Professional Documents
Culture Documents
Instn
Civ. Engrs
groundwater control
Geotech. Engng,
2000, 143, Oct.,
177±190
M. Preene, BEng, PhD, CEng, MICE, CGeol, FGS, MCIWEM Paper 12243
Written discussion
& Lowering of groundwater levels by Tr time factor for radial ¯ow closes 21 December
pumping increases vertical eective stress Tv time factor for vertical drainage 2000
in the soil around an excavation. The t elapsed time since pumping began
resulting ground settlements are normally Du reduction in pore water pressure Manuscript received
small, but in certain circumstances may be Duav average pore water pressure reduction in 7 January 2000;
large enough to cause damage to sur- a soil layer revised manuscript
accepted 31 May
rounding structures. This paper describes x linear distance from edge of ground-
2000
the mechanism of these settlements and water control system for plane ¯ow
presents a simple risk assessment gw unit weight of water
methodology for initial identi®cation of Ds0v Increase in vertical eective stress
zones where buildings may be at varying r compression of a soil layer
degrees of risk of damage. These risk r corr corrected compression of a soil layer
zones can help to determine the extent of rt eective compression of a soil layer at
preconstruction building condition surveys time t
needed, or to assess the need r total ground settlement
for settlement mitigation or avoidance r ult ultimate compression of a soil layer
measures. Potential uncertainties in the
assessments are discussed. Introduction
If an excavation is made below groundwater
Keywords: geotechnical engineering; level, groundwater control by pumping (also
groundwater; subsidence known as construction dewatering) may be used
as part of the temporary works. The practicali-
Notation ties of pumping will depend on the method
a length of groundwater control system used 1 but all methods, whether pumping from
b width of groundwater control system wells, wellpoints or sumps, will reduce pore
Cd correction factor for eective stress water pressures and increase vertical eective
cv coecient of consolidation of soil stress in the soil around the excavation,
D thickness of soil layer, thickness of perhaps for some considerable distance away.
con®ned aquifer 2. These increases in eective stress will
E 0o stiness of soil in one-dimensional com- result in ground settlements every time
pression groundwater control by pumping is carried out.
H initial saturated thickness of uncon®ned In the great majority of cases movements are so
aquifer, initial piezometric level of small that no distortion or damage is apparent
con®ned aquifer in nearby structures. However, in some circum-
h maximum drainage path length for ver- stances settlements may be large enough to
tical drainage cause distortions resulting in eects ranging
kh coecient of permeability in the hori- from slight cracking of architectural ®nishes to
zontal direction major structural damage. These eects may
kv coecient of permeability in the vertical extend several hundred metres from the
direction groundwater control system, and could aect
L distance of in¯uence large numbers of structures. This paper
mv coecient of volume compressibility of describes the circumstances and mechanisms
soil by which settlements may result from ground-
R average degree of consolidation of a soil water control operations, and proposes a risk
layer assessment method to determine, at the early
r radial distance from centre of ground- stage of a project, whether such settlements
water control system for radial ¯ow should be of concern.
re equivalent radius of groundwater control
system The problem
s drawdown 3. The magnitude and distribution of
s av average drawdown in soil layer ground settlements resulting from groundwater M. Preene,
sw target drawdown in the excavation control are dependent on the groundwater ¯ow Groundwater Team
(s/s w ) ratio of drawdown to target drawdown pattern, soil strati®cation and the spacial dis- Leader, Ove Arup &
in the excavation tribution of soil stiness. This is a complex Partners, Leeds
177
PREENE
three-dimensional and time-variant problem; avoided with adequate investigation and plan-
precise analysis would require a site-speci®c ning but, sadly, these continue to happen. 6,7 If a
numerical model. This has been done on major failure occurs local settlements can be much
projects with critical structures at risk. 2,3 larger than from an eective groundwater
However, such detailed work is unlikely to be control system.
economic or feasible for many smaller-scale 9. Powers 8 identi®es conditions when
projects such as basement construction, shaft inadequate groundwater control may lead to
sinking or open-cut pipelines. settlement. In practice, most problems can be
4. An important dierence between settle- avoided provided the groundwater control
ments from groundwater control and those from system is appropriate for the conditions. 1 One
tunnelling or excavation is the potential extent particular problem occurs if ®ne soil particles
of the zone of signi®cant settlement. Normally, are continually drawn from the ground during
settlement due to tunnelling or excavation is a pumping (known as loss of ®nes). During early
relatively local phenomenon, becoming negligi- stages of pumping it is normal for some ®nes to
ble at a distance of a few times the excavation be drawn through a well or sump, as a more
depth. In contrast, where a highly permeable permeable zone develops. However, if pumping
stratum is present beneath compressible allu- of ®nes continues for extended periods sub-
vial soils, settlements can occur at large dis- surface erosion channels may develop and
tances from the groundwater control system; in subsequently collapse, leading to unpredictable
a case described by Powrie, 4 damage was and potentially large settlements.
reported 500 m away from an excavation.
