QuatJEngGeol-v25-1992-InfluênciaÁguaResistênciaDeformabilidadeRocha-Hawkins&McConnell

Quarterly Journal of Engineering Geology, 25, 115-130. 0481-2085 92 $03.50 © 1992 The Geological Society

Sensitivity of sandstone strength and deformability to changes in moisture content

A. B. Hawkins & B. J. McConnell

Engineering Geology Research Group, Department of Geology, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 I RJ, UK

Abstract

The loss in uniaxial compressive strength between UCSar~, and UCS~,, for 35 British sandstone types is discussed. The difference between UCSd,y and UCSsa , for the clay-rich Cretaceous Greensand was 78% while for the Siliceous Sandstone the strength decreased by only 8%. Over the whole range of sandstones, however, the view that the weaker varieties are more sensitive to moisture content is not supported. The degree of sensitivity to moisture content is controlled primarily by the proportions of quartz and clay minerals present and to a lesser extent by the rock microfabric. Research indicates that the development of pore pressure during loading is negligible especially in pure sandstones and hence does not play a significant role in moisture related strength reduction.

The work has demonstrated a large variation in sensitivity to moisture content throughout the range of sandstones studied and has shown that an increase in moisture content of as little as I% from the dry state can have a marked effect on both strength and deformability. A classification scheme for moisture content sensitivity is proposed.

Introduction

The influence of moisture on the strength and deformability of sedimentary rock types was reported in detail by Colback & Wild (1965) who describe work on shale lithologies and a quartzitic sandstone. For both lithologies they demonstrated a 50% loss in uniaxial compressive strength from the dry to the saturated condition. Burshtein (1969) studied the effect of moisture on the strength and deformability of a number of Russian sedimentary rock types including a quartz arenite and clay-rich sandstone. For the quartz arenites he found that an increase in moisture content from 0 to 4% decreased the uniaxial compressive strength by 50% while for the clay-rich variety an increase in moisture content to 1.5% reduced the

compressive strength to approximately a third of its initial value. In addition, Burshtein found that tensile strength and modulus of deformation were both sensitive to moisture content for the two sedimentary rocks studied and that repeated wetting and drying cycles had a marked detrimental effect on the properties. Van Eeckhout & Peng (1975) studied the effect of humidity on the strength and elastic properties of a number of shales obtained from coal workings. Using relative humidities of 0, 48 and 100% they showed that with an increased moisture content there was a reduction in UCS and Young's modulus and an increase in Poisson's ratio.

The sensitivity of different rock types to moisture is highly variable. Priest & Selvakumar (1982) studied two sandstones and three limestones and established a relationship between uniaxial compressive strength and moisture content for each rock type. Dyke & Dober- einer (1991) demonstrated the variation in uniaxial compressive strength with moisture content for three quartz arenites ranging in dry strength from approxi- mately 34 to 70 MPa. They concluded that generally the weaker the rock, the more sensitive it is to changes in moisture content. Their results are summarized in Table 1 which shows that the reduction from dry to saturated strength varied from 24 to 34%; the Waterstone showing the greatest variation.

Many authors consider that stress corrosion (Atkin- son 1984) is the predominant process in moisture related strength reduction in rocks. The theory postu- lates that in crystalline silicates and silicate glasses the strained Si-O bonds at crack tips react more readily with environmental agents. For silicate glasses and quartz in moist environments the general expression for weakening has been revised by Atkinson (1979):

[H-O-H1 + [-Si-O-Si-] = [-Si-OH.HO-Si-]

TABLE 1. Summa O' of the uniaxial compressive strength results presented by Dyke & Doberemer (1991)

UCSdry U CS~.al Strength loss Sandstone type (MPa) (MPa) (%)

Penrith Sandstone 70 53 24 Bunter Sandstone 48 36 25 Waterstone 34 22.5 34

116 A. B. H A W K I N S & B. J. M C C O N N E L L

z~ :1 p

#'

• 23

II I o 8, 2 4 - 2 8

15 ,16 • 17

~ ' 1 2 + 1 3

• 5+6

~ 7 , 2 9 + 3 0 • • 3 1 + 3 2

32 3 4 + 3 5 33 •

FIG. 1. Sampling localities for the sandstones studied.

Atkinson & Meredith (1981) emphasize that this equation is derived from experimental work on glasses and express uncertainty about the extent to which any of the proposed stress corrosion weakening reactions may apply to complex silicate minerals such as feldspar, mica, pyroxenes and amphiboles. The role of stress corrosion in the deformation of calcitic rocks is even less well understood; there must therefore also be some doubt about the effect of this on many of the impure arenites which contain significant proportions of minerals other than quartz.

A number of other mechanisms of strength reduc- tion have been suggested including capillary tension decrease, pore pressure increase, reduction in friction and chemical and physical deterioration. These are described in detail by Van Eeckhout (1976). Although a number of these processes may occur simultaneously during the deformation of wet rocks, for certain rock types and conditions some will be more significant than others.

Although BS 5930: 1981 shows four different sandstone types, it is unfortunate that there is still a tendency to group all sandstones together and assume they have similar geomechanical properties, ignoring

their different mineralogies and textures. This paper highlights the influence of the mineralogy and texture on moisture sensitivity and demonstrates the effects of moisture content on a large number of different sandstones in terms of rock strength and elasticity.

Sandstones studied

Thirty five different sandstones were sampled from 21 localities within the British Isles, ranging in age from Precambrian to Cretaceous (Fig. 1). The sources were chosen to provide a wide range of sandstone lithologies in terms of mineralogy and texture, from very pure quartzites to immature clay-rich lith-arenites and greywackes (Table 2). The fresh block samples were obtained from working quarries, tunnels and road excavations. In some cases, such as the Pennant and Penrith Sandstones, a number of varieties were collected. This allowed a study of the effects of subtle differences in mineralogy and texture on the geomecha- nical properties within a single stratigraphic unit.

