Estimation of Building Damage Due to Excavation-Induced

Ground Movements
Moorak Son1 and Edward J. Cording, M.ASCE2
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Building damage due to excavation-induced ground movement is evaluated using a damage criterion based on the average
state of strain in the distorting portion of the structure, and by considering the effect of building shear stiffness on the distortions imposed
by the ground settlement profile. Physical model tests and numerical simulations, correlated with case studies of building distortion and
damage, have been used to evaluate these relationships for masonry bearing wall structures. The distinct element method was used to
numerically model each masonry unit as a block, with the contacts between blocks having the stiffness and strength characteristics of
mortar. In-plane displacements at the corners of the wall sections permitted determination of the average state of strain, and the compo-
nents of rigid body tilt, angular distortion, lateral strain at the base, and the contribution of bending strain to the lateral strain in the upper
portion of the wall. The increase in angular distortion with increase in the ratio of ground/structure shear stiffness (decrease in building
shear stiffness) was examined for both elastic and cracked building walls. Cracking significantly reduced effective wall stiffness making
the wall more conformable to the ground settlement profile, which increased angular distortion, causing it to approach the distortion
(change in ground slope) that would occur in the absence of the structure.
DOI: 10.1061/(ASCE)1090-0241(2005)131:2(162)
CE Database subject headings: Buildings; Damage assessment; Ground motion; Excavation; Soil–structure interaction; Stiffness.

Introduction tests with appropriate scaling relations. If the estimated damage

level is unacceptable, protective measures should be considered.
Open cuts and underground excavations in a limited urban space These may include controlling construction procedures, changing
are gradually increasing in frequency because of the development construction methods, increasing support stiffness, improving the
and upgrade of infrastructures and the construction of new build- ground, reinforcing foundations and structures, or selecting a dif-
ings. At the same time, public concerns have risen over the effects ferent site. Repairing and replacing structures may also be con-
of excavation-induced ground movements on adjacent structures sidered.
and utilities. Nearby structures can be distorted and damaged by Relationships among ground movement and building distor-
ground movement, causing problems such as loss of invaluable tion and damage resulting from adjacent excavation were inves-
historic property, third party impact, construction delay, and sub- tigated in a Univ. of Illinois study, which focused on three tasks:
stantial increase of project cost. field observations, physical model testing, and numerical model-
A reliable damage estimation and appropriate protection are ing. Observation of full-scale structures remains the key to under-
required to avoid or minimize these problems. In general, build- standing behavior, but often distortion and settlement data for
ing damage estimation is performed with the following steps: (1) damaged structures are incomplete and comparisons with free-
estimation of free-field ground movement; (2) consideration of field movements unavailable. Physical and numerical modeling
effect of building stiffness on free-field ground movement, de- allows such comparisons, and also provides an opportunity for
fined as the movement for the ground if no building were present; controlled variation of parameters. Reported in this paper are re-
(3) estimation of building distortion based on soil–structure inter- sults of both physical and numerical modeling of the distortions
action; and (4) estimation of damage level. Damage levels can be and damage to 1/10th and field-scale masonry building walls ad-
estimated using criteria based on field observations and model jacent to a braced excavation.
The angular distortion/lateral strain criterion for building dam-
1 age due to excavation (Boscardin and Cording 1989) is evaluated
Senior Researcher, Korea Institute of Construction Technology, 2311
Daehwa-Dong, Ilsan-Gu, Goyang-Si, Gyeonggi-Do 411-712, South in terms of the state of strain at a point, considering the average
Korea. E-mail: moorakson@empal.com strain across a distorted portion of the building. The method is
2 applicable to a wide range of building length/height ratios and is
Professor, Dept. of Civil and Environmental Engineering, Univ. of
Illinois at Urbana-Champaign, 205 N. Mathews Ave., Urbana, IL 61801. used in this paper to evaluate the distortions of field case histories
E-mail: ecording@uiuc.edu and the physical and numerical model tests on masonry bearing
Note. Discussion open until July 1, 2005. Separate discussions must walls.
be submitted for individual papers. To extend the closing date by one The data from field observations, physical model tests, and
month, a written request must be filed with the ASCE Managing Editor.
numerical tests were collected and are compared with a damage
The manuscript for this paper was submitted for review and possible
publication on September 15, 2003; approved on June 1, 2004. This paper estimation method to verify the suggested method. These data are
is part of the Journal of Geotechnical and Geoenvironmental Engineer- also used to examine factors that affect building response, and to
ing, Vol. 131, No. 2, February 1, 2005. ©ASCE, ISSN 1090-0241/2005/ develop a relationship to consider soil–structure interaction. Rea-
2-162–177/$25.00. sonable damage assessments require a better understanding of the

