Plate tests for the measurement of modulus and bearing capacity of gravels B P WRENCH" (Member)

Synopsis Because they are difficult to sample and test in the laboratory, gravels are best

tested in situ. Plate bearing tests are widety used to measure the modulus of soils and weak rocks and are also suitable for gravels.

Equipment and test procedures have been developed to allow the rapid testing of gravels. The widespread use of trial holes and test pits for foundation investigations in Southern Africa provides the opportunity to perform these tests on a routine basis.

The results of 175 plate bearing tests in gravels are analysed In this paper. The soil modulus has been determined from each test using both a simplified elastic analysis and a rectangular hyperbola method. Relationships are presented between modulus and observed consistency and also plate bearing capacity and consistency.

Samevatting As gevolg van hulle aard is dit die beste om die ingenieurseienskappe van gruis

deur middel van in situ toetse te bepaal. Plaatdruktoetse word algemeen gebrulk om die modulus van grond en rots te bepaal en is ook geskik vir gebruik in gruis.

Apparaat en toetsmetodes is ontwikkel vir die vlnnige toets van gruise. Die algemene gebruik van toetsgate en putte vir fondamentondersoeke in Suid-Afrlka maak dit moontllk om hierdie toetse op 'n roetine basis ult te voer. ·Director. Steffen. Robenson and Kirsten (Civi l) Inc

Resultate van 175 plaatdruktoetse Is geanaliseer. Moduluswaardes vir elke toets Is bepaal deur gebruik te maak van 'n vereenvoudigde elastiese metode en van die reghoekige hlperbool metode. Die verband tussen modulus en plaatdravermoe teenoor die dlgtheid van die gruis word aangetoon.

Introduction Most foundation designs on granular soils are determined by

acceptable settlement limits. As a result an important part of a foundation investigation is the 'determination of the compressibility characteristics of the subsoils.

It has long been reCQgnized that in situ testing prQvides a better assessment 'Of SQil cQnsQlidatiQn, characteristics than small labQratQry tests. Of the in situ tests available, plate bearing tests are widely used and numerQUS results are reported in the literature, mQstly fQr fine grained homQgenous SQils and weak rocks.

10 '.' .. Southern Africa, extensive shallow gravel horizQns occur as pedogenic gravels, pebble markers, dQIQmite rubble, and river or sea terrace gravels. Because they are dense,. these materials are 'Often the mostsuitable hQrizon on "l'hich tQ fQund structures that impQse light and

. medium IQads. ' Since gravels are difficult to sample it is practically impossible to

measure their cQmpressibility ·!h the labQratory and in situ testing is the 'Only way to 'Obtain reaSQnable design parameters. Plate bearing tests allQw the mass cQmpressibility :characteristics tQ be measured. These tests are used by my firm as a rQutine prQcedure in fQundatiQn investigatiQns.

This paper presents the results 'Of plate tests on gravel horizQns at 39 sites in the Transvaal and Orange Free State. The test equipment and prQcedures are described and methods of analysis are presented . A relatiQnship between Young's Modulus (E) and cQnsistency has been prQduced from the test data. A relationship between plate bearing capacity and plate cQnsistency is alsQ suggested. These relatiQnships are intended tQ prQvide initial estimates 'Of compressibility and bearing capacity during foundation invest igat iQns in gravels and to assist in the subsequent scheduling of in situ testing at the site.

Equipment and test procedures

Gravel horizons are usually hiQ'hlY variable in extent and cQnsistency . This is particularly so in pedQgenic horizons where the amount of gravel and its cementing may vary appreciably over short distances.

