Large Diameter Plate Tests On Weathered In-Situ Chalk
Large Diameter Plate Tests On Weathered In-Situ Chalk
Large Diameter Plate Tests On Weathered In-Situ Chalk
Abstract
he prediction of the settlements of spread
Keywords: Chalk, in-situ tests, plate-bearing tests, stiffness, Fig. 1. The Chalk outcrop in England, showing the geographi-
weak rocks cal locations of the plate tests.
The settlements of foundations on chalk are best pre- presented in Lord et al. 2002) to develop a new method
of settlement prediction. The purpose of this short paper
dicted using empirical methods (Kee 1974; Lord 1990;
is to provide detail and background to these tests and
Lord et al. 1994), based on the model of behaviour
draw together the results with other long-term testing
developed by Burland & Lord (1970) using data from
(Matthews 1993).
plate tests carried out on the chalk at Mundford (Ward
The objective of the plate-testing programme de-
et al. 1968). Because of the fractured nature of the chalk
scribed in this paper was to obtain realistic estimates of
(Lake & Simons 1975), and difficulties in extrapolating the behaviour of foundations placed on near-surface,
the results of small-diameter plates to large diameter highly weathered but structured chalks of differing
foundations (Lake & Simons 1970), it has long been intact stiffness. Tests were therefore carried out at
recognized that to provide representative parameters, three selected sites in England, near North Ormsby,
loading tests must be sufficiently large. Because of Lincolnshire, Leatherhead, Surrey, and Needham
its non-linear load-settlement behaviour (Ward et al. Market, Suffolk (see Fig. 1).
1968) loading should be controlled, and carried out to
sufficiently high levels to identify yield.
Despite the above, to the authors’ knowledge the Load-settlement behaviour of
detailed results of only four controlled large-diameter the chalk
(defined here, arbitrarily, as greater than 1 m diameter)
loading tests on chalk (the Mundford tank (Ward et al. On the basis of 865 mm plate tests carried out at
1968), two slab tests at Luton (Powell et al. 1990), and a Mundford, Burland & Lord (1970) found that the
test on structureless chalk at Salisbury (Burland et al. average bearing pressure-settlement curves for the more
1983) have previously been reported in the literature. fractured chalks were markedly non-linear but could
The data from an additional nine 1.8 m diameter tests be idealized by the simple bi-linear model shown in
carried out by the authors therefore represent a major Figure 2. This idealization permits the stress-settlement
part of the database used by Lord et al. (1994) (and behaviour to be described using four parameters:
Quarterly Journal of Engineering Geology and Hydrogeology, 37, 61–72 1470-9236/04 $15.00 2004 Geological Society of London
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
Fig. 3. Plan and elevation of BRE. 5000 kN plate load testing equipment.
was read visually. Recorded values were corrected to Optical precise levels using a parallel plate micrometer
allow for the self-weight of the loading column and plate and an invar staff are known (Cooper 1971; Deumlich
components. 1982) to provide a resolution and accuracy better than
In order to assess the magnitude of differential settle- 0.1 mm and have the advantage of good temperature
ment a total of 10 settlement measurements were made stability, and hence this was chosen as the main
using two independent measuring systems (see Fig. 4). measurement system. A Zeiss Jenar Ni 007 level
Four measurements were made using precise levelling of met these criteria. This instrument had a resolution
dome headed nails grouted into the plate to measure of 0.005 mm and a claimed instrument accuracy of
plate movement relative to a set of at least two tempor- +/ 0.05 mm.
