You are on page 1of 9

cold regions science and technology

ELSEVIER
Cold Regions Science and Technology 23 ( 1995 ) 191-199

Triaxial tests on dry, naturally occurring snow


R . M . L a n g a, W.L. H a r r i s o n b
aU.S. Army Cold Regions Research and Engineering Laboratory, 72 Lyme Road, Hanover, NH 03755-1290, USA bUniversity of Utah, Dept. of Geography, 270 Orson Spencer Hall, Salt Lake City, UT 84112, USA Received 27 May 1993; accepted after revision 18 March 1994

Abstract This study presents and discusses the results o f triaxial tests conducted on dry, naturally occurring snow. The applied strain rate for the fresh snow was 1.01 10-5 s-~ and the confining pressure was varied from 0 to 41.37 kPa. Stress-deformation curves are presented and snow behavior is represented within the context o f a critical state model (Roscoe et al., 1963 ). U n d e r these specific loading conditions the stress-deformation curves demonstrate that snow exhibits an instantaneous response which can be characterized as an isotropic, nonlinear, elasticplastic strain hardening material for computational purposes. The loading conditions did not promote the viscous behavior o f the snow and viscous effects are not considered. When the results are corrected to true stress, the snow continues to deform without an increase in load beyond the limit load. Also, the effect of increased ultimate strength with increased confining pressure was, in general, apparent for snow.

1. Introduction Data concerning the mechanical response of snow to measured, applied loads are very limited. The mechanical response of snow to an applied load is of interest in diverse fields of study, e.g. from the nature of flowing snow in an avalanche to the problems encountered in trafficability. The mathematical description of any material's mechanical behavior is always idealized and the idealization is formulated for the specific loading condition under consideration. A good example is the behavior of metals, which in many structural loading scenarios may be sufficiently described as a Hookean material but in other cases as perfectly plastic or perhaps hardening occurs outside the initial, linear portion of the stress-strain curve. When one considers the range

of deformations, deformation rates, temperatures, etc., to which a material in the solid state may be subjected, it becomes obvious that a constitutive law can only serve to approximate a material's response to loading in an appropriately restricted range. Formulating the appropriate, idealized constitutive behavior for any material is never a trivial or simple process. Snow does not conform to simple modelling in the same manner as a material like metals, as snow behavior is affected by hydrostatic pressure, and behaves differently under compression and tension. Under the specific, applied loading conditions that were considered in this paper, snow demonstrates an instantaneous response that may be interpreted as an elastic-plastic strain hardening behavior. This approach, or material model, is not intended to be used as an accurate model for the detailed behavior of the

0165-232X/95/$09.50 1995 Elsevier Science B.V. All rights reserved SSDI 0165-232X (94) 00005-I

192

R.M. Lang, W..L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199

Spr*ng

~J_
Mercury

Cylinder

!~.ale

Flow
Meter

Fle~l~eTube (S)Specimen
AJr Svppty

~: O.
0. C.

Fig. 1. Schematic oftriaxial test apparatus. (a) Chamber pressure control, (b) triaxial cell, (c) constant pressure pore volume change measurement device.

internal physical nature of snow, but as a computational tool for various macroscopic problems in snow science. These data are of particular interest for use in finite element and finite difference models. The experimental data are the result of triaxial tests on snow. The choice of triaxial testing results from the inherent assumption that snow response can be described by a plasticity theory for the loading scenario of interest, as snow exhibits the property of continuing deformation at a constant stress level beyond some threshold load (elastic limit), which depends on the confining pressure. This is the classical definition of "ideally plastic" or "perfectly plastic" behavior (see, for example, Chert, 1975 ).

2. Experimental results

2.1. Test apparatus and methodology


The triaxial apparatus used to perform the experiments with snow was adapted from that used

for soils. Modifications were necessary to accommodate the low compressive strength of snow, measurement at temperatures below freezing, and the use of a compressible pore fluid. The triaxial cell is of the Norwegian type (Andresen et al., 1957; Andresen and Simons, 1960) (see Fig. 1 ), which was modified to accommodate a load transducer inside the chamber so that the friction of the loading piston packing would not effect the stress measurements. For snow, deviatoric (axial) stresses of less than 0.1 MPa are of interest (Naito, 1973). The frictional forces due to the piston packing were estimated to be of the order of 0.04 MPa (Andresen et al., 1957) which is unacceptably high for measurement of the stress levels of interest without putting the load measuring device inside the cell. Also, it is necessary for the confinement membrane to have a negligible resistance to deformation when compared with the forces required to deform snow. The use of air as the working fluid in the