162 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 were conducted by placing model brick-bearing walls or concrete
frames on spread foundations on the soil prior to excavation of an
adjacent braced wall. During excavation, ground movements were
monitored adjacent to the wall, in the free field away from the
building, and distortion and cracking of the building wall were
measured. Relatively large model scales of 1 / 4 – 1 / 10 were used,
which allowed realistic excavation and support sequences to be
followed, and which also minimized the required reduction in
building stiffness and strength to meet the necessary scaling rela-
tions. Fig. 1 shows the geometry of the physical model test.
To model the building walls, it was necessary to scale their
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

strength and stiffness with that of the frictional soil mass. Because
the strength and stiffness of the soil are controlled by confining
Fig. 1. Physical model test geometry pressure, which is proportional to the model scale, the stresses
applied to the soil by the model structure, and consequently, the
strength and stiffness of the structure, must also be reduced in
complex soil–structure interaction, with information from field proportion to the scale. For the brick-bearing walls, two-story
observations, model tests, and numerical tests. model structures were used, and the model size was 610 mm
This paper provides a background for establishing guidelines high⫻ 1,830 mm wide. The model bricks used 共14 mm⫻ 57 mm
to control building damage and develops a procedure for estimat- ⫻ 28 mm兲 were partially fired to obtain a reduced compressive
ing damage in nearby structures due to excavation-induced strength of 5.5 MPa. A reduced-strength lime mortar was used to
ground movements. The effects of soil–structure interaction are obtain the brick/mortar contact strength, which was in the range
considered using numerical models in which cracking and distor- of 14– 50 kPa tensile strength. The strength is equivalent to
tion of structures are modeled and correlated with field data and 140– 500 kPa for a full-scale structure and represents the strength
physical model test data. of type N mortar, which has an intermediate strength among other
mortar types (M, S, N, and O). The elastic modulus of brick units
was approximately 689 MPa.
Physical Model Tests on Brick-Bearing Walls Before testing, some of the mortar joints in the brick-bearing
Adjacent to Braced Excavations walls had pre-existing visible and invisible cracks due to the pro-
cess of installation and curing. In a brick-bearing wall structure,
Model tests were conducted at the Univ. of Illinois’ Schnabel the pre-existing cracks can strongly affect the response and dam-
Large Soil Model Test Lab which houses a 3 m ⫻ 4.5 m ⫻ 5.5 m age due to excavation-induced ground movement. Based on the
soil test pit, overhead crane, and soil handling facilities. Tests compression and shear tests on brick/mortar samples, the normal