ary benchmarks. Six dial gauges were used to measure Dial gauges with a stroke of 25 mm and a resolution
movement relative to a datum bar (see Fig. 4). The use of 0.03 mm were chosen for the secondary settlement
of a sub-plate assembly (Marsland & Eason 1973) was
measurement system. A rectangular scaffold frame fixed
considered, but was rejected because of:
to the ground at four points located approximately 1 m
(a) the increased complexity during installation and the away from the edge of the plate was constructed under
need to provide power on site; the plate loading rig, to act as a reference from which to
(b) doubts over the ability of the system to anchor itself mount the dial gauges. The size of this reference frame
effectively; was kept as small as possible and was located entirely
(c) an inability to determine the stress changes required between the main beams of the loading frame in order
for the interpretation of the results. that temperature and wind effects could be minimized. It
Based on a review of published data the maximum was appreciated that the four datum posts would be
expected value of initial modulus, Ei was about subject to ground movements associated with the plate
1000 MN/m2. Assuming the plate to be rigid a plate settlement. However, an assumption of isotropic elas-
settlement of about 0.1 mm was calculated under an ticity suggested that movements of the reference system
applied pressure of 100 kN/m2. It was therefore necess- would be small relative to the average plate settlement if
ary for the instrumentation systems to have an accuracy the datum posts were kept 2 m away from the centre of
and resolution somewhat better than this. the plate. It was expected that the increase in stiffness of
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
Fig. 4. Layout Plan of plate test showing precise levelling points for plate settlement measurement and layout of dial gauges.
the ground with depth would further restrict the lateral period of at least 5 days. The plate was then lowered
extent of the settlement trough (Gibson 1967). The space onto a liquid plaster base spread onto the set concrete to
available over the plate allowed only six dial gauges to bed the plate well in (Fig. 3). (During dismantling of the
be used, which were positioned on the metal radial plate it was found that a layer of plaster less than 15 mm
stiffeners of the plate. thick formed between the plate and the concrete sur-
faces). A significant gap (generally 100 to 200 mm) was
left between the edge of the plate and the pits, which
Site preparation were generally square or rectangular.
1800 mm dia. plate by the 5000 kN loading frame. It was vertical sets, giving rise to trapezoidal-shaped blocks. All
decided to hold each 100 kN/m2 increment of loading the discontinuities had rough walls and apertures gener-
constant for a period of about 24 hours, rather than ally less than 3 mm. The spacing of the sub-horizontal
applying each load until movements reduced to less than discontinuities increased with depth to a much greater
a pre-determined rate of creep. extent than the sub-vertical discontinuities. Within 1m
Plate settlements were determined immediately before of the ground surface the spacing of the sub-horizontal
the application of each loading increment and again discontinuities was generally less than 50 mm, giving rise
shortly afterwards, once the rate of settlement had to tabular blocks of flaggy chalk (see Fig. 5a). The
reduced sufficiently. Subsequent readings were taken at fracture-block system was found to be loose within the
irregular intervals several times each day. The dial top 2 m, such that blocks of chalk could easily be
gauges were read at regular intervals during each stage removed from excavations without breakage. Below 1 m
of every test, to provide time-settlement data. The last depth the blocks became more equi-dimensional and the
settlement measured before changing the load was used flaggy chalk graded into blocky chalk. In general the
to construct average plate settlement versus applied chalk below the area used for the plate loading tests was
pressure graphs of the type shown in Figure 2. Average found, on the basis of fracture spacing observed on
applied pressure was calculated by dividing the load on adjacent quarry faces and drill-hole logs, to be equiva-
the plate by the plate area (based on its diameter of lent to Mundford (Ward et al. 1968) Grade IV at the
1800 mm). ground surface and Grade III below 2 m. According to
the recent CIRIA classification (Lord et al. 2002) it
would be graded as High density Grade C4/5 near the
The Chalk surface, becoming High density Grade B3 with depth.
Fig. 5. Chalk profiles observed at the three test sites (a) North
Ormsby, (b) Leatherhead & (c) Needham Market.
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
structureless chalk on which the plate was founded was graded as Low density Grade B4/5 near the surface,
dense and clast dominant. becoming Low density Grade B3 with depth.