SANDSTONE STRENGTH

TABLE 2. Names, ages and sampling localities ./'or the 35 sandstones studied

117

No Name Code Age Sampling locality

l Applecross A Torridonian Isle of Skye 2 Donegal Quartzite DQ Dalradian Donegal 3 Basal Quartzite BQ Cambrian Isle of Skye 4 Brownstones LORS Siluro-Devonian Forest of Dean, Glos 5-6 Pilton Pi Lower Carboniferous North Devon 7 Upper Cromhall UC Lower Carboniferous Bristol, Avon 8-11 Millstone Grit MG Middle Carboniferous DurharnNorthumberland 12-13 Holcombe Brook Grit HBG Upper Carboniferous West Yorkshire 14 Siliceous Sandstone SS Upper Carboniferous Bristol. Avon 15 Elland Flags EF Upper Carboniferous West Yorkshire 1 6 - 1 7 Thornhill Rock TR Upper Carboniferous West Yorkshire 18 Middle Coal Measures MCM Upper Carboniferous Tyne and Wear 19 Crackington Formation CF Upper Carboniferous North Devon 20-22 Pennant Pn Upper Carboniferous Avon/Gloucester 23 Annan Sandstone An Permian Annam Dumfriesshire 24-28 Penrith Pr Permo-Triassic Penrith, Cumbria 29 Redcliffe R Triassic Bristol. Avon 30 Midford Sands MS Jurassic Bath, Avon 31-32 Ardingly Sandstone Ard Lower Cretaceous East Grinstead, Sussex 33 Ashdown Sandstone AS Lower Cretaceous Tunbridge Wells, Kent 34-35 Greensand G Upper Cretaceous Devizes, Wiltshire

Petrographic and textural analyses were carried out on all the selected sandstones and the mineralogies obtained by point counting are summarized in Table 3. The main sandstone-forming minerals belong to rela- tively few groups as the less stable constituents will have disintegrated or been altered during transport and diagenesis. For each sandstone the detrital and authigenic constituents were distinguished: the opaques, micas and heavy minerals are included in the 'other minerals' column. The petrographic variation of thin sections demonstrated the potential variation in microfabric and mineralogy. The Donegal Quartzite for example was almost pure quartz with no clays while in the highly porous Cretaceous Greensand there was only 32% detrital quartz with the framework grains separated by more than 50% clay minerals. Despite its metamorphic texture, the Donegal Quartzite has been included as an end-member in this study in view of its high quartz content and degree of induration. Whilst clays normally form a weak cement or a sporadically developed pore infill, when recrystal- lized they can provide a relatively strong intergranular bonding material as noted for example in the Pilton and Crackington Sandstones which have been sub- jected to high stresses during the Variscan-Hercynian orogeny.

The sedimentological classification of sandstones proposed by Dott (1964) is not applicable for geomechanical purposes. For instance, both the Sili- ceous Sandstone and Ashdown Sandstone have over 78% detrital quartz and hence classify as quartz arenite but their densities, porosities and degree of

induration are very different, resulting in the former having a UCSdry of 198 MPa while only 31 MPa was recorded for the latter. Photomicrographs presented by McConnell (1989) demonstrate the textural variations, including such parameters as grain size, porosity and degree of welding and grain suturing. Textural analyses were carried out for each sandstone, with particular emphasis on packing density and packing proximity (Kahn 1956).

Laboratory testing

Cores from the block samples were taken perpendicu- lar to bedding and prepared in accordance with the recommendations of the ISRM (Brown 1981). Suf- ficient 55mm cores were taken from each block to ensure that specimens could be tested in both the dry and saturated state and at intermediate moisture contents. Samples were oven dried at only 50~'C in order to prevent baking of the clay minerals (Pinches 1986). Initially it was hoped that saturation would be achieved by placing the samples in deionized water for a period of one month but with low-porosity samples, simple immersion under atmospheric pressure did not prove satisfactory. Trials using vacuum apparatus also gave poor results for the low-porosity/permeability varieties and, as a consequence, a cell was constructed in which fluid could be injected into the specimens at very low pressures. Using this cell it was possible to obtain good saturation at pressures of only 50kPa:

