HKIE Transactions

ISSN: 1023-697X (Print) 2326-3733 (Online) Journal homepage: http://www.tandfonline.com/loi/thie20

Methods Used to Reduce Settlements and Shorten

Construction Time at Macau International Airport

M. Carter CEng, BSc, MSc, MICE

To cite this article: M. Carter CEng, BSc, MSc, MICE (1996) Methods Used to Reduce
Settlements and Shorten Construction Time at Macau International Airport, HKIE Transactions, 3:1,
1-8

To link to this article: http://dx.doi.org/10.1080/1023697X.1996.10667690

Published online: 09 Apr 2013.

Submit your article to this journal

Article views: 14

View related articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=thie20

Download by: [Fudan University] Date: 17 February 2017, At: 06:10

 Methods Used to Reduce Settlements
and Shorten Construction Time at
Macau International Airport
M. Carter", CEng, SSe, MSe, MICE

Owing to its location on an artificial island founded on highly compressible marine

deposits, and the short time-scale required for its completion, control of residual
settlements was of prime consideration in the construction of the runway and taxiways
for Macau International Airport. The paper describes precautions that were taken to
keep settlements after completion of the island within acceptable limits, including
practical experience with ground treatment methods and ground monitoring.

Keywords: Settlements, vertical drains, vibrocompaction, ground monitoring

Background

Macau is situated on the western side of the Pearl River T erminat area
estuary, about 60km west of Hong Kong, and comprises the
peninsular of Macau itself and the islands of Taipa and Sea dyke and wall
Coloane. Since the total land area of Macau is only about
16km', the new Macau International Airport (MIA) had to be
built on reclaimed land. Also, to minimise noise and air traffic -!-
,
problems, the runway and associated taxiways and safety
areas, which formed the majority of the reclamation, were
constructed as an artificial island located between the islands
ofTaipa and Coloane (see Figure 1). Runway

The client, CAM Sociedade do Aeroporto Internacional de

Macau, appointed Shanghai Communications Construction and
General Contracting Corporation to design the artificial island
and approach bridges, and China Harbour Engineering +
Corporation as main Contractor. Quality control was the
responsibility of the Contractor and Laborat6rio de Engenharia
Civil de Macau, and programme and construction management Taxiway

was provided by Spp, a consortium of Hong Kong-based Scott

Wilson Kirkpatrick with Portuguese consultants Partex and Safety
Pengest. areas
Since the artificial island was formed by a combination of
dredging and sand filling in an area characterised by estuarine
deposits, settlements were of prime concern, especially for the ,
runway and taxiways. Ground monitoring and stability and
settlement predictions were the responsibility of the
Contractor, who was required to employ specialist monitoring
units. In order to fulfill its responsibility for checking the
validity of the monitoring specialists' predictions, SPP
duplicated much of the analysis. Figure 1 - Location of project