5. Unfortunately, the area covered by pre- Effective stress settlements from
construction building condition (or dilapida- groundwater control
tion) surveys often does not extend far enough 10. Assuming that groundwater control pre-
to include all structures which may be aected vents the excavation failures previously
by groundwater control. In that case it is not described, ground settlements result purely
possible to determine at an early stage how from the reduction in pore water pressure and
many structures should be investigated further increase in eective stress that occurs when
(either due to their historic nature or intoler- groundwater levels are lowered (or drawn
ance to settlements), or may require mitigation down) by pumping.
measures. Also, in the event of damage being 11. Pore water pressures in soil layers will
alleged, buildings often lie outside the zone respond dierently to groundwater pumping
initially surveyed, making assessment of the depending on their permeability in relation to
validity of claims more dicult. other strata. For the purposes of settlement
6. The limited extent of preconstruction assessments, strata can be approximated to one
condition surveys may explain the paucity of of three categories: aquifer, aquitard, and
case history data available for settlement and aquiclude (Fig. 1), de®ned as follows.
damage arising from groundwater control. This
contrasts with tunnelling-induced settlements (a) Aquifer. A signi®cantly permeable Fig. 1. Groundwater
where established methods exist 5 to estimate stratum (generally the most permeable ¯ow in aquifers,
the extent of the zone of signi®cant settlement. stratum present). Dewatering wells nor- aquitards and
This allows baseline data to be obtained prior mally pump directly from the aquifer; aquicludes
to construction, and helps generate de®nitive
case data of settlements and damage.
7. There is a need for a risk assessment
methodology suitable for use on smaller-scale
projects, where data may be limited, to assess
the zones where potentially damaging settle-
ments are possible. Such zones can help deter-
mine the extent of preconstruction building
condition surveys and identify critical struc-
tures at risk, for which more detailed assess-
ment may be necessary.
groundwater ¯ow will be predominantly will be equal to the reduction in pore water
horizontal. Due to the high permeability, pressure Du, therefore
eective stress changes (and hence com-
D
pression) occur contemporaneously with r ult Du
2
E0o
the drawdown of water levels. Con®ned and
semi-con®ned aquifers are overlain by 14. The assumption of constant total stress
aquicludes and aquitards respectively, and is reasonable, because many ®ne-grained soils
have an initial piezometric level above the remain eectively saturated for some depth
top of the aquifer so the aquifer is satu- above the groundwater level, with no change of
rated throughout. An uncon®ned aquifer is unit weight. Even when soils desaturate, the
characterized by the top of the aquifer reduction in unit weight is relatively small and,
being open to atmosphere, with a water given the uncertainties in other parameters, can
table or phreatic surface below ground be neglected without signi®cant error. For the
level (Appendix 1). case shown in Fig. 1 with the wells fully
(b) Aquitard. A stratum of lower permeabil- penetrating a con®ned or semi-con®ned aquifer,
ity than the aquifer (normally at least two groundwater ¯ow will be horizontal, the pore
orders of magnitude less). During pumping, water pressure reduction will be constant with
groundwater ¯ow in the aquitard will tend depth and Du gw s (where g w is the unit weight
to be vertical, draining towards the aquifer. of water). Compression at distance r from the
Eective stress changes and compression centre of the excavation can be related to
of the aquitard will occur slowly, at a rate drawdown at that point.
controlled by the consolidation parameters
D
for vertical drainage. r ult gw s
3
E0o
(c) Aquiclude. A stratum of very low perme-
ability. During the period of pumping from 15. This allows the compression at any
an adjacent aquifer, no signi®cant pore distance r to be determined, provided the
water pressure reductions occur in the drawdown s or pore water pressure reduction
aquiclude. This implies that compression of Du are known. This requires a model of the
an aquiclude resulting from groundwater distribution of drawdown with distance.