118 A. B. H A W K I N S & B. J. MCCONNELL

TABLE 3. Summary of sandstone mineralogies determined by point counting

" . ~ ~ •

.~.-~ ~ ~ E ~. ~-~ E ~=

A 20.0 52.3 2.3 3.0 0.0 13.3 0.0 0.0 8.0 1.1 DQ 88.7 1.5 2.5 0.0 0.0 0.0 0.0 0.0 0.0 7.3 BQ 79.0 5.7 0.0 0.0 15.3 0.0 0.0 0.0 0.0 0.0 LORS 40.9 7.1 26.9 3.1 0.0 13.0 4.4 0.0 0.0 4.6 PiA 54.0 4.0 10.0 16.0 9.3 0.0 0.0 0.0 0,0 6.7 PiB 61.3 10.0 5.4 4.3 0.0 9.7 5.3 0.0 0.0 4.0 UCS 37.0 1.0 19.7 0.0 0.0 0.0 10.7 31.3 0.0 0.3 MGA 74.6 2.7 4.6 13.9 0.0 0.0 0.0 0.0 0.0 4.2 MGB 69.0 10.3 6.7 12.0 0.0 0.0 0.0 0.0 0.0 2.0 MGC 74.2 5.3 3.0 12.7 0.0 0.0 2.7 0.0 0.0 2.1 MGD 79.9 3.5 2.1 9.2 0.0 3.5 0.0 0.0 0.0 1.8 HBGA 67.5 10.1 4.2 15.2 0.0 0.0 0.0 0.0 0.0 3.0 HBGB 71.7 6.1 3.1 16.7 0.0 0.0 0.0 0.0 0.0 2.4 SS 84.7 3.7 4.3 3.0 0.0 0.7 0.0 0.0 0.0 3.6 EF 53.9 8.5 19.4 12.6 0.0 2.0 0.0 0.0 0.0 3.6 TRA 53.9 6.9 28.8 5.9 1.3 0.0 0.0 0.0 0.0 3.2 TRB 72.9 8.6 2.1 14.4 0.0 0.0 0.0 0.0 0.0 2.0 MCM 59.7 14.7 5.1 9.4 4.3 2.5 0.0 0.0 0.0 4.3 CF 62.3 2.0 16.0 15.3 0.0 1.0 2.0 0.0 0.0 1.4 PnA 28.3 3.0 51.0 2.3 0.0 3.0 0.0 0.0 0.0 12.4 PnB 31.2 2.8 56.7 7.9 0.0 0.0 0.0 0.0 0.0 1.4 PnC 29.6 2.0 64.0 2.0 0.0 0.0 0.0 0.0 0.0 2.4 An 49.8 6.0 19.3 6.3 0.0 10.2 7.1 0.0 0.0 1.3 PrA 54.2 14.1 10.6 1.9 5.0 13.4 0.0 0.0 0.0 0.8 PrB 63.7 4.3 2.1 2.2 18.7 7.6 0.0 0.0 0.0 1.4 PrC 50.3 4.0 10.5 0.7 21.1 5.2 0.0 0.0 4.5 3.7 PrD 58.9 3.1 10.3 0.1 9.0 12.7 0.0 0.0 0.0 5.9 PrE 53.1 8.1 10.5 6.2 0.0 10.5 9.7 0.0 0.0 1.9 R 44.3 1.4 17.2 3.2 0.0 6.1 26.0 0.0 0.0 1.8 MS 38.3 0.7 8.0 1.7 0.0 18.3 3.3 27.3 0.0 2.4 ArdA 71.2 2.1 3.7 4.9 0.0 1.6 0.0 0.0 12.3 4.2 ArdB 66.6 3.8 0.0 8.9 0.0 0.0 0.0 0.0 18.5 2.2 AS 78.2 3.7 1.2 0.4 0.0 0.0 0.0 0.0 13.2 3.3 G 32.2 0.0 1.7 50.5 0.0 0.0 2.3 0.0 10.3 3.0 D 30.7 5.6 2.l 16.9 0.0 0.0 0.0 32.4 10.3 2.0

well below levels at which hydrofrac tur ing or other detr imental effects could occur.

In addit ion to establishing the dry and saturated strengths, 15 texturally and mineralogically different arenites were tested at intermediate levels to determine the relationship between strength and moisture con- tent. The different required humidities were established in humidi ty cabinets with a saturated salt solution present as outlined by Winston & Bates (1960). For a number of low-porosity sandstones, however, it was found to be difficult to attain intermediate moisture contents using control led humidi ty environments . In these cases the samples were saturated and subse-

quently allowed to lose moisture in a modera te humidi ty envi ronment until the required wet weight was attained.

The percentage moisture content by weight at saturat ion for the different rock types varied depending

on the effective porosity. The highest moisture content of 26.7% was achieved in the Greensand (38% porosity) while the Crackington Sandstone contained only 0.75% moisture at saturat ion (2% porosity).

Results

For most of the 35 lithologies tested, a substantial decrease in uniaxial compressive strength occurred on wetting from the 50~C oven dried state. For each sandstone the average dry strengths (over four tests), average saturated strengths (over four tests) and percentage strength loss are listed in Table 4. The max imum loss of strength was observed in the Greensand from Wiltshire (78%) while the min imum reduction (8%) occurred in the Siliceous Sandstone of

SANDSTONE STRENGTH 1 19

TABLE 4. Average dry and saturated strength values ,for each sandstone type and degree of moisture related strength reduction

UCSary UCSsa t Loss Sandstone type (MPa) (MPa) (MPa) (%)

1. Applecross A 141.3 99.3 42.0 29.7 2. Donegal Quartzite DQ 237.9 184.0 53.9 22.7 3. Basal Quartzite BQ 247.0 202.1 44.9 18.2 4. Brownstones LORS 152.0 107.4 44.6 29.3 5. Pilton (Type A) PiA 173.3 152.2 21.1 12.2 6. Pilton (Type B) PiB 227.3 208.4 18.9 8.3 7. Upper Cromhall UC 161.4 143.9 17.5 10.8 8. Millstone Grit (Type A) MGA 59.3 39.6 19.7 33.2 9. Millstone Grit (Type B) MGB 49.0 40.2 8.8 18.0

10. Millstone Grit (Type C) MGC 123.4 72.1 51.3 41.6 11. Millstone Grit (Type D) MGD 127.5 98.6 28.9 22.7 12. Holcombe Brook Grit HBGA 119.1 49.0 70.1 58.9 13. Holcombe Brook Grit HBGB 108.1 48.6 59.5 55.0 14. Siliceous Sandstone SS 198.4 182.2 16.2 8.2 15. Elland Flags EF 59.9 31.4 28.5 47.6 16. Thornhill Rock (Type A) TRA 89.9 38.4 51.5 57.3 17. Thornhill Rock (Type B) TRB 91.8 46.7 45.1 49.1 18. Middle Coal Measures MCM 37.1 25.3 11.8 31.8 19. Crackington Formation CF 298.2 232.3 65.9 22.1 20. Pennant (Type A) PnA 114.2 50.0 64.2 56.2 21. Pennant (Type B) PnB 106.2 52.6 53.6 50.5 22. Pennant (Type C) PnC 103.4 47,0 56.4 54.5 23. Annan Sandstone An 66.3 43.6 22.7 34.2 24. Penrith (Type A) PrA 66.0 53.1 12.9 19.5 25. Penrith (Type B) PrB 34.8 29.4 5.4 15.5 26. Penrith (Type C) PrC 82.0 72.8 9.2 11.2 27. Penrith (Type D) PrD 101,2 87.9 13.3 13.1 28. Penrith (Type E) PrE 59,7 40.8 18.9 31.7 29. Redcliffe R 36. I 22.4 13.7 38.0 30. Midford Sands MS 23.2 14.6 8.6 37.1 31. Ardingly Sandstone (Type A) ArdA 42.2 36.9 5.3 12.6 32, Ardingly Sandstone (Type B) ArdB 53,4 47.6 5.8 10.9 33. Ashdown Sandstone AS 30.6 32.3 - 1.7 -5.6 34. Greensand (Type A) G 10,5 2.3 8.2 78.1 35. Greensand (Type A-Dogger) D 74.5 46.7 27.8 37.3