* Geotechnical Specialist, Scott Wilson Kirkpatrick

TRANSACTIONS • Volume 3 Number 1 1

Geological Setting Property Marine mud Clays
Moisture content m % 60 - 75 ' 30- 45
The area is characterised by shallow waters with significant
Bulk density y tim' 1.55·1.65* 1.75 - 1.90
amounts of suspended sediment, silt and clay, brought down
liquid limit II % 60 - 75 60 - 65
by the Pearl River. This results in a sea bed of semi-fluid mud,
becoming more solid, but still soft, with depth. Mud Plastic limit Pl % 30 - 35 30 - 35
thicknesses are typically 10-15m. Voids ratio e. 1.8 - 2.0 ' 0.9 -1.2
This recent material overlies older, slightly over- Compression index C. 0.6 - 0.7 0.3 - 0.45
consolidated marine deposits comprising a layer of silty clay Coefficient of volume m. m2/MN 0.7-0.9' 0.2 - 0.4'
typically between 10m and 20m thick below which are compressibility (0.7-1.0) (0.2 - 0.4)
alternating sand and silt/clay layers of varying thickness to a Coefficient of . vertical C, m2/yr 1.8 - 2.6 1.9-3.8
depth of 50-60m below the sea bed. These, in turn, rest on consolidation (1.2 - 2.5) (6 -15)
completely decomposed granite which, for the purposes of Coefficient of - horizontal Ch m2/yr 1.6 - 2.4 1.7-3.4
consolidation (1.2 - 2.5) (6 -15)
settlement analysis, was considered to be incompressible.
Figure 2 shows a simplified soil profile along the runway Shear strength s kPa 4 - 18* 40 - 20
centreline. '" values tend to increase significantly with depth
+ values tend to decrease significantly with depth
Table 1 gives soil properties for the marine mud and clay
( ) values in brackets obtained from back-analysis
layers, based on values obtained from the various site
investigations conducted since 1989, including testing by the Table 1 - Typical Properties of Cohesive Soifs
Contractor's own site investigation specialists. For the purposes
of this paper, the ranges given are intended to provide a was proposed by SPP and carried out as part of the contractor's
"broad brush" picture of soil properties to avoid 100 much monitoring programme.
confusing detail; hence extreme values and detailed variations Around 100 boreholes and probes were completed
during the 1989 and 1992 investigations, to depths ranging
are not included.
between 20m and 80m, and a further 75 boreholes were
Ground Investigation and Monitoring drilled as part of the monitoring programme durinq the
construction period.
Information on ground conditions was obtained by site As a check on stability and settlements, ground
investigations carried out in 1989 and 1992. Due to changes in monitoring was carried out during the construction period. In
both the location and design of the runway, which was addition to the extra boreholes, vane and cone penetration
originally designed as a piled structure, the original tests were carried out at various stages to check strength and
investigations did not fully meet the needs of the amended strength gain of the marine mud to assess stability of the sand
scheme, and a supplementary program of drilling and testing filling. The bulk of the monitoring effort was, however,

Distance from north sea wall (m)

g
g g
o
o "'

Hydraulic sand till

-20

-Cl
<) -30
~

-E

..
c -40
-
0
~
>
~
w -50

-60 Granite bedrock

-70

-80 Note: mean sea level is 1.8m above Macau Chart Datum (MCD)

Marine mud Clays ___I Sands - - - Limit of vertical drains

Figure 2 - Geological section along runway centreline

2 The Hong Kong Institution of Engineers

 - - 269m
~_._._---------.
Wer;l EllS!

P",,,pel woll

s,.t"ty "''''''
II RLlrlW"Y s .....ty ""~".

'.....
. -- ...... " ,'.-
. -,'. . - - '.-,'-

- - -
.'... -.'
-
: - :".
-,
-. .. _
- ..
., '->.-:....L.~••.:.
--:. ~.
:.',:, ..:-.;..:.,. .
.._......
..::;.:t.
-. ..
-- - - - -- -- - ~ ~
- ~

-
~

Vibrocomp:<lcted und 1 Settlement plate

Silt I Cla~ I Surface settlement point ± Settlement gauges

Completel~ decomposed granite 1 Temporary settlement block

! Piezometers

Figure 3 - Section A-A. typical section for northern part of the artificial island