control will be negligible. Several expressions are available from hydro-
geological theory, relating drawdown s to a
12. Figure 1 shows a section through a constant pumped ¯ow rate. However, most
schematic drawdown curve for radial ¯ow in a groundwater control systems are designed to
con®ned or semi-con®ned aquifer. Pumping achieve a target drawdown in the excavation,
results in pore water pressure reductions in the rather than a target ¯ow rate. They tend to
aquifer and any overlying and underlying operate under conditions of constant drawdown
aquitards (but not, by de®nition, in any aqui- at the excavation, not constant ¯ow rate. A
cludes). The eect on an individual soil layer useful parameter, which will be known for most
can be considered using an elastic analysis groundwater control systems, is the target
which gives the ultimate vertical compression drawdown at the edge of the excavation, s w .
r ult , of a soil layer of thickness D subject to a This gives
uniform increase in vertical eective stress Ds0v s D
as r ult g w sw
4
sw E0o
which, if the variation of s/s w with distance is
D
r ult Ds0v
1 known, allows estimation of a compression
E0o plotted against distance curve. Curves of s/s w
in con®ned or semi-con®ned aquifers are avail-
where E0o is the stiness of the soil in one- able for radial ¯ow to excavations modelled as
dimensional compression. Equation (1) is some- equivalent wells (of radius r e ) from Rau 9 (Fig. 2),
times written in terms of m v (the coecient of and for plane ¯ow to linear excavations (such
volume compressibility, where E0o 1=mv ). as pipeline trenches) from Powrie and Preene 10
However, the use of E0o is preferred because (Fig. 3). Fig. 2 or 3 gives the drawdown
eective stress compression occurs in both distribution in an aquifer layer; substitution of
coarse and ®ne-grained soils, and m v is nor- the aquifer thickness and stiness into equation
mally only quoted for ®ne-grained soils such as (4) will give the compression of the aquifer
silts and clays. 16. The relationships shown in Figs 2 and 3
13. It is useful to express equation (1) in are time-dependent and describe how, as
terms of pore water pressure reduction Du or pumping continues, a zone of drawdown propa-
drawdown s, because these parameters can be gates outward in the aquifer away from the
measured in the ®eld. Neglecting any change in dewatering system. The period for which
the unit weight of soil, the vertical total stress pumping continues is an important factor when
is unchanged by groundwater lowering, so Ds0v assessing settlements. In given ground condi-
179
PREENE
s /sw
dency that must be considered for low perme-
ability aquitards. In general, r ult can only fully 0·6
develop when sucient time has passed for
water to drain out of the soil in response to the
0·8
groundwater lowering. An aquifer, by de®ni-
tion, is signi®cantly permeable and for all Tr = (k hE ′o/γw)t/re2 using aquifer parameters
r t Rr
rult (5) 0·6
Table 1 Tentative limits of building settlement and tilt for damage risk assessment 5
Notes:
Maximum settlement is based on the nearest edge of the structure to the groundwater control system.
Tilt is based on rigid body rotation, assuming that all of the maximum settlement occurs as dierential settlement across the width of the
structure, or across an element of the structure.
The risk category is to be based on the more severe of the settlement or tilt criteria.
182
ASSESSMENT OF
SETTLEMENTS CAUSED BY
GROUNDWATER CONTROL
* Location of damage in the building or structure must be considered when classifying degree of damage.
{ Crack width is only one aspect of damage and should not be used as a direct measure of it.
information on ground conditions beneath the criteria for the appropriate type of structure (e.g.
building. This can allow a more detailed assess- Boscardin and Cording 13 for masonry structures
ment of the dierential settlement and resulting or various authors collated in Burland and
distortion anticipated at each structure. Predic- Wroth 12 for framed structures) to estimate the
tions can be compared with published damage level of damage more accurately.
Negligible Super®cial damage unlikely None, except for any buildings identi®ed as being sensitive, for which a detailed
assessment should be made
Slight Possible super®cial Building condition survey, to identify any pre-existing cracks or distortions. Identify
damage, unlikely to have any buildings or pipelines which may be sensitive, and carry out detailed assessment.