the Bristol area (Fig. 1). Generally the least sensitive sandstones were the relatively quartz-pure varieties such as the Siliceous Sandstone and the Penrith Sandstone. With the exception of the Ashdown Sandstone, the average loss of strength was 32%. In the case of the Ashdown Sandstone the uniaxial compressive strength actually increased on saturation by an average of 5.6% over ten tests. Such a large variation in the change of strength on wetting indicates that more than one process is active; the predominant one depending ultimately on the mineralogy and texture of the particular sandstone type.

As can be seen from Fig. 2, there is little correlation between dry uniaxial compressive strength and percent- age strength loss on saturation for the 35 sandstones. This contrasts with the conclusions of Dyke & Dobereiner (1991) who imply that the weaker rocks are more sensitive to moisture. In a number of samples

reported in the present research, sandstones with greater than 200MPa uniaxial compressive strengths show a higher percentage strength loss than sandstones below 60 MPa. A similar scatter of points is seen in Fig. 3 which plots the effective porosity against the percentage strength loss suggesting that the ability of the liquid to reach propagating crack tips is not the only controlling factor in strength loss.

Figure 4 shows the sandstone strengths at various intermediate moisture contents, each point representing the average of at least two tests. Rather than a gradual decrease from UCSdry to UCSsa ,, there is generally a sudden loss in strength at low moisture contents followed by a reduced rate of loss as the moisture content increases. It can be seen that in the majority of cases most of the strength loss occurs between zero and 1% moisture content with only 10 to 20% of the compressive strength reduction occurring at moisture

120 A. B. H A W K I N S & B. J. MCCONNELL

300

0 . . 200

" 0

U 1 O0

oO

. . . ."

•g

I I I I

20 0 20 40 60 80

% Strength Loss

FIG. 2. UCS,try plotted against degree of strength loss on full saturation.

contents above 1%. As density and porosity varies between sandstone types, the est imated moisture content at which the bulk of the strength loss occurs can be expressed as a percentage of the moisture content at saturation. Thus, as can be seen from Table 5, 80 -90% of the moisture related strength reduct ion will have occurred before moisture levels reach approx- imately one third of the moisture content at saturation. The main exception is the clay-rich Greensand (Type A) where most of the strength reduction does not occur until about 15% moisture content.

The plots of uniaxial compressive strength against moisture content shown in Fig. 4 demonst ra te definite

trends which can be summarized by best-fit experi- mental curves. A more useful representation, however, is a mathemat ical best fit as applied by Priest & Selvakumar (1982). The relationships can be described by a negative exponential function of the form:

a c = a e - b ' + c

where

a¢ is the uniaxial compressive strength (MPa), w is the moisture content (%) and a, b and c are constants.

v

t, I,,,.. 0

a .

0 >

0

l, lJ

40

30

10

0 2O

• •

• . :: . . . -

o%

• • •

I I I I

0 20 40 60 80

% Strength Loss

FIG. 3. Effective porosity plotted against strength loss on full saturation.

TABLE 5. Moisture contents at which 80-90% o f strength loss will occur, expressed as a weight percentage and as a percentage o[ saturation moisture content

Sandstone type

Actual moisture content at 80-90% strength loss

(%)

Actual moisture content as a percentage of saturation

moisture content

Donegal Quartzite Lr Old Red Sandstone Millstone Grit Type D Holcombe Brook Grit Type B Thornhill Rock Type A Crackington Formation Pennant Type A Pennant Type B Pennant Type C Penrith Type A Penrith Type B Penrith Type C Penrith Type D Greensand Type A Greensand Type B

0.5 1.0 1.0 1.0 1.5 0.4 0.75 0.8 1.0 1.0 2.0 0.75 1.0

15.0 1.0

25 33 22 25 28 53 25 30 27 29 33 29 43 53 13

S A N D S T O N E S T R E N G T H 121

WdW

o . . . . . ~ o o ~ o ~ o o

WdW WdW

m

,=. ~ • o

_ o o ° o o ~ o o o o

mdW

-co

0

o r-.

W~dW

v -w ~ i

,, • l o m-- ,~' " 'I - , - , - o . • , . , I~ , . 0

Wdlfl WdW

w

.'.

• c • • , • , • , • , • , • o • ' " ' " ' " c "" , • , " , o L . , . , . , . , • , . o

o ~ o o o ~ o ~ o o o o o o o ~ ° o o o . o _ o g ~ o ~ o ~ o ~

edW ed W wdfl WdW

~ " ~ ~ •

• e , i "I"

I . . . . . •

~P

" m "

_. .. , • .

'-' ' " ' ' " ' " , " "" | • , • , • , . o • . - , . , • o -= - - I o

edW WdW ~¢IW WclW

~n

E

E

0

0

o

C.

?, cn

u.

6 U.

122 A. B. H A W K I N S & B. J. M C C O N N E L L

150

100

200

200

1.50

100

200

13_ 150

O3 09 1oo UJ rr

O0 5o

200

150

100

D

0.5

Applecross

• 1 c $,

300

200

100

D

O5

Basal Quartz i te

lO 15

D

oi 5 Pilton (Type AI i

1 o 1'5

400

300

200

100

D j s O5

Pilton (Type B)

1'.o

D ,/s ols

U~)per Cromhall 10

D

/;S s,on.Gr,,TypeA, 0 5 Ii0 1

/ 0.5

Mi l ls tone Grit (Type B)

200

150

100

/ , / ~ D

Mil ls tone Grit (Type C)

o 5 1'0 1 5

D

I combo Brook Grit , , (Type A)

0.5 1.0

200

150

!00

AXIAL STRAIN (%)

( S i l i ceous S a n d s t o n e

i 05 10

FIG. 5. Measured pore pressures plotted against imposed stress during uniaxial tests.