Table 1. Classification of Visible Damage to Walls with Particular Reference to Ease of Repair of Plaster and Brickwork or Masonry [after Burland et al.
(1977)]
Approximate
Degree of crack widthb
damage Description of typical damagea (ease of repair is underlined) (mm)
Hairline cracks of less than about 0.1 mm are classed as negligible.
1. Very slight Fine cracks which can easily be treated during normal decoration. Perhaps ⬍1
isolated slight fracture in building. Cracks in external brickwork visible on
close inspection.
2. Slight Cracks easily filled. Redecoration probably required. Several slight fractures 1–5
showing inside of building. Cracks are visible externally and some
repointing may be required externally to ensure weathertightness. Doors and
windows may stick slightly.
3. Moderate The cracks require some opening up and can be patched by a mason. 5 – 15
Recurrent cracks can be masked by suitable linings. Repointing of external or a number of
brickwork and possibly a small amount of brickwork to be replaced. Doors cracks
and windows sticking. Service pipes may fracture. Weathertightness often 艌3
impaired.
4. Severe Extensive repair work involving breaking out and replacing sections of 15– 25
walls, especially over doors and windows. Windows and door frames but also depends
distorted, floor sloping noticeably. Walls leaning or bulging noticeably, some on number of
loss of bearing in beams. Service pipes disrupted. cracks
5. Very severe This requires a major repair job involving partial or complete re Usually ⬎25
building. Beams lose bearing, walls lean badly and require shoring. but depends on
Windows broken with distortion. Danger of instability. number of
cracks
a
In assessing the degree of damage account must be taken of its location in the building or structures.
b
Crack width is only one aspect of damage and should not be used on its own as a direct measure of it.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 163

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 Table 2. Angular Distortion with Respect to Soil/Building Stiffness for Brick-Bearing Walls with Ratio of Change in Ground Slope to Mortar Tensile
Strength
Maximum Maximum
principal strain Damage levelc crack size Damage levelc with
Case ␧ p共⫻10−3兲 with ␧ p (mm)a maximum crack size
PIT5E [Fig. 3(a)] 2.18 M–SE 1.0 M
PIT5ENb [Fig. 3(b)] 2.73 M–SE 1.2 M
N1 [Fig. 4(a)] 1.75 M–SE 0.96 M
N2 [Fig. 4(b)] 3.96 SE–VSE 1.89 SE
N3 [Fig. 5(a)] 1.52 SL 0.35 SL
N4 [Fig. 5(b)]
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

6.03 SE–VSE 3.25 VSE

N5 [Fig. 6(a)] 0.96 SL 6.60 M
N6 [Fig. 6(b)] 2.76 M–SE 12.0 M
a
For 1/10th model-scale, multiply by 10 to obtain field-scale crack width.
b
Numerical simulation for PIT5E.
c
N⫽negligible; VSL⫽very slight; SL⫽slight; M⫽moderate; SE⫽severe; VSE⫽very severe.

stiffness of brick/mortar joints was approximately 13.5 GPa/ m additional load. An additional load of 133.5 N and a wider foun-
and the shear stiffness of joints was in the range of dation were put at the front of the model structure to simulate the
350 MPa/ m – 1.65 GPa/ m. The friction angle of the brick/mortar downdrag effect of a façade wall connected to a bearing wall,
joint was 35°. Soil beneath the foundation was composed of clean which would tend to increase the degree of damage on the bearing
uniform sand and had a unit weight of 15.7 kN/ m3 in dry condi- wall. The total weight of the foundations was 890 N.
tions (Laefer 2001). The soil stiffness of the influence zone affected by the brick-
One of the 1/10 scale brick-bearing walls (PIT5E, Test 5, east bearing structure and foundations was determined and it was ap-
wall) was a two-story brick-bearing wall and was seated on mor- proximately 1.72 MPa (17.2 MPa at field scale). This soil stiff-
tar spread footings, which were 127 mm wide and 254 mm deep. ness is in the range of loose to medium sand.
The foundations were fully embedded in soil and firmly bonded Fig. 3(a) illustrates the measurements on the PIT5E model test
to the bearing wall. The brick-bearing wall was placed perpen- structure adjacent to the braced excavation that was excavated in
dicular to and 150 mm laterally from the future excavation (Fig. the model test pit (Fig. 1). The relatively flexible excavation wall
1). Wall footings were separated along the length and not tied or resulted in a free-field settlement profile at the foundation level
reinforced longitudinally. Three levels of tiebacks were tensioned having 5.5 mm of settlement at the wall and a differential ground
and locked off as the excavation was carried down to a depth of slope 共⌬GS兲 between the front third and center third of the build-
1.1 m. Building weight was 45.4 kg force, and an additional load ing of 2.25⫻ 10−3. Maximum lateral displacements near the ex-
of 854.4 N was put on the top of the structure. The first window cavation wall were 3 mm. (All lengths and displacements should
level and the bottom floor level were each subjected to 222.5 N of be multiplied by 10 for comparison with full-scale structures).