Gravels generally consist of hard angular fragments in a matrix usually

' Director, Steffen. Robenson and Ki rsten (Civi l) Inc

DIE SIVIELE INGENIEUR in Suid-Afrika - September 1984

Brian Wrench 'Obtained his BachelQrs degree in Civil Engineering IrQm the University 'Of Natal. FolfQwing a periQd 'Of site work with Frankipile and Soiltech in Durban he spent two years in the United Kingdom with the firm Soil Mechanics LId. In 1975 he returned to South Africa and worked for Soiltech. He jQined Steffen Robertson & Kirsten in 1977 and is responsible for the Soils and Foundations Division in Johannesburg.

of clay, silt 'Or sand. In order tQ reduce the effects of individual large fragments, tests should be perfQrmed 'On as large a scale as possible. Plate load testing is a gOOd. way to measure the compressibility parameters of these materials, since the plate size can be varied to take account of the size of the largest fragments and the req.uired stress range.

Widespread use of large diameter augered trialhQles and test pits for foundation investigations in Southern Africa makes routine in situ testing 'Of gravels practicable. Plate equipment was therefQre develQped to test in auger holes and test pits. This in situ testing is easily carried out during foundation investigations at little extra CQSt. Because of the variability of gravel horizons, frequent testing at each site is necessary and c'Onsequently a Simple test procedure was developed.

Due tQ the confined working space the dimensions and mass 'Of the equipment is limited to allow 'OperatiQn by two men.

The equipment comprises a light hydraulic jack with attachments to connect circular plates at both ends. A series of intermediate connections and threaded turnbuckles provides fQr adjustment of the equipment length to suit the test position. A separate pump connects to the jaCk by long hydraulic hosing. Pressure gauges are fitted both t'O the pump and the jack. For tests in deep trial hQles a prQfiling chair has been modified to SUPPQrt the equipment. A photograph 'Of the plate bearing equipment is shown in Fig 1.

Tests were carried out both horizontally and vertically. The majority of the tests were horizontal since the necessary reaction was easily provided by the opposing walls of the trial hole or pit. Vertical tests require an adequate reactiQn which is provided either by jacking against a heavy earthmoving plant, or against a cross beam held by shallow flight augers turned into the surrounding soils. Because it is difficult tQ prQvide reaction at depth vertical tests are seldom carried out in deep pits or trial holes .

Stiffened steel plates 20 mm thick and 100, 200 and 300 mm in diameter are used. Plate displacements are ideally measured directly, either with an independent frame carry ing dial gauges, or by settlement sensors passing through the plate and anchored intQ the soils behind the

Fig 1: Photograph of plate equipment.

429

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

· .. Cement marketing IS rising to meet every challenge

with a solution. From the bonding together of

giant Roman aqueducts to casting 50 storey miracles of modem technology - cement has proved to be truly an indispensible product.

But this solid reputation doesn't mean we can rest on our laurels. The future is within our grasp. Moulding it requires the right balance of innovative technology and personal involvement.

That's why Blue Circle have instituted intensive on-going research and development and marketing programmes. We've already invested millions of rand into increasing production capabilities for the future.

The programmes are also being supported by a vigorous new

marketing approach. By thoroughly investigating the problems and needs of all users, large and small, we have been able to identify and provide some much needed facilities. Such as the 24 hour service at Lichtenburg. Traffic flow has been improved at all Blue Circle Depots resulting in improved turn-around times. These are just the beginnings of a number of service innovations you can expect from Blue Circle.

We know that the goals we set ourselves today can only help build dynamic industry of tomorrow.

O BLUE CIRCLE CEMENT

~Kuper Hancb 4780

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

plate. The settlement sensor method has been described by Burland (1970) and comprises long thin aluminium pins, projecting through holes drilled in the plates and anchored into prebored holes in the soils behind or below the plate.

Provided the pins are anchored far enough behind the plate, they are not affected by stress changes during the test, and therefore allow direct measurement of the plate movement. This method is used in homogeneous soils and whilst it gives good results, it is both time consuming and difficult to execute. In gravels however, the method is found to be impractical due to difficulties in installing the pins to the required depth.

In the vertical tests, plate displacements were measured with a triangular frame fitted with two dial gauges. The frame is supported on pins driven into the surface soils sufficiently far from the plate to ensure that the frame is stable.