S A N D S T O N E S T R E N G T H 123

100

~ D

05 Elland Flacjs

o15 Middle Coal Measures

1.0

~ D

0.5 Pennth (Type E)

1.0 1 5

80

D

~ , , S Midford Sands i 0.5 1 0

Ardingly Sandstone (Type B) 0.5 10 1.5

AXIAL STRAIN (%)

D

S

ill ,Rock (Type B) 0.5 lo

S

Annan Sandstone 0.5 1'.0

O5 Redctiffe

110 15

D

~ A t d i n g l y Sandstone (Type A) 05 1.0 15

~ S

05 Ashdown Sands m

10 15

124 A. B. HAWKINS & B. J. MCCONNELL

TABLE 6. Numerical values of constants a, b and c and respective r-values for best-fit exponential equations

Sandstone a b c r-value

DQ 39.03 1.9601 184.23 0.93 LORS 29.34 0.7646 105.23 0.78 MGD 12.30 0.6821 96.27 0.71 HBGB 36.13 0.7794 48.65 0.88 TRA 45.73 1.5942 40.29 0.97 CF 84.01 6.4167 230.98 0.91 PnA 83.76 0.2306 51.02 0.86 PnB 28.81 0.5506 49.37 0.62 PnC 47.12 1.5439 47.65 0.95 PrA 7.01 0.0752 56.30 0.70 PrB 4.16 0.4061 28.90 0.87 PrC 17.27 1.0675 67.75 0.85 PrD 20.37 1.2629 87.29 0.88 G 6.14 0.1104 2.97 0.93 D 19.12 0.2567 45.79 0.77

The strength at zero moisture content is given by (a + c), the strength at saturation is given by c for high moisture content values and the parameter b is a dimensionless constant which defines the rate of strength loss with increasing moisture content. The constants a, b and c and the correlation coefficients r obtained for each of the 15 sandstone types are listed in Table 6.

These results agree approximately with those of Priest & Selvakumar (1982) but less closely with those of Burshtein (1969). The relatively sudden decrease in strength at moisture content values well before satu- ration indicates that pore pressure increase during loading of the rock is not a major controlling factor; pore pressure effects would only be significant when the rock pores are saturated or close to saturation. To confirm this, tests were carried out on a number of saturated sandstones using sensitive pore pressure measuring apparatus. Although the probes inserted into the centre of the specimens detected changes in pore pressure in a number of interconnected pores, the results indicated that only extremely small positive

pore pressures were developed and then only in the most deformable sandstones such as the Greensand (Fig. 5). More significantly, some negative pore press- ures developed during the later stages of the tests at or close to the yield point as a result of dilatancy occurring in the rock specimen, i.e. that at the start of Stage IV (Fig. 6) of the stress-strain curve as reported by Brace, Paulding & Scholz (1966).

It is of note that in a number of the more deformable sandstones (Fig. 4) the failure stress at saturation is slightly higher than that measured at 1- 2% below saturation. This may be a response to pore pressure effects when the pores are full but further work will be required to elucidate the full significance of this apparent increase in strength.

Although not demonstrated here, this research confirmed that the surface tension of the saturating liquid has a profound influence on sandstone strength. The process of change in surface energy outlined by Rehinder & Lichtman (1957) appears to be applicable as the fracture energy reduction occurs mostly during the latter part of Stage II and during Stages III and IV of the stress-strain curve when microfracturing takes place (Brace et al. 1966). The overall effect is to lower the fracture propagation energy, resulting in lower strengths, greater strains to failure and less brittle deformation.

The control of mineralogy and texture on moisture sensitivity

Sandstones which contain abundant clay minerals (Greensand) or sedimentary rock fragments (Pennant) show significant strength loss on wetting. In the clay- rich varieties the change in strength is likely to be related to the softening and possible expansion of the clay minerals. Sasaki, Kinoshita & Ishijima (1981) demonstrate the importance of expansive clay minerals within argillaceous rocks in controlling their moisture sensitivity but emphasize the difficulty in quantifying

TABLE 7. Semi-quantitative clay fraction XRD results for seven sandstone types

Types of sandstone Minerals present PiA MGD HBGA HBGB PnC PrC AS

Quartz 85-95 20-30 20-30 20-30 60-70 95-100 60-70 Feldspar 10-15 5-10 1-3 5-10 5 10 0-5 - - Kaolinite 2-3 50-60 20-30 5-10 1 0-20 - - 20-30 Illite 2-3 15-20 40-50 50-60 1 0-20 - - 3-5 Chlorite 0-1 0-3 - - - - 0-3 - - - - Calcite 0-2 5-10 1-2 - - - - - - 1-4 Muscovite - - - - 1-2 . . . . Pyrite 0-1 . . . . . . Swelling Clays . . . . . . . % Strength Loss 12.2 23.5 58.9 55.0 54.5 11.3 - 5.7

SANDSTONE STRENGTH 125

250

~ o

0

60

. . . / Cra(~ngton Formation

1.0

IV

,,-7

Penrith (Type A)

1.0

PrA i | ,

Penrtth (Type B) PrB

CF | i

0 lkPa

0 1 kPa 1.0

0 1 kPa

Penfith (Type C) PrC

1.o ' 6 lkN,

Greensand (Type A)

110 ' 0 lkPa

AXIAL STRAIN (%)

Greensand (Type B- Dogger) D i , |

1.o 0 lkPa -VE +VE

PORE PRESSURE

FIG. 6. Representative stress-strain curves and pore pressure curves for sandstones tested in dry (D) and saturated (S) conditions. The four stages of the stress-strain curve proposed by Brace et al. (1966) are shown for Penrith (Type A) diagram.