Two sub-vertical sets of discontinuities, trending NW
and NE, and one sub-horizontal set were observed in the
trial pits. The sub-horizontal discontinuities comprised Results
bedding planes and stress relief fractures. Evidence of
dissolution was observed, particularly along bedding The measurement and interpretation of deformations
planes. In general the bedding planes exhibited wavy caused by plate loading on a weak fractured rock such
walls, with apertures ranging from 0 to 8 mm. In the as chalk is not straightforward.
open sections the lower walls were often found lined (a) Disturbance during excavation (for example using
with putty chalk and in some cases rounded fine gravel hydraulic excavators or piling plant) may loosen the
size chalk fragments were seen. chalk, giving low measurements of stiffness. In this
The upper part of the flaggy chalk generally exhibited study great care was taken to minimize disturbance
a loose fracture block system but not to the same degree through hand excavation down to the test level.
as that observed at North Ormsby. In the lower part (b) Since the material under test is relatively stiff,
of the flaggy chalk and in the underlying blocky any bedding between the plate and the chalk can
chalk, blocks could not be removed without breakage, introduce significant errors. The test procedure
indicating a tight fracture block system. adopted avoided bedding through the use of
On the basis of structure and fracture spacing, the concrete blinding and a very thin plaster filling.
chalk at the Leatherhead site was considered equivalent (c) If the plate is placed in a pit or bore, a depth
to Mundford Grade V (0.5 m thick), overlying Grade IV correction factor is required, and for deeper tests the
(1.0 m thick), and Grade III–II material. According to effect of the gap or of bonding between the side of
the CIRIA classification it would be graded as Grade Dc the plate and the rock may be significant and
near the surface, becoming Medium density Grade B3/4 uncertain, as noted by Lord et al. (2002). The results
and then B3/2 with depth. reported here do not require depth correction since
the tests were performed in shallow pits, and a
significant gap was left between the outer edge of the
Needham Market (low density chalk) plate and the side of the pit.
(Fig. 5c) (d) The plate is to some extent flexible, and because the
ground is variable, differential settlement will occur.
This test site was located in a working chalk quarry on In this test 10 measurements of plate settlement
the southern side of the Gipping valley near the town of (four precise levels and six dial gauge readings) were
Needham Market, Suffolk. The outcrop here is the made to minimize this effect.
Culver Chalk Formation, part of the Upper Chalk and (e) The discontinuous nature of the chalk, and uncer-
is much younger than that found at the North Ormsby tainty with regard to the relative rigidity of the
site. It was characterized by a low flint content. Tests plate, means that interpretation on the basis of
carried out on samples from the site gave a very low simple stress distributions, for example assuming the
density (dry density=1.34 Mg/m3) and an average uni- chalk to be an elastic continuum and the plate to be
axial compressive strength of only 0.9 MN/m2. The rigid, cannot be relied upon. For this reason the
intact rock was easily crushed to a putty between the term ‘average applied plate pressure’ is used to
finger and thumb. Its intact stiffness, derived from describe the degree of loading on the chalk.
unconfined triaxial tests with local strain measurement, (f) The stress paths and strain levels imposed upon
was found to be 7.6 GN/m2 at an axial strain of 0.001% the chalk vary with depth, and distance from the
and 3.3 GN/m2 at an axial strain of 0.01%. centreline. Sophisticated interpretation of the stress-
The rock mass at this site was observed in three deep strain behaviour of the chalk is impossible. Even the
trial pits which were dug beneath the plate loading test use of a sub-plate assembly would not overcome the
locations. It was characterized by vertical and sub- difficulties, because the stress changes at measure-
vertical rough and often open discontinuities, together ment positions cannot be predicted with any confi-
with wavy sub-horizontal discontinuities that showed dence owing to the fractured, discontinuous nature
evidence of dissolution. There were also numerous of the rock.
minor horizontal and vertical impersistent, rough and (g) Settlement of the ground around the plate can affect
relatively tight fractures. The fracture block system at measurement datums, if these are placed too close.
this site was sufficiently tight that no blocks could be Where they are placed at a distance, and even where
removed from the face of the trial pits without breakage. the reference beams are constructed of Invar, tem-
On the basis of fracture spacing the chalk was Mundford perature effects lead to measurement instability. To
Grade IV at the ground surface, and Grade III below avoid this, the data reported in this paper were
1 m. According to the CIRIA classification it would be acquired using two independent measuring systems.