.
concentrated on settlement measurement, since this was of location Spacing No
more concern, and was considered to be a critical factor for Shore • 2 surface settlement blocks every 100m, 150
successful operation of the runway_ Settlement of the sand fill protection during construction
(runway, taxiway and safety areas) was measured mainly dyke • 1 surface settlement point on sea parapet 75-
through settlement plates which comprised 1m-square metal wall after cO,nstruction
• 6 inclinometers in sand fill behind wall at 6
plates or 3m-square frames, attached to a vertical tube. For the locations agreed between CAM's designer
sea dyke, temporary settlement points, consisting of O.Sm' and specialist advisers
concrete blocks, were used initially, to be replaced by points
Runway • settlement plates along the centreline every 115
set in the concrete sea wall once this was cast. and 100m and along the edges every 200m
Special instrumented sections were located at either end taxiway • temporary surface monitoring points on 400
of the artificial Island, which were the first areas to be completed pavement, 3 points every 40m
constructed, and at other areas of special concern. In these • permanent surface monitoring points on 55
areas, the normal settlement plate monitoring was pavement shoulders, 2 points every 200m
supplemented by settlement gauges, piezometers and Safety areas • settlement plates every 200m 45
inclinometers. Typical layout of these instruments is shown in Special • piezometers located at 7 sections of special 22
Figure 3. instrumented interest
Settlement gauges were the magnetic probe type, sections • settlement gauges (extensometers) located 28
comprising magnetic rings inserted at various depths in a at 7 sections of special interest
borehole, with a plastic tube through the centre. A torpedo- Table 2 - Spacing and Numbers of Instruments
type probe inserted down the tube gave readouts of each
plate location. Up to 8 plates were inserted down each
major influence on construction methods and programming of
borehole so that the compression of individual layers of soil
the work. Several measures were taken to speed up the
could be measured. Piezometers were of the pneumatic type,
settlement process and thus reduce the amount of settlement
inserted in boreholes to measure pore water pressures in the
remaining on completion of construction. Vertical drains were
marine mud and upper clay layer. Inclinometers were the
inserted through the marine mud and upper clay layer, as
keyed tube type, inserted into boreholes. The spacings and
indicated in Figure 3, and runway and taxiway areas were
numbers of instruments used are indicated in Table 2.
surcharged using sand fill which was subsequently spread over
the safety areas. Deep densification of the sand fill was
Constraints of the Project specified beneath the runway and taxiways.
The island was formed by dredging the recent marine mud For the runways and taxiways, the aim of these ground
beneath the sea dyke, runway and taxiways, and replacing it improvement measures was to limit settlements after
with sand fill. In the less-critical safety areas, only the upper 1- completion of the pavement so that the surface profile would
2m of mud was dredged. This is illustrated in the typical cross remain within ICAO (1990) guidelines. The long~term
section shown in Figure 3. settlement profile of the runway was of particular concern
For a project of its size and compleXity, the construction since the choice of a concrete pavement will limit the scope for
period was extremely short; initial dredging for the island regulation of the surface at a later date. The limiting factor, for
began in February 1992, and flight trials were conducted, on compliance with ICAO guidelines, is curvature of the
schedule, in June 1995. This short time-scale, coupled with the longitudinal profile, especially combinations of adjacent sag
need to minimise settlements of the completed runway, had a and hump curves, so that it was not possible to set simple limits

TRANSACTIONS • Volume 3 Number 1 3

on settlements or differential settlements. As a general rule, Predicted Residual Settlement'll
however, predictions indicated that a satisfactory surface Total Primary Including
profile should be maintained provided consolidation Feature Settlement (\I consolidation. secondary
settlements of the runway after paving are less than about (mm) only (mm) compression Ul
150mm. Settlement due to secondary compression, though (mm)
expected to be more than this, was considered by the Runway 800·2000 0-100 70 - 200
Designers to be less critical because they believe that resulting
Taxiways 600 -1500 o- 150 50 - 250
differential settlements will be less marked. ICAO tolerances
for taxiways are much less severe. Safety areas 1200 - 4000 200 - 550 -
The Designers believed that the settlement criteria would Sea dykes 400 - 1200 100 - 700 -
be met if the upper clay layer were consolidated under Notes: (1) Total settlement values are based on actual measurements
surcharge to achieve 100% of the consolidation expected plus estimates of settlement which took place before
under long-term load conditions. The surcharge height and readings began and expected future settlement.
vertical drain spacing were determined so that required degree (2) Residual settlements are based on predictions of
settlement that will take place after completion of any
of consolidation would be achieved within 4 months, whilst
section (concreting of pavement, concreting of sea wall.
limiting the quantity of surcharge material to that available finishing of to'psoil in safety areas).
from the surrounding safety areas. This resulted in a surcharge (3) Secondary compression estimates for the pavements were
height of 4m for the runway and 3m for the taxiway, and based on a 20-year design life. Secondary compression was
vertical drain spacings beneath the runway and taxiway of not estimated for the sea dyke and safety areas as the
combination of reduced ground investigation coverage and
106m to 2m. Later calculations, based on site monitoring reduced sensitivity to residual settlements, compared with
results, showed that the required 100% long-term the pavements, did not warrant this refinement.
consolidation of the upper clay layer was easily achieved within
the designed 4 month surcharge period, but compression of Table 3 - Typical Settlements
lower clay layers, beneath the 35m depth which could be
penetrated by the drains rigs, would not be complete if they and taxiway, so its effects had to be included. Finally, a major
were of too great a thickness. Where necessary, the surcharge consideration was that settlement estimates had to be updated
period was extended to achieve sufficient consolidation of frequently to take account of monitoring data, adjusting soil
these lower layers so that the runway profile would remain properties so that calculated settlement-time plots matched
within ICAO guidelines during its 20-year design life. The measured values, since estimates based on laboratory values
ground monitoring and settlement calculations were thus used alone were not considered accurate enough. This is discussed in
to control the surcharge period so that surcharge could be more detail later.
removed as early as possible whilst limiting residual settlements Neither a suitable commercial computer program nor
to acceptable values. Actual surcharge periods varied between publications dealing with calculations of settlements with
about 3 and 8 months. varying load and drainage conditions could be found. It was
Settlements of the sea dyke and safety areas, though less therefore decided to develop suitable equations and computer
critical than those of the runway, were also of concern as model for the frequent back-analyses that were required. A
settlement predictions were used to determine suitable feature of the analysis is the way it deals with load changes.
settlement allowances; constructed levels of the sea dyke and The equations developed from classical consolidation all
safety areas being increased above the design level by an assume that load will remain constant with time, which is
amount to offset all or most of the anticipated future clearly not the case during the reclamation period, when load
settlement. changes occur frequently. A common approach is to calculate
each change of load as a separate loading case. The results of
all the load cases are added together to give the combined
Settlement Analysis effect of the load variations. This direct approach appears to be
As stated above, settlement predictions were required straightforward but has the disadvantage that it requires a
throughout the construction period as an aid to programming separate set of calculations for each change of load; and its
of the work. The scale of settlements and anticipated residual application for unloading can be complicated and raises
settlements after construction can be seen from Table 3. From questions about its validity. In view of these problems, a
this it can be seen that total settlements were very high and different approach was chosen. The load for each time period
yet, especially for the runway and taxiway, residual settlements is taken as the total load applied during that period but a
had to be kept low and had therefore to be predicted with a modified time value is used, as indicated in Figure 4. For
high degree of accuracy for the purpose of making decisions instance, in 4(a), settlement has actually occurred over time I,'
affecting surcharge removal dates and settlement allowances. under load p" but for the purpose of calculating settlements
In order to achieve this, the analysis method had to be able to under load p" time t, is modified to t,', which is the time that
take account of the frequent load changes. Including would have been required to produce the same amount of
unloading and subsequent reloading, that occurred during the consolidation had load p, been applied throughout. These
construction period, and the effects of the insertion of vertical equivalent times will be shorter than the actual time in the
drains part way through the loading sequence. It can also be case of a load increase, 4(a). Equivalent times will generally be
seen from Table 3 that secondary compression accounts for a longer in the case of a load decrease, 4(b), but this situation is
significant portion of the residual settlement of the runway slightly complicated by the fact that consolidation under load