126 A. B. H A W K I N S & B. J. M C C O N N E L L

the clay minerals. When X-ray diffraction was carried out on a number of the sandstones in this research and the semi-quantitative results determined (Table 7), no swelling clays were detected in any of the sandstones, indicating that in these rocks expansive forces are not a mechanism in strength loss. High proportions of non- expansive clay minerals were detected in a number of the sandstone varieties, some of which showed strength losses of over 50% which again suggests that clay softening in the presence of water is important in controlling moisture related reduction in strength. However, this cannot be responsible for the loss of competence in the pure quartz arenites.

In an effort to elucidate the factors controlling moisture sensitivity, multiple regression analysis was carried out for all the sandstones studied. This involved the regression of the dependent variable (i.e. moisture related strength reduction) against a total of 36 textural and mineralogical variables such as poro- sity, mineralogical content, packing density, grain size and sorting. The results of the multiple regression model confirm that it is the relative proportions of quartz to clay minerals which to a large extent control the strength loss. The nature of the cement also influences moisture sensitivity. The rocks containing ferruginous cements (as observed in some Pennant varieties) and calcite cements (Triassic and Jurassic examples) are generally more susceptible to moisture than those with siliceous cements, but it is difficult to quantify the relationship.

Effect of moisture on the mode of deformation

As described by Hawkins & McConnell (1991), the mode of deformation observed in uniaxial compression testing of sandstone is influenced to a large extent by the strength of the specimen. Although the strongest varieties fail by axial cleavage and the weakest types by cataclasis, shear failure is the most common mode. Despite extensive overlap of the failure modes, wetting of the specimens results in a change from axial cleavage in the dry state to shear failure and/or a combination of shear failure and axial cleavage. Similarly in weaker varieties where shear is the dry failure mode, a combination of shear and cataclasis took place on saturation. Although no clear pattern has been recognized, the observations suggest that the changes take place in the low saturation state, i.e. less than 1% moisture.

On a microscopic scale the failure planes often showed a higher degree of disaggregation leading to a rolling mode of failure. This indicates a weakening effect of moisture on the cement/matrix resulting in microfractures propagating more easily around grains

than in the dry state. The main effect of this type of microfracturing is to produce a much less violent mode of failure, as observed in all the sandstone types tested.

In addition to showing an increased strength on saturation, the Ashdown Sandstone suffered less disaggregation of the specimens during failure than in the dry state. The geomechanical behaviour of this rock type and other poorly indurated sandstones is not well understood and will require further work but the observations suggest that pore pressures may play an important role.

Effect of moisture on deformability

Average values of tangent Young's modulus at 50% ultimate stress and secant Young's modulus at failure for both dry and saturated specimens are listed in Table 8. Representative stress-strain curves for sand- stones tested in dry and saturated conditions (Fig. 6) indicate that the secant Young's modulus decreases from the dry state. Sandstones which showed a high degree of elasticity in the dry state did not show significant change in the shape of the t~-e curve, despite experiencing marked decreases in Young's moduli on saturation (e.g. the Cambrian Basal Quartzite). In many cases, however, saturation produces a more 'S- shaped' stress-strain curve indicating a less elastic style of deformation. It is particularly noteworthy that for many sandstones the early part of the stress-strain curve (Stage 1, Brace et al. 1966) becomes more concave upwards. This indicates that the strength reduction processes are active at the lowest stresses, weakening the framework and causing a high degree of pore closure and cement deformation.

Figure 7 shows the sequential stress-strain curves for the Lower Old Red Sandstone and Pennant (Type A) tested at intermediate moisture contents. The variation in Eta ° and Ese c for these two sandstone types is plotted against moisture content in Fig. 8. It can be seen that the progressive decrease in elasticity of deformation is initiated at low moisture contents and the E,a . and Ese c show similar trends to those produced by plotting strength against moisture content. Values generally decrease sharply with increase in moisture content but above approximately 1% there is little change. Ese~ appears to show a more uniform relationship with moisture content than does E,a ., which for many sandstones even at low moisture contents shows a slight increase from that in the dry state. This may be attributed to the development of a more S-shaped profile at low moisture contents, without substantial increase in axial strain to failure.

It is suggested that the overall brittleness of the rock is substantially reduced by wetting. Hudson & Morgan (1975) defined a Brittleness Index which could

SANDSTONE S T R E N G T H

TABLE 8. Average values o f Et~ . and Ese c measured on dr)' and saturated sandstones

127

Sst type UCSdry tan Edry t a n Esa t sec Edry sec Esa ,

A 141.3 33.3 23.5 19.9 12.4 DQ 237.9 38.5 24.0 30.8 22.3 BQ 247.0 32.3 30.0 33.6 28.9 LORS 152.0 32.3 26.2 23.4 18.9 PiA 173.3 39.3 38.2 43.3 37.5 PiB 227.3 39.0 33.4 31.2 28.0 UCS 161.4 29.0 22.2 29.3 22.5 MGA 59.3 12.8 11.7 8.9 6.9 MGB 49.0 15.4 17.4 l ! .4 9. I MC 123.4 22.6 17.0 13.7 11.4 MGD 127.5 23.1 16.4 19.0 13.9 HBGA i 19.1 19.9 7.6 13.5 5.5 HBGB 108.1 14.1 9.0 11.4 7.1 SS 198.4 46.0 37.6 34.0 29.4 EF 59.9 10.6 5.8 8.6 5.3 TRA 89.9 14.5 5.9 10.6 5.3 TRB 91.8 11.8 6.7 10.0 5.5 MCM 37.1 11.6 5.8 8.7 5.6 CF 298.2 36.4 19.9 30.0 18.5 PnA ! 14.2 21.2 12.2 15.9 15.7 PnB 106.2 26.6 15.2 19.3 12.2 PnC 103.4 18.9 13.3 15.0 10.0 An 66.3 17.0 5.1 8.8 6.8 PrA 66.0 17.7 15.4 12.4 9.9 PrB 34.8 8.1 7.9 6.4 5.8 PrC 82.0 28.0 16.2 20.4 14.6 PrD I 01.2 24.4 33.3 25.0 24.3 PrE 59.7 I0.1 7.8 8.5 6.8 R 36.1 8.8 5.3 5.4 4.0 MS 23.2 6.1 5.1 3.4 2.8 ArdA 42.2 7.9 8.6 7.1 7.4 ArdB 53.4 6.3 5.9 5.9 5.4 AS 30.6 6.5 5.7 4.8 4.7 G 10.5 1.6 0.9 1.2 0.8 D 74.5 16.0 15.4 14.9 11.3