For horizontal tests in trial hole and test pit applications, independent dial gauge frames are impractical and likely to be disturbed during the test. Plate displacements are therefore measured by a single dial gauge mounted on the jack. The dial gauge records the ram movement and this arrangement measures the total movement of both plates. Plates of equal diameter are used and the average deflection is calculated.

All of the horizontal tests reported in this paper, were carried out on soil at its natural moisture content. For these tests, flat, parallel surfaces were trimmed on opposing faces of the trial hole or pit. It was seldom possible to achieve a smooth surface due to the larger fragments in the horizon , and care was needed to ensure that large fragments did not protrude from the surface. After this preparation a layer of plaster of Paris was placed between the soil face and the plates and a small load applied to support the weight of the equipment. Testing commenced when the plaster of Paris was hard.

A load increment was applied to the plates and the deflections measured. The load was maintained and time-deformation readings were taken . The results were plotted on site and were used to determine the time intervals for additional load increments to be used during the remainder of the test. Since the gravels tested were all partially saturated, the deformations under load occurred quickly. In most of the cohesionless horizons load increments were applied at approximately 10 minute intervals. Load was cycled at least once (usually at plate pressures between 100 to 400 kPa) . An unloading cycle was measured at the end of the test. Depending on the materials and the plate size, tests were usually taken to maximum pressures of between 1 200 to 2 000 kPa. Where possible a small block sample of matrix material was recovered from close to the test position for laboratory determination of density, moisture content, and degree of saturation.

Ana!y~is of plate tests In gravel horizons The 'plate loading tests have been interpreted using elastic theory.

Numerous solutions have been published .to compute the settlement of loaded, rigid, circular plates resting on a semi-infinite, isotropiC medium. For the simple case of a ci~cular rigid plate resting on a homogeneous medium, the Young's Modulus (E) is given by:

~' . (1-1'2)71ro

E = 21' (1 )

where:

\' = Poissons Ratio (for drained tests on soils usually between 0,2 and 0,3) r = plate radius a = plate stress /' = plate deflection.

This equation is commonly used to analyse plate load test data (Burland and Lord 1969, Lawrence 1977).

More rigorous analytical techniques are available. Poulos and Davis (1974) present solutions for homogeneous single and two layer mediums. Carrier and Christian (1973) present solutions for the case where stiffness increases with depth. Due to the complexity of these solutions, and the difficulty of determining parameters at each site, these solutions are seldom used in the routine analysis of plate test data.

To assess the applicability of ttle simplified model and Eqn 1 to the plate tests in gravels, consideration was g iven to how the requirements of the theory are met in practice. The factors considered were:

• Homogeneity For the plate diameters commonly used significant stress changes

DIE SIVIELE INGENIEUR in Suid-Afrika - September 1984

are only induced to about 0,2 to 0,6 m behind the plate. In most soil horizons consistency and stiffness vary with depth even over relatively small distances. Greater variations are usually encountered in gravel horizons, due to varying amounts of gravel and possible cementing . Horizontal variations are generally less marked . For the horizontal tests a measure of the homogeneity of the horizon was obtained by comparing the settlement of each plate into the face at the end of the test. In most of the tests, provided the test locations was carefully selected and the test surfaces well prepared, these settlements were found to be similar.

The size of the largest fragments in the horizon relative to the plate diameter is also important. Large fragments that occur in the zone of loading, result in a non-uniform stress distribution and the elastic theory does not apply. As a general rule tests were not carried out where the largest fragments visible in a horizon were greater than 0,3 times the plate diameter.

Thus although gravel horizons are not homogeneous the effect on plate tests can be minimized by careful selection and preparation of test locations. It is also preferable to carry out a number of tests in each horizon to establish average characteristics.

• Isotropy Anistropy in gravel is caused both by the stress history of the

deposit and by possible orientation of the hard gravel fragments. While some orientation of fragments is observed in gravels (and also other soil particles) , particularly in pedogenic materials, the effect on the test results is probably small. In such cases plate tests should preferably be carried out both horizontally and vertically.