4 The Hong Kong Institution of Engineers

 Drains in the safety areas were installed to 255m depth,
- - ... _-, - .._- -- .. --_..._ .... -_..
~. I.!.·_~"
-- -~
---j} ' ... _
....- through the sand fill and marine mud, and terminated typically
""' ., L
...

'- ~~..iilpr _..__ .

.. ,....._------
, . .. -I
___ between 2m and 6m into the underlying clay, depending on
12' - the level of the mudlclay interface. Rigs were supplied by
!, ,. • ,.
•
••
•
•
•,•
•
, ~ ....
_~d""'"
W.
"
'M
~,
de"..,
,..,. ••• .' several subcontractors and comprised a variety of modified
••• !!'tgl"
" COfI'IlJidIDon
",o<tu.. ~ by .... <1 ", .."t, ..... ll. fIl.
_ ... ~I
1rI..., ,,2 UIlt_ boo ea41:..,.e<l
back·hoes. The drain material was fed down the mandrel from
U""" 10)'" ,,2 Is
,. •" U·II',p2
<109''- ConsotOclIUO"
a roll mounted on the rig and cut off at the surface.
lim.. 11' II thl tim. tr..t would haYI b. .n r.Q\lI<." 10 prod"'"
t~l. ~'Qf"' 01 eon.oIklltlon und.,lood 1'1, The value 01 pI' I. Although vertical drains are commonly used in Macau,
C1I'cul.ted l'OM "'I "''''011'10''.
Un'.-C<ln ...kl.li(>n 10""1",,,
lim. those required for the airport were deeper than is usual locally,
---._.~-

,., -_.--._--- -- which led to some initial difficulties. Successfui insertion of

.... ~._-- -... - -_ .. --- - . ._. - drains depended on the expertise of the operators, and some
I ~ _ _ _ 11 operators took several weeks to reach satisfactory production
----t
"
l~pl
I •
!•
.-
•
-
...."
• "I\: .j
--~ ..
". ._.
--
l(NId p2