be de te rmined f rom the comple te s t ress-s t ra in curve for rocks tested in a servo-control led machine . Priest & Se lvakumar (1982) used this m e t h o d to obta in brittle- ness values for the Bunter and Carbon i fe rous sand- s tones at var ious mois tu re conten ts and found a sharp d rop in brit t leness occurs f rom zero to 1% mois ture content , parallell ing the observed d r o p in s trength. They also found that in some cases the bri t t leness increased gradual ly above 1% but were unable to explain the p h e n o m e n o n .

The presence of mois tu re generally causes an increase in Poisson 's ratio but the effect is less well defined than that observed with uniaxial compress ive s t rength and Y o u n g ' s modu lus . A n u m b e r o f sand- stones including L O R S and PnA showed slight decreases in Poisson 's ratio at low mois ture con ten t s fol lowed by a general increase. These results are similar to those repor ted by Priest & Se lvakumar (1982). It is cons idered that in a n u m b e r o f sands tones the differing values o f Poisson 's ratio measured ci rcumferent ia l ly

may be associated with the variable radial strains which occur over relatively small areas o f a specimen, as a result o f an increased degree o f local izat ion o f de fo rma t ion in the presence of mois ture .

Discussion and conclusions

It has been shown that:

1. The sensitivity o f s t rength and deformabi l i ty is highly variable across the range o f sands tone types. Loss o f s t rength f rom dry to sa tura ted condi t ions varied f rom 8.2% in the Siliceous Sands tone to 78.1% in the Greensand with a range o f in te rmedia te values. On the basis o f this research it is possible to suggest division o f sands tones into classes accord ing to their loss in s t rength f rom the dry to the sa tura ted cond i t ion (Table 9). It can be seen that the lower sensitivity

128 A. B. HAWKINS & B. J. MCCONNELL

2°° l 180

1 4 0

~" 120

2 0 "t

(LORS)

0% 0.06%

/ / 09°A ,/,

" " * 1.6°o 2.1% 2. . . * 2.9"1o

r 0.5% AXIAL STRAIN INCREASING MOISTURE CONTENT

0 I P e n n a n t (type A)

~ ~ ~ 02% / / . . 0 5 %

J / 1.1% 1 9% ~4~ 15*/, ' 2.9°V,

--, ~

0.50/0 AXIAL STRAIN INCREASING MOISTURE CONTENT

FIG. 7. Sequential stress-strain curves for the Lower Old Red Sandstone and Pennant (Type A).

classes contain sandstones which have relatively high quartz contents and that the sensitivity increases progressively with higher proportions of rock frag- ments and clay minerals.

2. The degree of sensitivity to change in moisture content is controlled primarily by the mineralogy of the specific sandstone type and to a lesser extent by the rock microfabric. The important considerations are the relative proportions of quartz to clay minerals. This is supported by the multiple regression analyses.

3. In contrast to previous work it is now apparent that the weaker sandstones are not necessarily more sensitive to moisture. In a number of cases, sandstones with uniaxial compressive strengths higher than 200 MPa show greater percentage strength loss than sandstones below 60 MPa.

4. Although stress corrosion cracking is important in quartz-rich rocks, clay softening becomes increas- ingly significant in creating moisture sensitivity in the more clay-rich sandstones. Expansive clay minerals were not detected in the sandstones studied.

5. Except in the most deformable sandstones where

it is related to pore closure and subsequent dilatancy during loading, the effect of pore pressure development with increasing moisture content is negligible.

6. Observations on sandstones tested at dry, inter- mediate and saturated strength indicate that in the majority of cases most of the strength reduction has occurred by 1% moisture content (i.e. at approxi- mately one third of the moisture content at saturation).

7. The relationship between UCS and moisture content for any sandstone can be represented by a best-fit exponential curve. Such a curve could be used to predict strength at certain moisture contents if a minimum of three or four points are established by testing.

8. An increase in moisture content has a similar effect on the Young's modulus to that observed on uniaxial compressive strength. Ese c shows a clear relationship with moisture content while Eta" shows a greater spread of results. This is due to the develop- ment of a more S-shaped a - e curve in the presence of moisture. Stages I, III and IV type deformation are thus more significant, indicating that:

SANDSTONE STRENGTH 129

(a) pore closure and collapse are more intense in the presence of moisture, and

(b) stable microfracture occurs earlier and to a greater degree towards failure.

and it is recommended that specimens should be maintained in a moist condition between collection and testing.

9. Intergranular microfracturing is more intense in ~c wet rocks, due to an increased strength disparity between the cement and the framework grains resulting in the rolling of grains along failure planes as opposed to fracturing through the grains.

10. Moisture has a marked effect on the stiffness of 3o the rock at low stresses, significantly less than those at failure, i.e. that part of the stress-strain curve of most ,~ importance in engineering practice. ~

The results emphasize the importance of knowing the moisture content at which rocks have been tested when considering how a rock will perform, for example in a road cutting or subsurface excavation. The work highlights the question of whether tests should be undertaken at a dry, saturated or intermediate mois- ture content. It is reasonable to argue that in their natural state most rocks will have a moisture content of more than 1%. As the results presented above demonstrate that in most cases there is negligible loss of strength when the moisture content is increased from 1% to saturation, it is realistic to suggest that rocks should be tested in the wet condition as standard procedure.