R.M. Lang, W.L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199
I00
I I '

193

Tro4og-I p-aOS k a / m ~
P.o
50

r,-z'e ~

111

I J ~ . ~

"r't0409-2 I p,328 k~/m3 P.iO.341d~ J T.-z'e /

.~f'----'7",
,

I I I
-

12
8

rll

o
150 770409.31

,I

! : i

4 0

~
~

I I
/11

!
100

p-328

kq/m3

P,20.68 kPa

I~

p.284 kq/mS P-31.03 kPa ._T.-Z "C.

770409-4

15% Strain I

~16 ~12

50

6 4

w
I

wl
2 3

displacement

o (cm)
I

I00
m
|sample

770409-5 taken vertically)

50 - T , - Z"C

og
~4

f _ . - . - - " -', -displacement

, I
(cm)

Fig. 2. Stress-deformation curves from the first season tests, with p the snow density, P the confining pressure and Tthe temperature. Dashed line is corrected for the diameter increase in the sample (true stress ). zJP'gives the corresponding value of volume change due to axial (true) stress for the indicated stress-displacement value at 15% strain.

chamber and pore spaces eliminated the problem of finding a fluid compatible with snow. The use of a compressible fluid does not introduce any problems when volume changes are measured at constant pressure, but when the pressure varies, Boyles Law (Brady and Humiston, 1978) must be applied. Accurate measurements of the initial pore volume and of the pore volume change are also requisite. The pore volume change measurement device is a U-tube manometer with a stationary arm and a spring suspended arm connected by flexible tubing. The volume change is read on the stationary arm which is graduated in O. l ml divisions. The spring suspended portion

of the manometer positions itself to correct the imbalance in mercury levels caused by volume changes. In order to do this, the spring constant, K, is
K = AcPng/ ( 2 -- Pa/ PHg ) -- k

where Ac is the cross sectional area of the suspended mercury reservoir, Prig is the density of mercury, Pa is the density of air and k is the weight of flexible tubing and mercury per centimeter. A change in volume of the system causes a slight pressure differential across the arms of the manometer. The mercury flows so as to equalize the pressure and the spring adjusts to the change in

194

R.M. Lang, W.L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199
ISO

J
7'704 iO-"~

7704t0-4

I
I / "

, J
I00

p ~320 kqlm 3 P-13.79 kPa


mTI

E2

SO

///~"

o
150 m

- ,,o,lo:o
T - - 2 C

p - 3 2 0 kg/m 3

'1
/"

I i/ l
l i

I00

501

T
I

II
2

I-1
3 0

i
I

I I

displacement (cm)

Fig. 3. Stress-deformation curves from the first season tests, with p the snow density, P the confining pressure and T the temperature. Dashed line is corrected for the diameter increase in the sample (true stress ). A Vgives the corresponding value of volume change due to axial (true) stress for the indicated stress-displacement value at 15% strain.

weight, maintaining the preset pressure across the device. The volume change is recorded as a change in level of mercury in the fixed manometer arm. An x - y recorder was used to record the axial load and deformation. Pressure and flow control regulators allowed complete control of the chamber air supply and pressure. A Soiltest brand precision motorized compression device was used in the field to control and apply the axial load. The snow samples were cylindrical with a volume of 391 cm 3, a length of 14.95 cm and a diameter of 5.77 cm. After weighing for density, the samples were encased in a confining membrane with a solid circular cap on top and a porous platen on the bottom. The porous platen allowed interaction of the pore air with the air in the pore volume measurement device. Tests conducted at atmospheric pressure in the chamber were performed with and without membranes to determine effects of the membrane. No effects were discernible.