• Boundary conditions The elastic solution used in the analysis of the results considers a

rigid plate resting on the surface of a semi-infinite medium. Vertical tests at ground surface satisfy this condition provided that either a much stiffer or less stiff horizon does not underlie the plate at shallow depth. Horizontal and vertical tests in pits or trial holes may not satisfy this requirement however, and boundary conditions may significantly alter the stress distribution beneath the plate. In a test pit, for example, the length of the pit may be sufficient but tests should not be carried out too close to the surface or floor of the pit. As a general rule tests are not performed within 1 m of the top or bottom of a test pit or trial hole.

Boundary effects may be more pronounced however for a horizontal test in a trial hole. The depth of the hole is usually very large in relation to the plate diameter. whereas the whole diameter may not be. The depth of embedment of the plate in the trial hole wall also varies. In this case the boundary conditions may significantly change the stress distribution under the plate from that assumed for a plate resting on a flat surface. The magnitude of this effect is of interest and computer models were therefore generated forthe two cases of a rigid plate resting a flat surface and two plates partially embedded in opposing walls of a circular trial hole. A tWO-dimensional plane stress finite element analysis was used.

The model of the plates in the trial hole is complicated since for an elastic analysis, tensile forces are generated around the centre of the trial hole between the plates. Since gravels are predominantly cohesion less, the tensile stresses are relieved by displacements and the computer analYSis does not accurately model the actual conditions. An apprOXimation of the mechanism was obtained in the model by assuming a tension crack across the trial hole midway between the two plates. Analyses were carried out for the three plate sizes commonly used and for a 760 mm diameter trial hole. All plates were loaded to the same contact pressure. \

For the geometry analysed the analyses show reduced plate settlements in the trial hole due to the confining pressure of the surrounding soil. For a given trial hole diameter the effect becomes more pronounced as the plate diameter increases. A reductio'n in seltlement of approximately 30 per cent was obtained for a 300 mm diameter plate in the 760 mm trial hole compared to the seltlement of the plate on a flat surface.

The above analysis is considered to overpredict the effect of trial­hole shape on reducing plate seltlements. While the magnitude of the influence may have been overpredicted, it is nonetheless good practice to limit the plate size to less than one third of the trial-hole diameter.

431

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

From the above is evident that the elastic method of analysis .is not theoretically ideal- particularly for horizontal tests in trial holes. Due to the simplicity of the analytical procedure the method is, however, widely used to analyse plate test data. For the tests reported here the method is considered sufficiently accurate to warrant its use, provided the precautions mentioned above are observed.

The information obtained from the plate tests can be further analysed by using an alternative method. Researchers have found that load test data on soils can be represented by a rectangular hyperbola (Kondner 1963). Transformation of a hyperbolic function results in a straight line. This of particular use in interpreting load data since, while it is difficult to deduce parameters from a hyperbolic curve, it is relatively easy to use straight lines and intercepts.

The plate tests described in this paper also produced hyperbolic stress strain relationships. Following the reasoning given by Kondner the hyperbolic relationship between plate loading (a) and plate settlement (p) is:

a= p

a+bp

Where a and b are material constants.

(2)

The transformed plot of pia versus p yields a straight line as shown in Fig 2.

The ultimate value of stress is obtained from the limit of Eqn 2 as the displacement becomes very large.

ie a ult = lim a = 11b p- 00

(3)

Thus the ultimate plate bearing stress is given by the inverse of the slope of the str<:light line on the transformed plot.

Differentiating. with respect to displacement and evaluating the derivative at p = 0 gives

(da/dp)p=o=1Ia (4)

The'value ofa is given by the intercept of the straight line on the pia axis. It may be noted that 11a has the units of the coefficient of subgrade reaction (k).