.1
L -
rates. The main problems were due to the drain coming back
up with the mandrel as it was withdrawn or tearing of the
drain inside the mandrel. Both were affected by the size and
•
••
••
•
• .. . stiffness of shoe used for insertion. Shoes, small rectangular
• ..
plates with handles for attaching the vertical drain, block off
the bottom of the mandrel during insertion and then anchor
-_. ._-----_._---- ._._.- .__ ...- ..
TIm.
---_..- -- •.. _._.- the drain in the ground during withdrawal. The use of water
I" to flush the mandrel during insertion was particulariy effective
0 in curing these problems and was eventually adopted by most

, ---
"
lO"'_~"=~_-= (Oacl~~ __ _ "
1.___._ subcontractors, sometimes after initial resistance owing to the
added complication of providing the necessary pumps and
! hoses.
~
• Drains beneath the runway and taxiway areas were
installed to between 325m and 3S.Sm depth but, because the

._._-_._-- ,... mud in these areas had been replaced by sand fill, it was not
- -_._----- --_ .. .. _ _---_.~-

- necessary to install drains up to the surface. Instead of using

'<' drain on a roll, drains were pre·cut to the required length and
pulled up the mandrel using a cord attached to the top of each
Figure 4 .. Treatment of load changes in the settlement
length. After insertion the cord was either snatched out of the
analysis: (a) increase of load; (bi decrease of load with
top of the drain before withdrawing the mandrel, ready for reo
subsequent consolidation; (cJ decrease of load without
use, or simply cut off at ground surface after withdrawing the
further consolidation
mandrel. This method slowed operations by about 15% but
P, may represent more than 100% consolidation under load p" saved over 50% of drain length. Drains were cut so that at least
as shown in 4(c), in which case the concept of effective time 3m of drain extended into the sand fill, to allow for variations
will be meaningless. This approach allows any number of load in dredge depth and contamination of mud in the lower fill.
changes to be accommodated within a single set of settlement Because of the additional length of towers needed for
calculations although its application becomes somewhat these deeper drains, and the more difficult driving conditions,
complex when the effect of vertical drains is included in the heavier rigs were required, based on cranes. Most rigs used
calculation. The detailed development of the consolidation static force to insert the mandrel, as with the backhoe·based
equations for this approach and their application in a machines, but a few used vibrators mounted on top of the
computer model are therefore the subject of a separate paper. mandrel. The vibratory machines were generally slower.
Penetration was a major problem with the deep drains,
Experience with the Ground Treatment Methods and, in addition to changes in shoe design and the use of
water flushing, the cross·section of the mandrel was reduced
Vertical drains on some machines to obtain the required penetration. In a few
Altogether, about 380,000 vertical drains were installed in the areas where the clay was exceptionally strong, full penetration
artificial island, with a total of around 8,500 km of drain being could not be obtained but this was not considered to be a
used. All work was carried out from the sand fill once filling major worry, since stronger clays are likely to be less
reached just above high tide level. Two types of drain were compressive.
used, both comprising a 100mm·wide flat plastic core with a Drain spacing was specified on a square grid but in
geotextile filter wrapped around it, inserted into the ground practice every second row was staggered to give an isosceles
using a hollow rectangular·section mandrel. Drains from triangle pattern. Spacing in the safety areas was I.Sm except
TIanjin and Nanjin, China, had heavy, ridged cores to provide a for the last areas to be completed where it was reduced to
drainage path whilst Nylex drains from Malaysia were of 103m. Drain spacing beneath the runway and taxiways was
lighter construction and had dimpled cores to give the same typically 2.0m but this was reduced to I.Bm in the last sections
effect. Notwithstanding the various arguments in favour of to be completed and to 1.6m in isolated sections of the runway
one type of drain or another, both types appeared to give and taxiway where exceptionally soft soils remained beneath
satisfactory performance during installation and in service. the dredged trench.