2.2. Snow properties


The field laboratory was located in Albion Basin, Alta, Utah. Alta is well known for the consistent precipitation of dry, low density snow during the winter season. The snow samples used for the naturally occurring snow tests during the first test season were collected and tested after sundown when the ambient temperature was below freezing. The grain size displayed negligible gradation. The densities of the samples ranged from 170 k g / m 3 to 376 k g / m 3. The seasonal snowpack depth was subnormal (approximate depth of 1 m) for the test area during the first test season. The pack contained snow which ranged in age from 2 weeks to 3 months, all within the 1 m depth. The snowpack conditions for the second test season were normal for the area. The grain size distribution was studied in order to assess the structural properties of a snowpack sample in relation to the critical density of

R.M. Lang, W.L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199

195

lOG
A a

780221-02

pp:0230 k,/m' 50 ~p,259 ko/m 'L P,O


T,-9"C /

0
displacement (cm)

ZOO
A

I
780221-03

p -Z32 kg/m3 P-O 50 - T - -8"C

i
f i
i

IL o
i I '

I/

,.2.2 . , . 3

,
3
displacement (cm)

,
2

."
"~

5oi-/'-V

raozz,-o6-

I I
displacement (cm)

IPx~~'e z

P,O

Fig. 4. S t r e s s - d e f o r m a t i o n curves from the second season tests, with p the snow density, P the confining pressure and T the temperature.

dry snow. Critical density is defined as the density that is reached through unconfined compaction where volume change ceases (see, for example, Drucker, 1967). In order to determine grain size distribution, a 500 cc sample of snow was gently separated and placed in the top of a stack of two sieves 2.38 mm, and 0.84 mm, and a pan to retain any grains passing the 0.84 mm sieve. The stack was agitated for 5 minutes and inspected. The result was a general regrouping of the particles with little exhibition of particle distribution. Temperature control had been carefully maintained and it was concluded that grain separation would have occurred only if the agitation had been severe. During the second test season at Alta, it was noted that the grain size profile of the snowpack was uniform except for the layer of depth hoar at the ground-snowpack

interface. This was 0.45 m in depth in a total pack depth of 3.0 m at the test site. 2.3. Test results The stress-strain curves of the undisturbed field samples are shown in Figs. 2 through 4. The densities p, chamber pressures P, volumetric changes AV, and ambient temperatures for each sample are given in Table 1. R2 is the stress at 2 cm deformation ( 13.5% strain). The volumetric changes A V are expressed as the value at the applied confining pressure followed by the value at 15% strain. The strain rate for the curves in Figs. 2 through 4 was 1.01 10- 5 s - ~. One salient feature of the snow under loading was the continuous smooth deformation of the sample, i.e., no indication of collapse-regain patterns as re-

196

R.M. Lang, W.L. Harrison / CoM Regions Science and Technology 23 (1995) 19 I- 199

Table 1 Parameters for dry, naturally occurring snow tests Test No. p (kg/m 3) 168 178 306 328 328 284 312 d 328 328 320 320 376 320 320 258 230 232 282 276 P (kPa) 0 10.34 0 10.34 20.68 31.03 0 0 13.79 27.58 41.37 34.47 0 6.89 0 0 0 0 0 AV~ (mE) 0/3.6 b c c 9.4/9.8 7.7/14.4 15.8/5.7 0/3.5 0/10.4 2.6/10.9 11.3/7.5 17.2/7.0 13.1/e 0/10.0 3.7/14.8 R2 (kPa) 29.3 28.9 68.0 82.0 110.0 80.0 43.0 90.0 97.0 104.0 110.0 83.0 70.0 75.0 43.0 36.0 85.0 84.0 T (C) -4 -4 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -9 -8 -8 -6 -6

770330-1 770330-2 770409-1 770409-2 770409-3 770409-4 770409-5 770410-3 770410-4 770410-5 770410-6 770410-8 770410-9 770410-10 780221-1 780221-3 780221-5 780221-5 780221-6

aA Vbefore " / " is due to confining pressure only; AVafter " / " is the value at 15% strain. bCorrugated membrane. CMembrane leakage. dVertical sample. eVirgin compressive test only.

ported by others (Kinosita, 1967; Shinojima, 1967) For the applied strain rate, the snow samples deformed continuously, first exhibiting linear behavior in a very narrow range up to the limit load. The behavior after the linear portion of the curves is that of a plastic strain-hardening material with a dependence on confining pressure (Figs. 2 and 3). The dashed lines indicate that the stresses have been corrected for the increase in diameter, i.e. converted to a "true stress". The "true stress"-deformation curves would imply that the snow is approaching plastic behavior as the snow samples continue to deform although the applied stress level is not increasing. In general, increasing the confining pressure increased the ultimate strength of the snow sampies. This behavior is not displayed in the transition from test #770409-3 and #770409-4. The test pressure was increased from 20.68 kPa to 31.03 kPa, but the density of the sample in