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
In Figure 6 average settlements are given separately pressure (900 kN/m2) for some 40 days. The results, in
for the two independent sets of measurements, made terms of average plate settlement versus the logarithm of
with dial gauges and using precise levelling. It can be time, are shown in Figure 8.
seen that until yield there is good agreement be-
tween the two measurements. The greatest diver- Discussion
gence occurs during unloading – elsewhere, as
expected given the position of the dial gauge datum The general behaviour of large plates and instrumented
posts (Fig. 4), the dial gauge readings give average foundations on chalk, previously observed by Ward
settlements less than those calculated from levelling, et al. (1968), Burland & Lord (1970), Kee (1974),
but the difference is small. Burland & Davidson (1976), and Burland & Bayliss
(1990) for the more weathered near-surface chalk is also
seen in these tests. Initially the chalk behaves in a stiff
and more-or-less linear and elastic manner. Stiffening
as a result of bedding fracture closure (Barton 1986)
was not seen. After yielding, settlements increased
dramatically, and were largely irrecoverable.
Despite founding these large plates at the highest
possible level, the weathered chalk at all three sites
displayed very high stiffness before yield. The average
initial Young’s modulus for the nine tests was 593 MN/
m2, with a range from 328–1381 MN/m2. The linear
range extended to an average plate pressure of between
200 and 400 kN/m2, and settlements of the 1.8 m diam-
eter plate under an average applied stress of 300 kN/m2
in general did not exceed 2.3 mm. These figures suggest
that it should only be necessary to pile exceptionally
heavily-loaded foundations, such as for silos, when
building on structured chalk.
The large-diameter plate tests reported in this paper
were deliberately carried out on chalks with the widest
possible range of intact density, but with a similar high
degree of weathering, in an attempt to allow an assess-
ment of the influence of density variations on the mass
Fig. 6. Plate settlement as a function of applied pressure, compressibility of more-weathered chalks. Table 1 sug-
comparing results from dial gauges and precise levelling. gests that density has relatively little effect on either the
initial modulus (Ei) or the stress at the onset of yield (qe).
Given that the stress levels and strain paths in the As has already been seen (Fig. 7 and Table 1), the
discontinuous chalk mass must remain unknown, a Needham Market chalk was initially very stiff, despite its
simple interpretation of plate test data is necessary. The high porosity. It had the highest average Es value, and
plate-loading test should be regarded as a model footing also the highest average yield stresses (both qe and qy).
test, and average settlement plotted against average The average initial modulus of the hard North Ormsby
applied plate pressure. Stiffness values derived from such chalk was the lowest, because of its loose state. Post
curves are not true moduli but can be used to estimate yield the Needham Market chalk was extremely com-
foundation settlements, and as suggested by Ward et al. pressible, presumably as a result of its low intact yield
(1968) can be regarded as ‘figures of merit’. stress (Clayton et al. 2002), but the North Ormsby chalk
The results of the 9 plate tests are plotted in Figure 7, had a similar modulus to the Leatherhead chalk, despite
using data from the precise levelling. All tests were taken their different densities and intact stiffness.
significantly past yield, and at each site at least one plate The data in Table 1 suggest that, if anything, increas-
location was loaded to an average plate pressure greater ing intact strength and stiffness may be associated with
than 1 MN/m2. From these data values of Ei, Ey, qe and decreasing yield stress (qy). However, intact strength
qy (Burland & Lord 1970) were calculated, and are may be having an important effect on stiffness reduction
shown in Table 1. post yield, since this varies significantly between the
As with previous investigations, and because of time different types of chalk. The ratio Ei / Ey is on average
constraints, there was little opportunity to investigate 6.3, 9.2 and 44.3 for the North Ormsby, Leatherhead
the creep characteristics of the chalk. Most loads were and Needham Market plate tests respectively.
held for only about 24 hours. One plate test, on the Creep is obviously an additional factor that must be
North Ormsby chalk, was however left at its highest taken into account. On the basis of both the plate and
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
Fig. 7. Applied pressure v. observed settlement for the nine 1800 mm diameter plate tests, based upon precise levelling results
(a) North Ormsby results, (b) Leatherhead results & (c) Needham Market results.
