coal pillars played a very important role in pillar stability.
Mine description and pillaring plan

SOLID PILLAR

PARTIPL PILLIRING

Fig. 1 - Plan view of mine pillaring layout at the study area.

Preliminary study of pillar stability

The mine was located in northern West Virginia and extracted coal from the Upper Freeport seam, which had an average thickness of 1.2 m. The seam depth ranged from 44 m to 87.5 m with the majority being about 63 m. The immediate roof was gray shale, which was overlain by sandstone. The floor rock was fireclay and shale. The room-and-pillar mining method was used to extract the coal. During the first mining (or development), both entry and crosscut were 5.5 m wide whereas pillars were square and at 21 m center. During the second mining (or pillaring), the 16 m x 16 m coal pillar was split into four stump pillars at the comers, the size of which was 5.5 m x 5.5 m. In other words, a 5-m wide cut was made along the center of the pillar in two perpendicular directions. In order to protect the residential houses and water wells on the surface, a solid coal pillar, 36.5 m x 36.5 m, with each residential house at its center was left unmined. Also, one or two rows of coal pillars surrounding the four sides of the large solid pillar were left unsplit. Fig. I shows the plan view of a typical mine pillaring layout with a big solid pillar at the center. The mine visits confirmed that in spite of its small size (5.5 m x 5.5 m), the stump pillars remained intact as the pillaring proceeded and that no ground control problems ever occurred.
MINING ENGINEERING

In order to estimate pillar stability in the pillaring area, a preliminary study was conducted by using the pillar strength formulae. The tributary area theory was used to calculate the average pillar stress whereas the pillar strength was determined by the laboratory compression tests on small coal samples that considered the size and shape effects of the pillars. There are many pillar strength formulae. In this study, only three strength formulae that are commonly used in US roomand-pillar mining are employed (Peng, 1986): Obert-Duvall Formula g = 0, (0.778 + 0 . m wlh) Holland Formula (1)

q , = 4 (wlh)

1R

Bieniawski Formula g = 0, (0.64 +0.36wlh) where ap is the in situ strength of coal pillar, 0, is the strength of a cubical pillar at the critical specimen size, w is pillar width, and h is pillar height.
D.H.Y. Tang, member SME, is design engineer with MorrisonKnudsen Inc., Boise, ID. S.S. Peng, member SME, is professor, department of mining engineering, West Virginia University, Morgantown, WV. SME preprint 87-81, SME-AIME Annual Meeting, Denver, CO, February 1987. Manuscript November 1986. Discussion of this paper must be submitted, in duplicate, prior to Nov. 30, 1988.
SEPTEMBER 1988 893

Based on the results of laboratory tests on coal samples and the above mentioned formulae, the safety factors of stump coal pillar were determined and listed in Table 1. Clearly, all the safety factors of coal stump pillars from three formulae were less than 1.O. In other words, the stump pillars were not safe (or stable), which were not supported by underground observations.
~~

where
1 J =-[(CT,-Or' 2 6
2 +(02-03)

(03 0,)

J, = 0, + o2+ o3= ox+ 5 + oz

Table 1

- Safety factors of coal stump pillars determined by


various strength formulae
Overburden depth, m 45.1 63.4 0.447 0.533 0.565 87.8 0.324 0.385 0.41 1

Investigator

Obert Holland Bieniawski Note: 1. 2.

0.630 0.750 0.800

The size of coal stump pillar is 5.5 m x 5.5 m and the width of entry is 5 m on one side and 5.5 m on the other. The overburden stress is calculated using an average unit weight of 2.6 t1m3.

, , and 0 , are principal stresses, q, 5, and 4 are where a,.a normal stress inx-, y-, and z- direction, respectively, @ is the internal angle of friction, Co is the uniaxial compressive strength, q is the triaxial stress factor, and T is the tensile strength.