TRANSACTIONS • Volume 3 Number 1 5

 The time taken to insert a single drain and move to the equipment available. The vibrators used were 400mm
next position was typically around 2.5 minutes but the overall diameter, 3m long, with specially thickened walls and weighed
rate of production was reduced by mishaps such as jamming about 2,5OOkg. Their 120kW motors generated a centrifugal
and tearing of the drain, minor repairs and servicing, and by force of 220k N.
the need to move periodically to new areas. Actual The usual procedure was followed, whereby the vibrator
productivity of the rigs varied widely, from an average of was run at low power and steadily lowered to full depth. It was
below 100 drains per 16-hour day for the poorer machines to then run at full power and withdrawn in 0.5m stages.
more than 400 drains per day for the best rigs, with an overall Vibration at each stage was normally maintained until the
average of about 250 drains per day. These figures are only for motor current reached 200-250W, or for 2 minutes if the
normal working days and do not include days broken down, current failed to reach 200W, which sometimes occurred due to
holidays, or standing idle due to inclement weather or lack of the presence of silty lenses within the sand fill. Sand and silty
working area. layers with a high initial density caused surges in the current to
over 400W if the rate of penetration or withdrawal was too
Vibrocompaction rapid, resulting in frequent burn-outs of vibrator motors until
Densification of the sand fill was deemed necessary by the operators developed the skill and vigilance needed to avoid
Designers to reduce creep and variability within the fill. such problems. After tests using varying amounts of jetting
Densification of sand fills is also known to increase resistance to water, both subcontractors found that best results were
liquefaction in the event of an earthquake. The Designers did obtained by using large amounts of water, and added extra
not give details of the improvement in performance expected pumping capacity.
in terms of creep settlement and, indeed, this would be The time taken to complete insertion and compaction at
difficult to quantify given the present poor state of knowledge a single point varied according to soil conditions from as little
and wide variations in reported observations, but it is as 30 minutes to over 2 hours. Typically, 24-28 points were
generally acknowledged that densification of loose fills greatly completed per machine during a 24-hour working day. It was
reduces creep settlements. For instance, densification may found during trials that the required degree of compaction
reduce creep settlements from around 1% compression of the could be achieved with spacings, on a triangular grid, of as
layer to around 0.2% over a 20-year period, but results vary wide as 5m, but spacings of 4.5m and 4.75m were actually used
widely. to allow for variability of conditions. These spacings are much
The design included a two-stage deep densification of higher than typical values quoted in most texts and probably
the sand fill beneath the runway and taxiways. reflect the greater power of modern equipment.
Vibrocompaction was to be used once sand filling had been Once experience had been gained with local conditions,
completed to just above high tide level, with dynamic the required minimum compaction criteria were easily
compaction (heavy tamping) being specified for the remaining exceeded in clean sands. Where bands of material failed to
4.5m of sand fill once filling reached pavement level, just prior meet the required standard, investigations showed it was due
to surcharging. To save time, SPP proposed that densification to a high silt content within the sand fill. This problem was not
be carried out in one operation using vibrocompaction from widespread and was most common in the lower parts of the fill
the top of the surcharge. This effectively saved all the time where some mixing of trench bed deposits and sand fill had
originally scheduled for both densification operations, since occurred during placement. The specification of two
the vibrocompaction was carried out during the surcharge acceptance standards, based on SPT and CPT results, did not
period for each section of runway or taxiway, when no other create a problem, as a pass or failure in one test was almost
work was possible. invariably accompanied by a similar result in the other. The
The degree of compaction was specified in terms of both specification called for one SPT and one CPT every 5000m' but
standard penetration test (SPT), N-values and static cone (CPD additional tests were carried out wherever there was concern
resistance, the minimum requirements being 25 blows (SPT) about the quality of compaction. Automatic records of
and 12MPa (CPD down to 3.5m below formation level, and 20 vibration time and motor current against depth were also
blows and 10MPa below this. Initial CPTs and SPTs showed the useful for quality control.
sand fill to be loose to medium dense, with cone tip resistance
of typically 2-8MPa and N-values of 4-17 blows.
Experience with the Monitoring Methods
The work was carried out by two specialist
subcontractors, Keller of Germany and a joint venture of Agra Settlement plates should ideally have been installed at or close
from Canada and Hong Kong company Express. In addition to to the base of the sand fill so that they would be unaffected by
the time saving referred to above, working from the top of the settlements of the sand fill itself, and especially by
surcharge also overcame the problem that vibrocompaction is vibrocompaction operations. Initially, therefore, settlement
not fully effective in the upper 2-3m of ground. However, this plates were planned to be installed deep in the sand fill during
change also meant that the sand fill had to be densified to bottom dumping operations. However, the first settlement
25m depth, through material which varied greatly in initial plates to be installed were all rapidly destroyed by barges, and
density and fines content owing to the different methods of it wal decided to install plates higher in the sand fill after
placement (bottom dumping from barges, hydraulic fill and bottom dumping had finished. This meant that plates would
bulldozer) and the large number of sand sources used. This be affected later, and possibly damaged, by vibrocompaction
combination of depth and variability caused initial difficulties operations. Care was taken to minimise problems: the
and required the use of the most powerful vibrocompaction vibrocompaction probe positions were modified in the vicinity