Fig. 2. Numerical model geometries and boundary conditions for 1/10th model-scale brick-bearing structures (dimensions are
multiplied by 10 for field scale)

164 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Distortion and cracking in model and numerical brick-bearing wall, 5E

The structure was noticeably cracked and deformed in the first buildings subjected to ground movements and to give justifiable
of the three bays as a result of the large change in slope of the basis for extended numerical tests which were performed under
free-field vertical ground movement between Bays 1 and 2. The controlled variation of parameters. The simulation was performed
maximum vertical movement of the structure was 4.46 mm, based on the properties of the model structures, which had been
which was similar to the maximum free-field vertical ground obtained from model sample tests and on the load conditions of
movement of 4.5 mm at the same equivalent location as the struc- the structures (Son 2003). Fig. 2 shows the geometry of the two-
ture. The maximum horizontal movement of the structure was and four-story numerical model using UDEC (Itasca Consulting
2.31 mm, which was slightly smaller than the free-field horizontal Group Inc. 2000).
movement of 3.0 mm and indicates that a slight slip occurred Foundations were not modeled physically in the numerical
between the structure and the ground. The structure was slightly simulations. Instead, the weight of each physical foundation was
tilted up at the back corners of the third bay due to the structure’s applied at the foundation level, in addition to the loads transferred
tilt. This behavior has also been observed in field structures built from the building. Side friction of the foundations was considered
on shallow foundations. The cracks were concentrated around by adjusting the properties of the interface between the building
window openings because of low confinement and high stress element and the soil mass element. In the physical model, the
concentration in these areas. Crack formation near window open- foundation located at the end of the wall nearest the excavation
ings is frequently observed in full-scale structures adjacent to was made twice as wide as other foundations and additional load
open cutting or tunneling. Some cracks in the model test structure was placed on the end of the wall to simulate the downdrag effect
were attributable to curing and installation. In addition, material of an adjacent façade wall. Thus, in the numerical model, the
nonuniformity induced some local cracks in the structure. The shear stiffness at the interface between the soil mass and the test
maximum crack width was 1.0 mm (10 mm at field scale). walls, for the first foundation, was selected to have much higher
Damage level, determined from both generalized state of strain values than the shear stiffness at the interfaces where other foun-
criterion (Fig. 8) and maximum crack size criterion (Table 1), is dations were located, and additional load was applied along the
“moderate” (Table 2). front edge of the wall to simulate the downdrag effect.
The soil beneath the wall was modeled elastically (Fig. 2),
Comparison of Physical Model Test and Numerical with a stiffness selected to provide the same pressure/
Simulation displacement relation for the two-dimensional (2D) model as that
for a 127 mm wide footing on an elastic foundation with a
The physical model test situation described above was subjected Young’s modulus matching that of the soil in the influence zone
to numerical simulation to provide insights into the behavior of below the footing. The lateral and vertical displacement profile

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 165

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Numerical analyses of model brick-bearing wall, varying brick/mortar tensile strength