Using Eqn 1 and substituting 11a fa r alp we have fora drained test that

E= 1,4. _r_ a+bp

(5)

In practice it is often found that a straight line is obtained only on the transformed plot for large deformations which correspond to the rnode.rate and high stress ranges in the test (Kondner 1963). In assessing the modulus value for plate tests however, it is preferable to evaluate the results overs a lower stress range which corresponds to a reasonable factor of safety against bearing failure of the plate. Comparison of modu.lus calculated using Eqns 1 and 5 shows that the hyperbolic transformation methqd yields higher modulus values than by direct measurement. The main advantage of hyperbolic representation of the data is that it greatly simplifies analysis of the test results. The technique is easily computerized to allow a best fit line to be fitted to the transformed data. In this way consistent interpretation is assured. In the interests of ensuring interchangeability of the two methods, Eqn 5 has been modified to bring it into line with modulus values obtained from Eqn 1. The following relation is proposed:

E = 0,95r a

(6)

This equation yields similar modulus values to those obtained from high quality tests using Eqn 1.

The test results The results of 175 tests on gravel horizons have been analysed using

the method described above. The majority of the results as for horizontal tests carried out in test pits and trial holes. All tests were carried out on soils at natural moisture content corresponding to low degrees of saturation and all materials were essentially cohesion leSs and relatively free draining.

Typical results of nine tests in gravel are shown on Fig 3 which shows plate stress plotted against deflection. At the start of a test a bedding in effect is occasionally found. This is caused by seating of the plate into the plaster of Paris and disturbance of the test surface during trimming.

432

b = TAN A

E=f(l/a)

O'ULT= lib

PLATE SETTLEMENT,p

Fig 2: Transformed hyperbolic representation of plate stress-settlement (after Kondner 1963)

With careful preparation this effect is usually minimal. Thereafter the curve varies in slope. At very high loads bearing capacity failure may even be approached.

In analysing the test data the deformation modulus was first calculated in the conventional manner from the load-deformation curve using Eqn 1. Since the plot is non-linear the modulus value is dependent on the choice of the stress range. All results were calculated over the range from assumed overburden pressure, Po. to Po + 200 kPa.

The test results were then replotted as transformed hyperbolic functions. Transformed plots of the nine typical tests are given in Fig 3. These show for each test that, apart from the initial readings, the results plot as straight lines. For large strains these load-deformation results may therefore be modelled by a rectangular hyperbola. Analysis of all the plate load tests available for gravel horizons shows that for more than 90 per cent of the results straight lines could reasonably be fitted through the transformed plots of the data. Those results that do not plot as straight lines were found to give slightly concave plots on the load­deformation graph, indicating unusual settlement characteristics. It may be that the presence of larger fragments behind the plates influenced these results and in all probability these tests are not representative. It seems therefore that plate load test results in the gravel horizons follow a hyperbolic function.

Modulus and plate bearing capacity values have been evaluated for all of the tests, using the transformed plots and Eqns 6 and 3 respectively.

Values of mOdulus obtained from the two analytical methods for the examples are given in Fig 3. The values are similar and for design the value of modulus obtained from the hyperbolic method is usually used.

Interpretation of the results In many instances it is useful to have an appreciation of the

approximate stiffness of a soil horizon during the early stages of a foundation investigation. Experienced engineers and engineering

. geologists build up this appreciation for soils and rocks from the results of laboratory tests and the observed behaviour of structures founded on these materials. Due to the difficulties of testing gravels in the laboratory, the same level of confidence is often not felt for these materials.

As an aid to the assessment of the engineering properties of gravels the results of the plate tests have been compared with the visually observed consistency. The consistencies of the gravel horizons at each test site were recorded using the procedures recommended by Jennings et al (1973). In gravels it is difficult to accurately assess the mass consistency using this system and the tendency was to record the consistency of the matrix. This could have resulted in an underestimate of the true consistency of the gravel mass. In a recent paper Kirsten (1982) presents qualitative descriptions to assess the consistency of detritus materials. These descriptions compliment the original work by Jennings et al.

Average modulus values plotted against observed conSistency are presented in Fig 4.

The modulus values for a given consistency show a wide scatter, often up to one order of magnitude. This scatter is probably due to

DIE SIVIELE INGENIEUR in Suid-Afrika - September 1984

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

::::

z o f­OJ W ..J t.. W

" W f­a: ..J a.

u

10

I::'

o

I.?