6 The Hong Kong Institution of Engineers

of the plates and, in the event, damage by vibrocompaction monitoring but, to be reliable, this requires long periods of
proved not to be a major problem. It also proved not too monitoring under constant load. This approach was tried by
difficult to estimate the effects of vibrocompaction on the the Contractor's monitoring specialists but, since loads were
settlement readings and make the necessary corrections. continually changing during the construction period, the
Interestingly, plates tended not to settle as much as would method was considered unsuitable. The only approach
have been expected from surface settlements, though there considered to be sufficiently reliable was to use back-analysis,
was considerable variation between plates. It was assumed with soil profile data obtained from borehole records and soil
that, except for the disturbance during vibrocompaction, consolidation properties adjusted to fit observed results.
settlement of the sand fill was negligible compared with As settlement readings became available, improved
consolidation settlements. settlement estimates were made by performing back-analyses,
Judged in terms of useful information gained, the simple the accuracy of predictions steadily increasing as the
settlement plates were by far the most effective of the monitoring period increased. The problem with back analysis is
installed instruments. Their simple, robust design meant that that there are several unknowns which may be varied in any
they were less prone to damage than other types of number of permutations so that the analysis fits the measured
instrument, could be more easily repaired when damaged and data. Each permutation will produce a different final
were cheap enough to be used in large quantities. Settlement settlement, with often widely differing results depending on
gauges had the advantages of giving settlements of individual the assumptions made. Unknown variables are:
soil layers and avoiding the need to correct for
• the initial settlement that has taken place before
vibrocompaction. They were also reasonably robust though not
measurements started,
as durable as the settlement plates. Their main disadvantage
was their cost, since each set of plates required a borehole. This • effective field m, and C,lCh values for each compressible
severely limited the numbers that could be used. Piezometers stratum, and
also proved to be of limited value as readings often did not
• the soil profile, especially relating to drainage paths.
give values that would logically be expected so were difficult
to interpret; a problem that has been reported by engineers on In fitting the analyses to the data it was found that
other land reclamation sites in the Pearl River delta area, so various combinations of m, and c, could be used to produce a
would appear to be a common occurrence. Again, numbers of calculated rate of settlement that fitted measured values, with
piezometers had to be severely limited due to their cost. the effects of assumed higher m, values balancing those of
Inclinometers provided little useful information and did not assumed lower c, values, and vice-versa. The combination of
justify their high cost. high m" low c, produced higher predicted residual settlements
than assuming low m" high c,. By trying various combinations
of parameters, whilst using judgement to keep all values
Experience with the Analysis Method
within reasonable bounds, upper and lower bound limits could
Initial analyses were based on values of the coefficient of be obtained for future settlements although, initially, m, values
volume compressibility, m" and the coefficients of vertical and were kept reasonably close to laboratory values for reasons
horizontal consolidation, c, and Ch, from laboratory discussed above. As the degree of consolidation increased, the
consolidation tests. Laboratory values of m, are normally settlement/time plots developed pronounced curvature, and
considered to give a reliable guide to field behaviour but, the combinations of m, and c, that could be used to give both
especially in marine deposits, laboratory c, and Ch values are the correct slope and the correct curvature were greatly
known to be unreliable since they do not include effects due to reduced, with the result that the range between upper and
the macro-structure of the soil, such as interbedded sand layers lower bound estimates narrowed until, at pavement
which tend to speed up consolidation. Thus, it is usually construction, predictions could be made with reasonable
assumed that total long-term settlements can be calculated confidence. Throughout, best-guess values were chosen to
with reasonable confidence but that the rate of settlement will reflect the more pessimistic assumptions so that, as time
generally be under-estimated based on laboratory values. progressed, estimates would improve or, at worst, were not
On the Macau International Airport project, rates of likely to get worse.
settlement were as important as total settlement, since the all- Predictions were found to be sensitive to the assumed soil
important residual settlements after completion of profile. For instance in areas where ground conditions varied
construction, especially of pavements, were determined by the greatly between boreholes, it was sometimes not clear
rate of settlement as much as by its final magnitude. It was whether the clay beneath individual settlement plates existed
therefore essential to obtain the best possible estimates, and as a single thick layer or as several layers, interspaced with
relying on laboratory test values alone was not sufficient. sand. The calculated settlement curve could be made to fit
One approach sometimes used to obtain improved values predictions by assuming either a single thick layer with a high
of c, and Ch is to calculate their values from a combination of c, or several thinner layers with lower c, values. It might be
field permeability tests and laboratory consolidation tests, thought that, provided the calculated curve fitted the observed
using the relationships between m" c, and permeability. This settlements, the two assumptions would produce much the
approach is not without its problems and was not possible at same predictions but' in practice the single-layer, high-c,
Macau International Airport since no field permeability tests assumption produced substantially higher predictions of future
were carried out. Another approach is to use curve-fitting settlement. In the event, the more optimistic assumption of
techniques to predict future rates of settlement from field several clay layers invariably produced predictions that were