#770409-4 is 13% lower. Although density is not an ideal indicator of strength, it may have restricted the ultimate strength in this particular sample. Fig. 4 shows the second season tests (series 78) which were all conducted with zero confining pressure. Ultimate strengths vary from 40 kPa to 97 kPa with densities varying only 18% (plus or minus 26 k g / c m 3). The lower values of strength do correspond with the lower densities and the highest values with the highest values of density. The series 78 tests have not been corrected for diameter changes in the samples as measurements on diameter changes were not recorded. The volumetric change, A V, in Table 1 is shown as two values for all tests except for number 770410-8 which was a virgin compression test (pure hydrostatic stress). The number preceding the " / " gives the volume change due to cell pressure P (virgin compression) and the value following the "/'" gives the volume change due

R.M. Lang, W..L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199

197
I i I i I i

ok
4O 30

'

'

'140 I

i I i itj

0.284 to 0.376

Z
420 - ~ V C L

/c,
.=

~
.c

SL

E 380

IO
/ /

// 0.210

0.178~ --t
j J

J
F t

,---~"~
0

20 volumetricchange, (ml) AV
IO

30

360 1

t I v v,hl
t0

I , 1,'7

naturalIogahthm the sphehcal pmseum, In P(kPa) of


Fig. 6. Initial volume minus the volumetric change, V-ztV, versus the natural logarithm of the spherical pressure, P, first season of Alta tests (77 series). C~ is the extrapolated volume at unit pressure on the critical state line and V~ is the extrapolated volume at unit pressure on the virgin compression line. Lines are best fit.

Fig. 5. Virgin compression lines (VCL) of the samples in the pressure-volume plane. Sample densities are indicated. Test temperature was - 2 C.

to the "true stress" at 15% strain. This second value of AV is also given in Figs. 2 and 3 and coincides with the 15% strain (vertical line). Fig. 5 shows the virgin compression lines (VCL) of the samples in the P-Vplane, for the range of snow densities tested. The significance of the virgin compression lines is the ability to predict (extrapolate) the volumetric response of the snow under an increase in pressure for a given initial density range. The typical soil response is a linear relationship between the natural logarithm of increasing hydrostatic pressure and decreasing void ratio (see Chen, 1975, for example). Fig. 6 shows the relationship of initial volume minus the volumetric change, V-AV, versus the natural logarithm of the spherical pressure, P, containing the critical state line and the virgin compression line using all possible data accumulated during the first season of Alta tests (77 series). The best fits for the virgin compression line and critical state line may be interpreted to display linear response. In considering the VCL, the implication is that if the snow is at

569

374

A > e _E
>o 3~ 384

sy
o ~ o VCL
I v t "~t

58==

394

20

40 P(kPe)

60

spherical pnmaure,

Fig. 7. Projection of the critical state line (CSL) and the virgin compression line (VCL) in the P - V plane using the data from the first season Alta tests (77 series).

198

R.M. Lang, W..L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199

some current pressure on the VCL and it is the largest value that has been applied to the snow, then upon the application of a pressure increase the snow will load along the VCL. If the pressure is removed, the snow will unload along some other path under the VCL at a reduced value for the volume, but upon reapplication of a pressure which exceeds the initial value, the reloading response should intersect the VCL and proceed to respond linearly along the VCL. Ideally, the VCL would be intersected upon reloading at the same point of unloading, but realistically the intersection will be made at an decreased value of volume and a corresponding increased value of the natural logarithm of P. In order to describe the general state of stress of a snow sample in the triaxial test, it is allowable to postulate that in a three dimensional space of the spherical stress P, the second invariant of the deviatoric stress H'~, and the void ratio ev or volume V, a unique state boundary surface can be constructed from the triaxial test data (for a complete discussion see Roscoe and Burland, 1968 or Chen, 1975 ). Admissible state points lie below the surface and represent the initial linear region of material response. Points on the state boundary surface represent hardening and inadmissible points lie above the surface. The critical state line is a line on this state boundary surface where unlimited distortional strain may occur without an accompanying change in volume or state of stress. The critical state line projection into the lnP vs. V - A V plane for the snow samples is depicted in Fig. 6. The depicted parameter, C1, is the extrapolated volume at unit pressure on the critical state line and the parameter V~, is the extrapolated volume at unit pressure on the virgin compression line. Fig. 7 shows the projection of the critical state line (CSL) and the virgin compression line (VCL) in the P - V p l a n e using the data from the first season Alta tests (77 series).