I G

2 [) r-----j-----t-

-----] MOOU wS FRaoI CURVE

E : 3,8P#o ]I( E : J,8MPo

)t E • 2"P,-MP_O ___ --J

MOOULUS F'RCW CURVE

E ;: 13 Mlo ]I( E:16MPo )i E:r1MPo

24 L'c-,--~2..Jl,cc"c- .Hh~ t;f)0 80:0

F'LATE PPES;URE, cr ( ~ Pa

., . 1:

~ E

t> Q.

,-. "-1:

E E ,~

t> Q.

II) MEDIUM DENSE

z

:1' ' [) ------,--------'---,_._--

MODUWS FRQMCVf!l/ES

E ;: 71 Wo )( E=40liFa )t E :37..,.

6 0';------""30"'''''"----""''6-0''''0---.----- gOO 1200

FLRTE PP.ES';URE. (J' ( 'P. I

Fig 3: Examples of plate bearing test results.

DIE SIVIELE INGENIEUR in Suid-Afrika - September 1984

, E E

t> "-

1500

1111 DENSE

10tl

90

Stl ~--- ~

.,-V ~tl --1--- ~-

-;;7 --; 50

V po--- ,A(

.-- /' 50 V"

40 /' ~~ ~ I----<-.---

---0---30 .---20

0

~1..!.!~Fa!f=~

Hl 0 . E = 3,4Wta

X E'I,UF. )( E • 3,DMPa

0 2 ~ 6 8 10 12 14 16 18 2e

p ( rom 1

2[]

/ V

16

V ./'

/ 0

12 -./' //

L ~ --------" B ~-

~ ~

~

[J 0

4.0

3.5

3.0

2.5

~ 2.0

1.5 :.-----

o 0

1.0

. 5

0·~.tl

fIIXlULUS FQ3: F=;} . E"2 _

X E·18 _

"Jt. E:I~MPa

4 8 12 16 20 24 P { mm )

....--~

~ ~. ____ r"

~ V .---------~ f---"-

~ ~ -----~

~ ~

~

MOOULUS Fgj F: ~ . E _62..." X E -42'" "Jt. E .44..-a

< 0 1.0 I_S 2.0 ,.5 3.0 3.5 4.0 4.5 5.0 P ( mm }

433

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

Piling and special founCiations.

unh three 808A

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

inaccuracies in assessing the mass consistency and the variability of gravels. Curve fitting techniques were used to obtain a best fit line for the data set. This has been achieved by expressing the observed consistency as an approximate solid density. An exponential function is found to fit the data best and yields a straight line relationship on the semi-log plot. A correlation coefficient of 0,75 is obtained; this is considered to be good forthe type and nature of the data. The best fit line is plotted on Fig 4 and provides an estimate of the average modulus for a given consistency.

Although Fig 4 contains data only from gravel horizons a similar relationship exists for other cohesionless soils. Test results for medium and coarse sands have been checked against the data for the gravels and show similar values and trends. Analysis of test results for soils which contain significant percentage of silt and clay, however, show that apparent cohesion influences the modulus an-d bearing capacity. The degree of saturation of these materials is therefore of prime importance, I n using plate test results in design, adequate allowance must be made

4tXJ

3tXJ

200

D ~ •• ( .. (

· 0 'U

/00 90 80

• S- eD /

&. 0 II .; ,...

// ll.

~ ./ ll. / J\'tl

• V 0

~tr •

10

60

50

40

30

/ Oil'

:/ • x • • 0

~

/ll.e6 . ll. .

'V ll.

/: ~ 0

/ I)"! • / ~

/ . /~ ll.