observed in the free-field for the physical model was applied to loading with the same weight as with the model test structures.
the elastic soil mass beneath the model (Fig. 2). A vertical dis- After the initial stress was assigned to the soil mass, initial equi-
placement profile was applied to the bottom of the soil mass after librium was obtained with building self-weight and additional
confirming the displacement profile created at the top of the soil loads applied except for the façade load. At this stage, the bound-
mass for the free-field condition. A horizontal displacement pro- ary condition for the soil mass was roller supports at each end of
file was applied to the top of soil mass interconnected with the the two vertical boundaries of the soil mass and at the bottom
building wall at the soil/structure interface. The horizontal dis- boundary of the soil mass. All displacements were reset to zero
placement was applied at the top, not the bottom, of the soil mass and the boundaries at the soil mass were made free before a
because application of the horizontal displacement at the bottom free-field ground movement that was measured from the test pit
of the elastic soil mass would produce a distorted horizontal dis- was imposed to the soil mass, and at the same time, the downdrag
placement at the top, for the free-field condition. However, the force due to an adjacent façade wall, which is induced by ground
soil–structure interaction for horizontal ground movement was settlement below the wall, was applied. After applying the free-
still allowed because an equivalent soil shear stiffness was ap- field ground movement and façade load, the boundaries at the soil
plied at the interface element. The shear stiffness at the interface mass were roller supported and final equilibrium condition was
was selected to provide the same horizontal pressure/horizontal ensured. Then, distortions and cracking in the walls were exam-
displacement relation for the 2D model as that for the foundation ined.
soil (Son 2003). Thus, the soil–structure interaction between the The results of the numerical simulation are summarized for
building wall and the soil mass was allowed both for vertical deformations and crack patterns in Fig. 3(b) and compared with
settlement and for horizontal ground displacement. The model the physical model test in Fig. 3(a). In this, and in other cases, the
used an interface frictional resistance that was adjusted to account overall structure deformations, crack patterns, and crack width
for the horizontal side friction of the foundations that were not distributions were similar between numerical and physical models
explicitly modeled. Tensile capacity between foundation and soil having similar properties. However, small changes in load distri-
due to vertical side friction of the foundation was also modeled. bution and minor variations in the properties of the brick/mortar
Thus, the model allowed for interaction between the soil mass and contacts will cause the crack formation and opening to occur in
a structure for the given ground movement profiles. Son (2003) different locations. Such conditions occur in real structures and
described the approach and methodology of numerical analyses in may occur in the model because of the variation of properties
more detail. through the physical model test structures and the initial cracks
At the beginning of the numerical analysis, an initial stress is due to shrinkage and disturbance during curing and installing the
assigned to the soil mass to simulate the preloading performed in model structures. The numerical simulations were performed
the test pit. The initial stress was obtained by simulating the pre- under the assumption of uniform strength and stiffness for the

166 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Numerical analyses of four-story brick-bearing wall, varying soil stiffness

mortar and for the masonry units throughout the model structures reached, the tensile strength or cohesion is lost, but frictional
with no initial cracks. resistance remains. Further extension across a contact causes
Fig. 3 is one example of the good agreement that was found separation of the bricks and opening of a crack. In these cases,
between numerical and physical models. The numerical model additional forces were applied at the end of the wall nearest the
was able to simulate both the distortions and the crack widths and excavation to simulate the effect of downdrag forces distributed
patterns exhibited in the models and in actual case histories. The from the façade wall to the bearing wall.
correlations confirm the suitability of the numerical model as a Damage was concentrated in the first through the second or
base for the parametric studies described in the following section. third column of windows, where the distortions and lateral strains
in the foundation were the highest. Two-story masonry walls are
illustrated in Fig. 4. In Fig. 4(a), with mortar having 55 kPa
Distinct Element Analyses for Two- and Four-Story model tensile strength (equivalent to a high strength mortar), the
Brick-Bearing Walls