}
structural signi®cance Determine whether mitigation or avoidance measures are required locally
Moderate Expected super®cial
damage and possible
Building condition survey and structural assessment. Assess buried pipelines and
structural damage to
services
buildings; possible damage
to rigid pipelines
Severe Expected structural
damage to buildings and
Determine whether anticipated damage is acceptable or whether mitigation or
expected damage to rigid
avoidance measures are required
pipelines or possible
damage to other pipelines
185
PREENE
32. Depending on the nature of the project variation with distance s/s w , and soil stiness
and the number and types of structure at risk, E0o can be dicult to estimate appropriately and
some (or perhaps all) of the potential damage can create large errors and uncertainties.
may be deemed unacceptable. In such cases 35. The dimensionless curves of drawdown
measures will be required to mitigate or avoid plotted against distance in Figs 2 and 3 assume
damage to structures. Table 4 lists settlement uniform aquifer properties and boundary con-
mitigation or avoidance measures. In addition ditions. This is not a good assumption if the
to mitigation or avoidance there is a third aquifer is of limited extent, or where an
option, rarely considered explicitlyÐthat is, impermeable feature forms a barrier boundary
acceptance of settlement. Powers 17 suggests across the aquifer. In that case the drawdown
that, provided damage risk is not more than curve will be very ¯at, and distant drawdowns
slight, accepting minor claims from third and settlements will be rather greater than
parties may be more economical than deploying predicted from Figs 2 and 3. 4 Conversely, if the
large-scale mitigation measures. On many pro- aquifer is in hydraulic connection with a close
jects this would be controversial and could source of recharge (such as a body of surface
present a public-relations challenge, but might water) the drawdowns and settlements may be
be appropriate where relatively few structures smaller than predicted. Numerical modelling of
were classi®ed as being slightly at risk. A the drawdown response may help, but this
preconstruction building condition survey approach is most suited to cases when detailed
would be essential for this approach. aquifer data, such as pumping test results, are
available.
Uncertainties in assessments 36. Predicted settlements are clearly depen-
33. The methodology described in this dent on the values of soil stiness used in
paper is intended to be a ®rst step in assess- calculations. Stiness will generally need to be
ment of the damaging settlements that may estimated for coarse-grained granular soils, for
result from groundwater control. Because of the ®ne-grained cohesive soils, and perhaps for
need to base the assessment on parameters that more unusual soils such as peat. For small-scale
are available or determinable for even small- projects, the data from which soil stiness can
scale projects, the method contains a number of be inferred may be limited, perhaps consisting
uncertainties, and is best applied using engi- of standard penetration test results for granular
neering judgement and experience. Appropriate soils and oedometer test results for cohesive
use of the method requires an understanding of soils. Detailed discussion of measurement of
these uncertainties. soil stiness is beyond the scope of this paper.
34. Equation (4) shows the parameters Preene et al. l list some of the methods of
required to estimate ultimate settlement. The estimating soil stiness. Matthews et al. 18
drawdown s w at the excavation can be esti- compare stinesses estimated by geophysics
mated fairly accurately. The thickness D of an with values from other methods; considerable
individual soil layer is likely to be known variation is reported.
within reasonable limits (assuming appropriate 37. One of the most important factors when
borehole and geological mapping information is estimating soil stiness is that the values used
available). These parameters are unlikely to must be appropriate to the levels of stress and
result in large errors in assessments. In con- strain which the soil will undergo during the
trast, the remaining parameters of drawdown lowering of groundwater levels.
Protect individual structures Prior to the works, underpin the foundations of some or all of the structures at
risk
Reduce the number of Reduce drawdowns by reducing the depth of excavation below groundwater
structures at risk level
Reduce the extent of the risk zones by minimizing the period of pumping
Use cut-o walls to reduce external drawdowns
Use a groundwater recharge system to minimize external drawdowns
* If a cut-o structure is used to avoid external drawdowns it is essential that a groundwater monitoring regime
is in place to allow any leaks in the cut-o to be identi®ed before signi®cant settlements can occur.
186
ASSESSMENT OF
SETTLEMENTS CAUSED BY
GROUNDWATER CONTROL
Underlying 0 0
γwH γw(H – s) γw s
aquifer
Effective stress correction factor
55. As an aid to estimation of settlements a (b)
correction factor for eective stress changes
C d = s av /s can be de®ned in terms of s (draw-
down), D (thickness of con®ned aquifer) and H (e) Aquitard: with negative pore water Fig. 10. Pore water
(initial saturated thickness of uncon®ned pressures pressure reductions in
aquifer or initial piezometric level of con®ned aquitards: (a) zero
Cd 10 (18)
aquifer). C d can be expressed in terms of a pore water pressures
dimensionless drawdown s/H and an aspect The thickness of the aquitard used for this above phreatic
ratio for con®ned aquifers D/H, as: case should include the zone of negative pore surface; and (b)
water pressure above the phreatic surface. negative pore water
(a) Uncon®ned aquifer (D=H 1) 56. Values of C d are plotted on Fig. 5. pressures above
s phreatic surface
Cd 1 ÿ 05
14 References
H
1. P R E E N E M., R O B E R T S T. O. L., P O W R I E W. and
(b) Con®ned aquifer: drawdown level above D Y E R M. R. Groundwater Control: Design and
Practice. Construction Industry Research and
the top of the aquifer (D=H 0)
Information Association, London, 1997, CIRIA
Report FRCP/50.