20

tJ Q.. ~

---~ io

~ 9

~ 8

~ 1

~ 6

~ 5 ~ '<t Q:: 4 ~ ..,. •

3

• .

for possible long term changes to the moisture content of the soil. Ultimate bearing capacities calculated using the hyperbolic method

are plotted against observed consistency in Fig 5, together with the best fit line through the data, The exponential function gave a correlation coefficient of 0,67, A rough check on the accuracy of the bearing capacities predicted by the method has been obtained from a few tests on very loose and loose horizons. Some of these tests were taken to failure and the actual failure loads compare well with the predicted values. These results tend to confirm that the method is reasonable for gravel horizons,

The ultimate bearing capacity of a shallow circular footing may be calculated using the well known relationship:

Pull = 0,4 N )'l' 0 + 1,3 eNc + Po (Nq-1)

where )' = density of subsoil 0= footing diameter e = cohesive intercept of soil

/ ~/o

/(

(7)

./ 0

/00 :0 (~

0

ll.

p

~ KEY

2

VERY LOOSE

0,45 ,

0

LtXJSE' MEDIUM DENSE

CONSIST"ENCr

0,65

• WITWATERSRAND

0 WEST RAND

.0. EASTERN TRANSVAAL

)( ORANGE FREE STATE

-- EXPONENTIAL CUftYE .. IT

DENSE VERY DENSE

VERY SOFT RXK

APPROXIMATE SOUD DENSITY %

DIE SIVIELE INGENIEUR in Suid-Afrika - September 1984

Fig 4: Drained modulus versus ob­

served consistency for gravels

435

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

Po = overburden pressure and N )" Nc and N q are bearing capacity factors .

Values of the dimensionless N y and Nq factors have been shown to vary according to the angle of internal friction (0) and are widely reported in the literature. For cohesionless materials c is zero and the bearing capacity is governed only by 0 , the density of the material and the embedment of the footing .

In gravel materials is the angle of internal friction may be expected to vary widely depending on the size, packing and orientation of the gravel fragments and any cementing between the fragments. For this reason a is not accurately known for gravels and the estimation of ultimate bearing capacity for footings in these materials is problematic. For design purposes bearing capacity factors appropriate to coarse sands are often used.

Back analysis of plate test data could provide an indication of the N " and Nq factors for gravel s, provided that ulti mate bearing capacities are

4Cj 0

30, 0

10,0 9,0 8,0

r,o t+O

'i0

•

•

. . " •

known . Ind icat ive values of Ny for sands have been publ ished by Terzaghi (1943) and Meyerhof (1956) and vary between about ~U and 200 depending on the angle of internal friction and relative density. Values of Nq were also found to be similar to N y for loose to medium dense sands .

An attempt was made to back analyse values of N ). and Nq for gravels based on the tests reported here by us ing the assessed bearing capacities given by the best fit line on Fig 5. These tests have mostly been carned out horizontally in test pits and trial-holes which results in a problem in deciding on the boundary condit ions to be used and in particular the embedment of the plate. Rigorous analysis of these conditions is considered to be inappropriate in view of the nature of the bearing capacity data.

. To foUow an assessment of the magnitude of the factors a major SimplifYing assumption has been made that for these tests the plates are only partial ly embedded and consequently, that both the Nand N t . E 7 . Y q erms In qn contribute to the bearing capacity of the plate.

• •

/ - - - I-- 7 ./

• / /

• ./

/ • : / •

4P

~O

436

2,0

1,0 0,9 0,8 o,r 0,6

•

.

./ /

O~

0,4 V

O,J

0,2

0,1 ~RY LOOSE

0,45 I

,

./ /

/0

LOOSE

0," I

.. • :/V

• V • · ~/

, L- •

./

• •

I

: f • • · •

• KEy --

--EXPONENTIAL CURVE FIT

MEDIUM DENSE DENSE

CONSISTENCY 0,65 0,75 , I

APPROXlliATC SOL 10 DENSffY %

VERY DENSE

0,65 ,

VERY SOFT ROCK

0,95 I

Fig 5: Inferred ullimale bearing capa­city for 200 mm diameter plate . versus observed consistency.