technique for testing snow was directly adapted from methods that are used to obtain the mechanical properties of soils and other granular media subjected to three dimensional stress states. Initial densities were low to intermediate ( 170 kg/m 3 to 376 kg/m 3) and significant volume changes occurred under the applied loads. The stress-deformation curves exhibited an initial linear portion of the curve to internal failure, i.e. loss of cohesion, and the plastic hardening portion. A dependence on confining pressure is apparent. Ultimate strength increases from approximately 60 to 130 kPa at zero confining pressure to values in excess of 160 kPa at higher pressures. In granular ice, the role of confining pressure is to inhibit the growth of cracks and communication between cracks (Kalifa et al., 1992 ). The mechanism in snow is probably similar, i.e., the confining pressure may serve to repress the failure of the intergranular bonds in snow. In plotting the virgin compression lines of the snow which was tested, results exhibited the typical soil response of a linear relationship between the natural logarithm of increasing hydrostatic pressure and decreasing void ratio or volume. Although these data do not provide a statistical basis for the mechanical properties of snow, they do serve to provide guidelines for the load response of dry snow. In addition, mechanical properties for use in modelling can be easily inferred for comparable loading rates.

References
Andresen, A. and Simons, N.E., 1960. Norwegian Triaxial Equipment and Technique. Norwegian Geotechnical Institute, Publication #35. Andresen, A., Bjerrum, L., Dibiago, E. and Kjaernsli, B., 1957. Norwegian Geotechnical Institute, Publication #21. Bennett, W.D., 1973. Documentation of Snow Characteristics for Over-snow Vehicle Validation Tests. CRREL TN. Brady, J.E. and Humiston, G.E., 1978. General Chemistry Principles and Structure. Wiley, Chichester, 2nd Ed. Chen, W.F., 1975. Limit Analysis and Soil Plasticity. Elsevier, Amsterdam. Drucker, D.C., 1967. Introduction to Mechanics of Deformable Solids. McGraw-Hill, Englewood Cliffs.

3. Conclusions Constant strain rate triaxial tests were performed on dry, naturally occurring snow. The

R.M. Lang, W.L. Harrison / Cold Regions Science and Technology 23 (1995) 191-199 Harrison, W.L., Berger, R.H. and Takagi, S., 1975. On the Application of Critical State Soil Mechanics to Dry Snow. CRREL 1R-473. Kalifa, P., Ouilion, G. and Duval, P., 1992. Microcracking and the failure of polycrystalline ice under triaxial compression. J. Glaciol., 38 (128): 65-76. Kinosita, S., 1967. Compression of snow at constant speed. In: H. Oura (Editor), Physics of Snow and Ice: Int. Conf. on Low Temp. Sci., 1966, Proc., Vol. I, Pt. 2. Inst. of Low Temp. Sci. Hokkaido Univ., Sapporo, pp. 911-927. Naito, M., 1973. Mechanical Characteristics of Snow II. (Report # 5 ) TRDI, Japan Defense Agency, Experimental Section, Sapporo Experimental Station. Ramseier, R.O., 1963. Some physical and mechanical prop-

199

erties of polar snow. J. Glaciol., 4(36). Roscoe, K.H. and Burland, J.B., 1968. On the Generalized Stress-Strain Behavior of Wet Clay. In: J. Heyman and F.A. Leckie (Editors), Engineering Plasticity. Cambridge Univ. Press, London, pp. 535-609. Roscoe, K.H., Schofield, A.N. and Thurairajah, A., 1963. An evaluation of test data for selecting a yield criterion for soil. Spec. Tech. Publ. ASTM, 361:111-128. Shinojima, K., 1967. Study on the Viscolastic Deformation of Snow. In: H. Oura (Editor), Physics of Snow and Ice: Int. Conf. on Low Temp. Sci., 1966, Proc., Vol. I, Pt. 2. Inst. of Low Temp. Sci. Hokkaido Univ., Sapporo, pp. 875-907.

You might also like