Cd 10 (15)
2. K N I G H T D. J., S M I T H G. L. and S U T T O N J. S.
Sizewell B foundation dewateringÐsystem
(c) Con®ned aquifer: drawdown level below design, construction and performance monitoring.
the top of the aquifer GeÂotechnique, 1996, 46, No. 3, 473±490.
2 0 13 3. H A Y W A R D D. Wonderful challenge: engineers
D s forming vast station boxes for the Copenhagen
62 2 B 1 C7
Cd 054 ÿ H ÿ H ÿ @ A5 metro face extremely tight ground settlement
s D s D D s controls. Ground Engineering, 1999, 32, No. 11,
H H H H H H Nov., 30.
16 4. P O W R I E W. Ground movements due to construc-
tion dewatering. Groundwater Problems in Urban
(d ) Aquitard: no negative pore water pressures Areas (Wilkinson W. B., ed.). Thomas Telford,
s London, 1994, pp. 237±250.
Cd 1 ÿ 05
17 5. L A K E L. M., R A N K I N W. J. and H A W L E Y J. Predic-
H tion and Eects of Ground Movements Caused by
189
PREENE
Tunnelling in Soft Ground Beneath Urban Areas. associated with building on a variable depth of
Construction Industry Research and Information ®ll. Ground Engineering, 1999, 32, No. 7, July,
Association, London, 1996, CIRIA Report PR30. 32±35.
6. B A U E R G. E., S C O T T J. D., S H I E L D S D. H. and 15. S I M P S O N B., B L O W E R T., C R A I G R. N. and W I L K -
W I L S O N N. E. The hydraulic failure of a coer- I N S O N W. B. The Engineering Implications of
dam. Canadian Geotechnical Journal, 1980, 17, Rising Groundwater in the Deep Aquifer Below
574±583. London. Construction Industry Research and
7. G R E E N W O O D D. A. Re-levelling a gas holder at Information Association, London, 1989, CIRIA
Rhyl. Quarterly Journal of Engineering Geology, Special Publication 69.
1984, 17, No. 4, 319±326. 16. F R O G A T T I. J. Groundwater Disturbance and
8. P O W E R S J. P. DewateringÐAvoiding its Unwanted Surface Settlements Around a Dewatered Sand
Side Eects. American Society of Civil Engineers, and Gravel Quarry. PhD dissertation, Nottingham
New York, 1985. Trent University, 1997.
9. R A U D. B. Construction dewatering by vacuum 17. P O W E R S J. P. Construction Dewatering: New
wells. Indian Geotechnical Journal, 1973, 217±224. Methods and Applications. Wiley, New York,
10. P O W R I E W. and P R E E N E M. Time±drawdown 1992, 2nd edn.
behaviour of construction dewatering systems in 18. M A T T H E W S M. C., H O P E V. S. and C L A Y T O N C. R. I.
®ne soils. GeÂotechnique, 1994, 44, No. 1, 83±100. The geotechnical value of ground stiness deter-
11. P O W R I E W. Soil Mechanics: Concepts and Applica- mined using seismic methods. In Modern Geo-
tions. Spon, London, 1997. physics in Engineering Geology (McCann D. M.,
12. B U R L A N D J. B. and W R O T H C. P. Settlement of Eddleston M., Fenning P. J. and Reeves G. M.
buildings and associated damage. Proceedings of (eds)). Geological Society Engineering Geology
the British Geotechnical Society Conference on Special Publication No. 12, London, 1997, pp. 113±
Settlement of Structures, Cambridge, 1975, pp. 123.
611±654. 19. H E L M D. C. Field-based computational techniques
13. B O S C A R D I N M. D. and C O R D I N G E. J. Building for predicting subsidence due to ¯uid withdrawal.
response to excavation-induced settlement. ASCE In Man-induced Land Subsidence (Holzer T. L.
Journal of Geotechnical Engineering, 1989, 115, (ed.)). Reviews in Engineering Geology VI, The
No. 1, Jan. 1±21. Geological Society of America, Boulder, 1984, pp.
14. S K I N N E R H. D. and C H A R L E S J. A. Problems 1±22.
190