The response of structures to ground movements is influenced by

Table 3. Angular Distortion with Respect to Soil/Building Stiffness for
building stiffness and brick/mortar tensile strength as well as soil
Brick-Bearing Walls with Ratio of Change in Ground Slope 共⌬GS兲 to
stiffness. In general, building stiffness increases with the number Mortar Tensile Strength 共␧t兲
of stories.
The effect of soil stiffness, number of stories, and brick/mortar Case EsL2 / GHb ⌬GS/ ␧t ␤ / ⌬GS
tensile strength is investigated for two- and four-story structures. PIT5E [Fig. 3(a)] 34.95 3.36 0.84
Fig. 2 shows the geometry of model structures for numerical tests. PIT5ENb [Fig. 3(b)] 34.95 3.36 1.02
Figs. 4 and 5 present the results of the 2D distinct element N1 [Fig. 4(a)] 11.02 4.00 0.47
numerical analysis (UDEC) for 1/10th model-scale two- and four- N2 [Fig. 4(b)] 11.02 10.00 1.06
story brick-bearing walls subjected to a given ground movement. N3 [Fig. 5(a)] 5.93 8 0.18
Each brick was modeled as a separate block and the stiffness and
N4 [Fig. 5(b)] 23.74 8 1.08
strength of the contacts between blocks were selected to provide
N5 [Fig. 6(a)] 6.20 8 0.71
reasonable values of building stiffness and strength (Son 2003).
N6 [Fig. 6(b)] 6.20 8 1.13
When the tensile or cohesive shear strength of the mortar is

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 167

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Numerical analyses of four-story brick-bearing wall, varying façade downdrag force

building tended to flatten out the ground settlement profile, pro- change in free-field ground slope 共␤ / ⌬GS兲 is low (0.18), and the
ducing relatively more tilt and less angular distortion 共1.3 ground settlement curve is flattened. Damage level, based on both
⫻ 10−3兲 than would have occurred had the building moved with angular distortion and lateral strain, as illustrated in Table 2, is
the settlement profile. The ratio of angular distortion to change in slight. Damage determined from the maximum crack width is also
free-field ground slope 共␤ / ⌬GS兲 was relatively low (0.47). In in the slight category, using the crack widths described in the
Fig. 4(b), with mortar having 27.5 kPa tensile strength (equiva- damage criterion by Burland et al. (1977), scaled to 1/10 to fit the
lent to an intermediate strength mortar), the building tended to model scale. (It is recognized that crack width is only one indi-
move more closely with the ground settlement profile and angular cator of the damage level in the criterion, which is also related to
distortions in the model increased to 2.7⫻ 10−3, and the ratio of other aspects of the functional and structural performance of the
angular distortion to change in free-field ground slope 共␤ / ⌬GS兲 building). The lateral strain at the base of the structure is larger
was high [1.06: it is slightly larger than unity because structure than at the top, which is expected for small distortions and a stiff
load and stiffness modified the ground settlement profile a bit, see structure.
Fig. 4(b)]. For the 55 kPa tensile mortar strength [Fig. 4(a)], the In Fig. 5(b), the ground stiffness beneath the four-story build-
lateral strain was concentrated at the foundation level ␧lat共F兲, with ing has been increased (equivalent to a soil stiffness of 70 MPa
little lateral strain in the top floor, whereas, for the 27.5 kPa ten- for a full scale structure), causing the relative building/soil stiff-
sile strength [Fig. 4(b)], the lateral strain increased in the top, and ness to be low and the building to distort more closely with the
cracking occurred in the upper floors. Damage level, based on free-field ground settlement profile 共␤ / ⌬GS= 1.1兲. The resulting
both angular distortion and lateral strain as illustrated in Table 2, angular distortion is much higher 共3 ⫻ 10−3兲 than in Fig. 5(a).
was moderate for the Fig. 4(a) case and severe for the Fig. 4(b) Cracks extend through the full height of the structure, with the
case. largest cracks at the second window line. Lateral strains at the top
Four-story brick walls, as shown in Figs. 5(a and b), were of the wall are much higher than those at the base. Damage level,
significantly stiffer than the two-story walls. In Fig. 5(a), the soil determined from both the measured ␧max and the measured crack
stiffness and brick/mortar contact strength is the same as that of width is very severe (Table 2).
the two-story building in Fig. 4(b), but the lower L / H ratio for the The normalized angular distortion, ␤ / ⌬GS, increases with in-
four-story building causes it to be stiffer than the two-story build- creasing ratio of soil/wall shear stiffness 共EsL2 / GHb兲, where L is
ing. Thus, the building undergoes relatively more tilt and less the deformed length of the building, H is building height, and b
angular distortion 共0.5⫻ 10−3兲, the ratio of angular distortion to its wall thickness, Es is the elastic soil modulus in the influence