TRANSACTIONS • Volume 3 Number 1 7

closer to actual behaviour during later stages of the
construction period. Michael carter
Values of m. and c. obtained from back analysis can be For the past four years. Mike Carter has
compared with laboratory values in Table 1. From the table, it worked as the geotechnical specialist for Scott
can be seen that the assumption that laboratory mv values are Wilson Kirkpatrick on a construction
normally realistic but that c. values are too low proved management team at Macau International
reasonably true at Macau International Airport, since m. values Airport. A graduate of the University of Wales,
Swansea, he has about thirty years' civil
obtained by back-analysis are similar to those obtained from
engineering experience. with particular
laboratory tests, and that c. values from back analysis are
emphasis on geotechnical engineering and
about 4 times laboratory values for the clays. A surprise, materials, and has worked on a variety of projects in Asia and Africa as
however, is that for the marine mud the c. values obtained by well as the U.K. He also spent ten years as a university lecturer in the
back analysis are similar to the laboratory values. A further U.K. and Indonesia and two years as a training adviser in Sri lanka. He
surprise was that Ch values were similar to c. values for both is the author of three books on geotechnical engineering.
clays and marine mud, since it is common for such sediments to
exhibit chi c. ratios in the range of 2 to 4.
As a check on settlement predictions, settlements of the
completed runway were measured for the first 4-6 months
after paving and trends projected into the future assuming an
exponential decay in the rate of settlement (i.e. that a
settlementilog-time plot would give a straight line). Estimates
of residual settlements after paving using this method were
within the same range as those given in Table 3 and individual
results were within ± SOmm of values obtained from the
settlement spreadsheet calculations.
Overall, the analysis method developed and its
application in combination with a back-analysis approach
proved successful. The advantage of back analysis is that, even
though the consolidation characteristics and drainage
conditions of the soil are not known in detail, as is generally
the case with highly stratified marine deposits, the analysis still
produces a trend line which can be used to predict residual
settlements with reasonable confidence. A limitation of the
back-analysis approach is that it requires a high degree of
consolidation before it can be used with confidence. This did
not detract from its usefulness at Macau International Airport
because the settlement analyses were used primarily to
determine the time of surcharge removal, which was not
permitted until a high degree of consolidation has been
achieved. The method was also found to be sensitive to the
accuracy of the soil profile and, whilst it can help overcome the
deficiencies of laboratory test values of soil consolidation
parameters, it is not a substitute for good ground profile
information.

Acknowledgements
The author thanks CAM Sociedade do Aeroporto Internacional
de Macau and SPP Concorcio Internacional for permission to
publish this paper. Grateful thanks are also due to Adelina
Kong Santos, whose help and ideas during both the
development of the settlement analysis system and the
preparation of this paper were invaluable, and to colleagues
David Coffey and Denis McDowell for checking the drafts and
for their helpful suggestions.

References
ICAO (1990). International Standards And Recommended Practices.
Aerodromes: Annex 74 to the Convention on International Civil
Aviation, Vol. I, Aerodrome design and operations, International Civil
Aviation Organization.

8 The Hong Kong Institution of Engineers