Results and discussions


Fig. 2 shows the average pillar stress from both tributary area theory and finite element analysis for stump pillars and barrier pillars. It can be seen that for stump pillar the vertical stress from the mbutary area theory is greater than that from the finite element analysis. But for the barrier pillar the result is reverse. This means that for stump pillars the average loading from the tributary area theory is generally more than the real situation, which, in turn, will result in either overdesign or prediction of failure and yet it is safe in reality. But if the immediate roof is made of very weak rock, the pillar stress determined from the finite element analysis may be greater than that determined from the tributary area theory. Fig. 3 shows the safety factorsof coal pillars along the cross section E-E' (Fig. 1). It is clear that safety factors of all coal pillars are greater than 1.O. The closer the stump pillars to the large pillar, the greater is the safety factor. It is apparent that the large pillar carries part of the load that is originally sustained by the stump pillars. Therefore, the finite element results confirm the underground observations. The tributary area theory, on the other hand, usually overestimates the average stressin the pillar. In addition to this, the traditional method does not consider the effects of the interaction of the roof, coal, and floor (i.e. the structural behavior of the roof and floor of different material properties cannot be considered). In order to find out the effects of the interaction of the roof and floor as well as the effects of the change of their properties, a series of finite eIement analyses was done by assuming various models of differentmaterial properties of the roof and floor. Fig. 4 shows the safety factors of coal pillars for four different models in which the coal strength for all four models

Finite element analysis


In order to clarify the pillar stability at the study area, the stability analysis using the finite element method is recommended. In finite element analysis, not only the structural behavior of all underground members, such as roof, coal, and floor are simulated, but the interaction among them and the mine layout are also taken into consideration. In this study, finite element analysis using NASTRAN computerprogram (McCormick, 1981)wasemployed. Threedimensional models with hexahedron elements were used throughout the analyses. The mechanical and physical properties of coal and rocks (Table 2) were obtained by laboratory tests on the samples secured from the mine. In order to account for the in situ rock mass conditions, a scale factor must be used for the strength values obtained in the laboratory. Based on numerous case studies conducted by the authors (Su and Peng, 1986; Hsiung andPeng, 1987)in which correlation of finite element modeling results and field conditions was available, a scale factor of 115 was adopted. All strata were assumed to be homogeneous, isotropic, and linearly elastic. In order to evaluate the stability conditions of all the members of the underground structures, the modified Drucker-Prager failure criterion (Drucker and Prager, 1952) was used in the analysis:

Table 2
. -

- Physical and mechanical properties of coal and rocks


Poisson's ratlo Uniaxial compressive strength (MPa) 78.13 100.80 74.22 5.60 34 67 Triaxiai* stress factor 4.0 4.8 4.0 3.3 3.8 Tensile strength (MPa) 3.12 3.56 2.98 2.26 3.09

Unlt Welght (llm3)

Malerlal

Young's modulus

Sandy shale Sandstone Gray shale (Immediate roof) Coal (Upper Ffeeport seam)

23718 38218 13086 2020 8101

0.22 0.16 0.18 0.30 0.20

2.60 2.69 2.71 1.33 2.57

* Data, belng unavailable, are taken from the representative values in the Handbook.
MINING ENGINEERING 894 SEPTEMBER 1988

Fireclaylshale (Immediate floor)

Overburdon bpth, 83.4 m

7 A

! ! a 6
a -

Vertical Strna Tributary Area Thory

-8-$-=*
8Flnite E l t

a -*
kulyslr

5-

5
P I

?
m
n

4-

3-

Ckl~altI s t r w 0-8-0-

+-0

f
a -

2-

-&lnStr~r&.el_mPe

-/
o+I

-- -

- - -- -

1-

P I

Fig. 2 - Pillar stress at the mid-height of

coal pillars determined from both tributary area theory and finite element

analysis (overburden depth is 63 m).

is the same. Model A is the original model for this case, Model B has a weak roof, Model C has the same material for all strata (i.e. homogeneous) while for Model D, the Young's modulus of coal is two times of its original value. It can be seen that four models have different safety factors, with Model C having the largest and Model B having the lowest one. Fig. 5 shows the vertical stress of coal pillars for Models A, B, and D. By comparing Fig. 5 and Fig. 4, it is found that although Model B has the smallest vertical pillar stress, it has the smallest safety factor. This indicates that the vertical pillar stress, which is usually used in the determination of the safety factor by the traditional method, is not the only controlling factor for the evaluation of pillar stability. There must be some other controlling factors involved in the structural behavior of the coal pillar. In reality, each coal pillar is not only subjected to the vertical stress, it is also subjected to the horizontal stresses in both x and y directions (i.e., it is triaxially confined by the

____ :I
1.675 1672

stresses acting on it). Thus, the horizontal stress confinement must play a very important role in pillar stability. As shown in Fig. 2, the horizontal stress, with its magnitude larger than one-third of the vertical stress, is acting on both x and y directions. As a result, the confinement of the coal pillar due to horizontal stress increases the coal pillar strength to some extent. Fig. 6 shows the pillar stress and safety factor of coal pillars for two different models. One model is the same as model A describedbefore while the other one is made of weak roof and floor with everything else the same. It can be seen that the vertical stress from the original model is larger than that of the weak one. But the safety factor of the coal pillar of the former is also greater than that of the latter. The main reason is that the horizontal stress confinement for the former case is also greater than that of the latter one. Therefore, due to horizontal confinement, the real coal pillar strength is usually greater than that determined from the strength formulae.