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
Table 1. Stiffness parameters (Ei, Ey, qe and qy) derived from the 9 plate tests
North Ormsby
PNO1 328 351 200 299 58 1.3 13.9
PNO2 314 315 250 324 48 1.4 16.2
PNO3 454 409 200 319 68 1.2 11.5
average 365 358 217 314 58 1.3 13.9
Leatherhead
PLE1 463 463 200 363 57 1.3 12.6
PLE2 620 648 200 440 59 0.6 10.7
PLE3 635 635 200 364 69 0.8 10.3
average 573 582 200 389 62 0.9 11.3
Needham Market
PNE1 1381 724 400 576 30 0.6 14.0
PNE2 808 663 300 590 14 0.7 29.1
PNE3 338 371 200 399 13 2.3 41.2
average. 842 586 300 522 19 2.3 28.1
The equivalent Young’s modulus is calculated from the equation for a flexible circular loaded area on an isotropic homogeneous elastic half space:
E=/4. qD/?. (12).I
Where q=applied stress, D=Plate diameter, =Average settlement, =Poisson’s ratio, I=Influence factor (this is a controlled by flexibility of plate,
depth of embedment, partial loading of test pit) (see Lord et al. 1994)
the tank loading tests at Mundford it was concluded that which occurs at more than 2900 kN/m2. The Mundford
no measurable creep occurs in hard, high-density chalks, data suggested that the little creep undergone by the
even under loads as high as 1600 kN/m2 (Burland & Grade III chalk soon ceased, but that the creep de-
Lord 1970). Recent laboratory data using very-small flexions of grade IV and V chalk might ‘in the long term
local-strain measurement (Heymann 1998) similarly be considerably larger than the immediate deflexions’.
show that even for the high porosity Needham Market Burland (1975) reported that the ratio of the long-term
chalk there is insignificant creep below the yield stress, to short-term moduli for these materials was as low as
0.2, emphasising the importance of estimating creep
settlements when designing foundations on weathered
chalks.
Although the primary objective of this work was to
assess the influence of intact stiffness on immediate
settlement parameters, an opportunity did arise to make
some longer-term measurements. Whilst the third site
was being prepared for testing, the loading rig was left in
place at North Ormsby, on location PL3, with a load of
900 kN/m2 applied. The 24 hour settlement under
900 kN/m2 (loaded in 100 kN/m2 increments over a 9
day period) had been 10.3 mm, and this increased, as
shown in Figure 8, to almost 14 mm over the next 39
days. As noted by previous authors (Burland 1975;
Powell 1990), the settlement-time relationship is almost
linear when plotted on a logarithmic basis, suggesting
that settlement of the plate might approximately double
over a 30-year period. These data confirm that creep
must be taken into account when predicting the long
Fig. 8. Results of the long-term loading test at North Ormsby term settlements of foundations on weathered chalks
(average applied plate pressure=900 kN/m2). when under high applied stress levels (> qe).
Downloaded from http://qjegh.lyellcollection.org/ at University of Colorado Boulder on December 7, 2016
P, J.J.M. 1990. Discussion on Session III. Int. Chalk Bedfordshire. Int. Chalk Symposium, Brighton, Thomas
Symposium, Brighton. Thomas Telford Ltd, London, Telford Ltd, London, 327–341.
416–417. W, W.H., B, J.B. & G, R.M. 1968. Geotech-
P, J.J.M., M, A., L, T.I. & nical assessment of a site at Mundford, Norfolk for a
B, A.P. 1990. Engineering properties of Middle Proton Accelerator. Geotechnique, 18(4), 399–431.
Chalk encountered in investigations for roads near Luton.