DIE SIVIELE INGENIEUR in Suid-Afrika - September 1984

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

Calculations of the Ny and Nq factors have been made for various observed consistencies and show values of about 70 for very loose gravels, increasing uniformly to about 600 for very dense gravels. These gravels are somewhat higher than those reported for sands. The factors will be significantly altered by changing the assumed boundary conditions and the above values are presented as indicative parameters only. I t is suggested however that the plate test method does offer scope for the prediction of the bearing capacity factors although for this purpose tests should preferably be carried out verticall y and in large excavat ions.

Conclusions and summary The results of 175 plate bearing tests carried out in fine and medium

gravels in South Africa are reported . Test equ ipment and procedures have been developed to allow rapid testing in trial holes and test pits, thus allowing a number of tests to be ca rried out in each gravel horizon at a

site. The modulus of the materials is calculated from the load-deformation

plo ts using a simplified elastic analysis. By transforming the test results the estimated bearing capacity of the plate in the gravel horizon can be estimated. This method also allows the modulus to be estimated.

Plots of modulus and plate bearing capacity versus observed consistency are presented. Both plots show a wide scatter of results and curve fitt ing tech niques have been used to give a best fit line through the data sets. These relationships are intended to assist in the preliminary design of structures founded on gravel horizons. Actual design parameters should, however, be obtained from addi tional tests at the site.

Acknowledgements

The permission of the Board of Directors of Steffen . Robertson and

Kirsten Inc to publish the data given herein is acknowledged. I have 'had many valuab le discussions with members of the firm . particularly with Dr J Amir and Mr G A Jones and their assistance is appreciated

References

Burland, J B, and Lord. J A. 'The load deformation behaviour of Middle Chalk at Mundford, Norfolk: A comparison between full-scale performance and in situ and laboratory measurements.' Conf on In Situ Investigation in Soils and Rocks, London (British Geotechnical Society, 1969. Burland , J B. Private communications, 1978. Carrier, W D. and Christian. J T. 'Rigid circular plate resting on a non- ' homogeneous elast ic half-space.' Geotechnique 23. No.30, pp 67 -84 . 1973 Jennings, J E, Brink , ABA. and Williams. A A B. 'Revised guide to soil profiling for Civil Engineering purposes in Southern Af rica.' Civil Engr S AIr. Vol 15. No. 1. pp 3-1 2, t973. Kondner. R L. 'Hyperbolic st ress-stra in response: Cohesive Soils.' J Soi l Mech and Found Div, Amer Soc Civil Eng, Vol 89, 1963. Kirsten, HAD. 'A Classification system for excavat ion in natural materials .' ClVit Engr S Afr, Vol 24, NO.7. pp 293-308. 1982.

Lawrence. G J. 'Trench Wall Jack: An apparatus to measu re the equiva le~t elastic modulus of soil.' Transport and Road Research Laboratory. Supplementary Report NO.347, 1977. Meyerhof , G G . 'Penet rat ion tests and bearing capacity of cohesion less soi ls.' Paper No 66. J Soil Mech & Found Div Amer Soc Civil Eng. 1956 Poulos. H G, and Davis. E H. 'Elastic Solutions for Soil and Rock Mechanics.' J Wiley , New York. 1974. Terzaghi. K. Theoretical Soil Mechanics .' J Wiley. New York. 1943.

BIO-FILTER ROTARY DISC EFFLUENT PURIFICATION

PLANTS

LOOK AT THE FACTS a) 50-15,000 Person planls now available b) Low power consumption, c) Small space requirements d) Odour free e) Guaranteed to produce General Standards Effluent

DIE SIVIELE INGEN IEUR in Suid-Afrika - September 1984

.~.

by BECON CONTROL (Pty) Ltd. (Project Engineers) P.O. Box 504, SILVERTON 0127 Tel.: (012) 86-9646 PRETORIA

f) Attention & Maintenance free g) Self compensating to load fluctuations between no load. and

upto 400"/0 overload for short periods h) Complete plant design and installation service available. i) Over 200 plarliS already operating in Southern Africa.

437

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).

Rep

rodu

ced

by S

abin

et g

atew

ay u

nder

lice

nce

gran

ted

by th

e Pu

blis

her (

date

d 20

11).