0:

1.669

> 1.658
UI a

1.663-

7-STRATA M O D E L 63 4 m . D E E P

1.660

30

40

50

60

7 0

D I S T A N C E F R O M EDGE O F LARGE S O L I D PILLAR. M

Fig. 3 M I N I N GE N G I N E E R I N G

- Safely factors for coal pillars (along cross section E-E').


S E P T E M B E R1 9 8 8 8 9 5

1.6L

A-ORIGINAL 0 -Q

* . .

1.wY u

+
f

1.621.601.58-

D- STIFF FOAL

n ' u

DISTANCE FROM EDGE OF LARGE SOLID PILLAR. M


- - - p p

Fig. 4 - Safety factor of coal pillars for four different models (along cross section E-E').

6.4 0-STIFF

COAL

6.2
P

"-0

Y)

6.0 -

5.8 5.6 -

5.4

B-WEAK ROOF

;
I

0
I

0
I

52-

0 0 30 40 50 60 70

DISTANCE FROM EDGE OF LARGE SOLID PILLAR, M

Fig. 5 - Vertical stress of coal pillar for different models along E-E' cross section.

SOLID LINE, PILLAR STRESS WTTED LINE: SAFETY FACTOR 0 ORIGINAL MODEL WEAK ROOF AND FLOOR MODEL

7
VERTICAL STRESS
6 -

-20

.----.----. ----.-HORIZONTAL STRESS


0 . . . -

" a

e t

1 -

m = = - -

Fig. 6 - Pillar stress and safety factor for the original model and the modified model with weak roof and floor.
896 SEPTEMBER 1988

MINING ENGINEERING

" 4
a
0

j -

6 -

5 -

P I

4 -

E s

3a -

"

.
2

MODIFIED ROOF STRATA SEQUENCE

0
1 -

ORlGlNU STRATA SEQUENCE

0 TRIBUTARY AREA THEORY

:
0

Fig. 7 - A v e r a g e pillar stress determined by tributary area t h e o r y and FEM f o r ditferent models.

There is another factor that usually cannot be considered in the traditional method, the effect of strata sequence. Fig. 7 shows the average pillar stress from both the tributary area theory and finite element analysis. The average pillar stress by the tributary area theory is the same for both the original and the modified models with different roof strata sequence. However, the average pillar stress from the finite element analysis is smaller for the modified strata sequence than the original model.

For this mine, the traditional design methods predicted that stump pillars would fail whereas the finite element analysis indicated that they would be stable. The field observations during and one-and-a-half years after mining confirmed the results of the finite element analyses. The horizontal stress confinement within the pillars plays an important role in the underground pillar behavior. It increases the pillar strength to varying extent depending on the overburden depth, strata sequence,and the material properties of the roof and floor. 4

Conclusions
The traditional pillar design method usually overestimates the pillar size. For smaller coal pillars, the averagepillar stress determined from the tributary area theory is usually larger than the real pillar stress in underground. In other words, the pillar loading from the tributary area theory is a little overestimated. The three-dimensionalfinite element analysis not only can model the mine layout and mining sequence, but also can simulate the interaction of the roof, coal, and floor. The pillar design based on this method considered the pillar, roof and floor as an integrated structure, which generally yields more accurate results.

References
Drucker, D.C.. and Prager,W.J., 1982,'Soil MechanicsandPlasticAnalysis of Limit Design." Applied Mathematic Quarterly, Vol. 10, No. 2, pp. 157-165. Hsiung. S.M. and Peng,S.S.. 1987, "Controlof Floor HeavewithProper Mine Design-Three Case Studies." Mining Science and Technology, Vol. 4, No. 3, pp. 257-272. McCormick. C.W.. 1981. The NASTRAN User's Manual (Level 17.5). National Aeronautics and Space Administration. Washington, DC. Peng, S.S., 1986, Coal Mine Ground Control, 2nd ed., Wiley, New Yolk, 491 pp Su, W.H.. and Peng, S.S., 1986, 'Investigation of me Causes of Roof Falls in a Deep Underground Coal Mine," Trans., SME, Vol.. 280, pp. 2019-2023.

MINING ENGINEERING

SEPTEMBER 1988 897

You might also like