BS 5930:1981 draws attention to the fact that the uniaxial compressive strength is dependent on the moisture content of the specimen but does not give any recommendations as to the moisture content at which rocks should be tested for engineering purposes.

This research has shown that with the exception of the most clay-rich sandstones it is appropriate to test the sample in its wet state; provided it has a moisture content of at least 1 to 2% full saturation is not necessary. In the most clay-rich sandstones the speci- mens should be tested at a moisture content greater than one third of the moisture content at saturation. It is also important to appreciate that oven drying of clay-rich samples may irreversibly change their state

20

10

L O R S

+ + Elan

• Ese c

÷

÷ ÷ ÷

÷

÷ ÷

÷ •

• t

o

2

W%

2O

w

10

P e n n a n t ( T y p e A)

+ Eta n

• Ese c

÷

• ÷ ÷ ÷ ÷ •

O i

~] O+o

FIG. 8. E,,, and E,~ plotted against moisture content for the Lower Old Red Sandstone and Pennant (Type A).

TABLE 9. Division o/" sandstones into sensitivity classes according to their loss of strength from the dry to the saturated condition

% loss Sensitivity class UK example Lithological type

0 10 Very low Siliceous Sandstone Quartz arenite 10-25 Low Penrith (Type B) Quartz arenite 25 40 Moderate Millstone Grit (Type A) Sublitharenite 40 55 High Elland Flags Sublitharenite 55 70 Very high Pennant Lithic wacke

>70 Extremely high Greensand Quartz wacke

130 A. B. H A W K I N S & B. J. MCCONNELL

Fo r ma jo r engineer ing works it is s trongly recom- m e n d e d that the UCSa~y is ob ta ined for the purposes o f s t rength classification while for the actual engineer ing design it is essential to establish the wet s t rength and ideally the mois ture sensitivity o f the rocks in order to assess their potent ia l change in s t rength and deforma- bility.

References

ATK1NSON, B. K. 1979. A fracture mechanics study of subcritical tensile cracking of quartz in wet environ- ments. Pure and Applied Geophysics, 117, 1011-1024. 1982. Subcritical crack propagation in rocks: theory,

experimental results and applications. Journal 0/" Struc- tural Geology, 4, 41 56. 1984. Subcritical crack growth in geological materials.

Journal of Geophysical Research, 89, 4077 4144. - - & MEREDITH, P. G. 1981. Stress corrosion cracking of

quartz: A note on the influence of the chemical environment. Tectonophysics, 77, TI 11.

BELL, F. G. 1978. Petrographical factors relating to porosity and permeability in the Fell Sandstone. Quarterly Journal Of Engineering Geology, I 1, 113 126.

BRACE, W. F., PAULDING, B. W. & SCHOLZ, C. 1966. Dilatancy in the fracture of crystalline rocks. Journal qf Geophysi- cal Research, 71, 3939 3953.

BRITISH STANDARDS INSTITUTION 1981. Code of practice .[br site investigations: BS 5930. H.M.S.O., London.

BROWN, E. T. (ed.) 1981. I.S.R.M. Suggested Method, .fbr Rock Characterisation, Testing and Monitorhtg. Perga- mon, Oxford.

BURSHTEIN, L. S. 1969. Effect of moisture on the strength and deformability of sandstone. In: Soviet Mining Science, N4, 573 576.

COLBACK, P. S. B. & WILD, B. L. 1965. The influence of moisture content on the compressive strength of rock. Proceedings of the 3rd Canadian Rock Mechanics Symposium, 65 83.

DOTT R. L. Jr. 1964. Wacke, graywacke and matrix--what approach to immature sandstone classification? Journal of Sedimentary Petrology, 34, 625~32.

DYKE, C. G. & DOBEREXNER, L, 1991. Evaluating the strength and deformability of sandstones. Quarterly Journal of Engineering Geology, 24, 123 134.

HAWKINS, A. B. & MCCONNELL B. J. 1991. Influence of geology on geomechanical properties of sandstone. Proceedings of the International Society ,/br Rock Mech- anics, Aachen, 1, 257-260.

HL'I)SON, J. A. & MORGAN, J. M. 1975. Compressive Failure of Chalk. Transport and Road Research Laboratory Re- port 681.

KAHN, J. S. 1956. The analysis and distribution of the properties of packing in sand size sediments, i. On the measurement of packing in sandstones. Journal of Geology, 64, 385-395.

MCCONNELL, B. J. 1989. Factors Controlling Sandstone Strength and Deformabilio' in Uniaxial Compression. PhD thesis, University of Bristol.

PINCHES, G. M. 1986. Geotechnical Implications of Weathering of Upper Trias in the Bristol Area. PhD thesis, University of Bristol.

PRIEST, S. D. & SELVAK~MAR, S. 1982. The Failure Character- istics of Selected British Rocks. Transport and Road Research Laboratory Report.

REHINDER, P. & LICHTMAN, V. 1957. Effect of surface active media on strains and rupture in solids. Proceedings 2nd International Congress on Surface ActivO', 3, 563-582.

SASAKI, T., KINOSHITA, S. & ISHIJIMA, Y. 1981. A study on water sensitivity of argillaceous rocks. Proceedings of the International Symposium on Weak Rock, Tokyo, 149- 154.

VAN EECKHOUT, E. M. 1976. The Mechanisms of strength reduction due to moisture in coal mine shales. International Journal of Rock Mechanics, Mining Science & Geomechanical Abstracts, 13, 61 67. & l~.'q6, S. S. 1975. The effect of humidity on the

compliance characteristics of coal mine shales. Inter- national Journal o[ Rock Mechanics and Mining Science, 12, 335-340.

WINSTON, P. W. & BATES, D. H. 1960. Saturated solutions for the control of humidity in biological research. Ecology, 41, 232 237.