168 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 zone below the footing, and G is the uncracked elastic shear most of the cracks concentrated at the first row of windows while
modulus for a section of masonry wall (Table 3). The relationship in the case with no façade loading the cracks are concentrated at
is described in more detail below. the second row of windows.
As the soil to building shear stiffness 共EsL2 / GHb兲 increases, For the structure with façade load, damage level, determined
there is less tilt for a given settlement slope and the ratio of from both the measured ␧max and the measured crack width is
angular distortion to the change in the free-field ground slope moderate (Table 2). However, for the structure without façade
共␤ / ⌬GS兲 increases. However, the effective stiffness of the wall load, damage level from the measured ␧max is slight and damage
cannot be described by elastic stiffness alone, because it is determined from the maximum crack width is in the moderate
strongly controlled by the extent of cracking. For a given value of category 共6.6 mm兲, but near the 5 mm boundary between slight
soil/building stiffness, as the strength of the brick/mortar contact and moderate damage.
decreases or the ground distortion applied to the structure in-
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

creases, the building undergoes more extensive cracking causing

␤ / ⌬GS to increase toward one. Table 3 indicates that ␤ / ⌬GS is
Damage Criterion Based on State of Strain Theory
approximately one for stiff soils (Es⬃ 70 MPa for full scale) and
low brick/mortar contact strength. Lower values of ␤ / ⌬GS were
obtained for various combinations of higher tensile strength, During construction of the Washington, D.C. Metro, building
higher (four-story) walls and softer soils. damage resulting from tunneling and excavation was observed to
In addition to the 1/10th model-scale numerical tests, two be greater than would be predicted for buildings settling under
field-scale numerical tests were performed for four-story brick- their own weight. Accordingly, a damage criterion was developed
bearing walls. The purpose of these numerical tests was to inves- in which the critical strain for a given damage level was deter-
tigate the effect of façade downdrag force on bearing walls. The mined from the combined effect of lateral strain induced by lat-
ground stiffness beneath both structures is 68.9 MPa and the eral ground displacements and angular distortion of the structure
structures have a tensile and cohesive strength of 344 kPa. Figs. (Boscardin and Cording 1989). Damage levels were described by
6(a and b) and show the results of the numerical tests. As shown Burland et al. (1977) and Burland (1995) presented a criterion
in the figures, the structure with façade load (0.11 times building that related deflection ratio and lateral strain. Boone et al. (1999)
wall self-weight) underwent more severe angular distortion evaluated damage in terms of cumulative crack width determined
共2.25⫻ 10−3兲 and damage, while the structure without façade load from strains.
flattens the ground settlement curve and undergoes relatively The damage criterion developed by Boscardin and Cording
more tilt and less angular distortion 共1.41⫻ 10−3兲. Table 3 indi- (1989) considered angular distortion 共␤兲 and horizontal strain
cates that for the structure with no façade load, ␤ / ⌬GS is lower 共␧h兲. The criterion was based on the state of strain of a simple
(0.71) than the 1.13 value for the structure with façade load. deep beam with L / H = 1, E / G = 2.6, and neutral axis at the bottom
The façade downdrag results in concentration of cracks closer of the beam. With little change from that established by Boscardin
to the end of the bearing wall: the case with façade loading has and Cording (1989), the damage criterion has been described and

Fig. 7. State of strain at point or average state of strain in distorting portion of structure

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 169

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Comparisons between damage estimation criterion and damage levels resulting from field observations, physical model tests, and
numerical parametric studies

updated to a generalized state of strain damage criterion. The “slight” was established at a critical tensile strain of 1.67⫻ 10−3,
generalized damage criterion is based on the state of strain at a which corresponds to an angular distortion of 1/300 for no hori-
point (Fig. 7), which is not dependent on L / H, E / G, and the zontal strain. This is the angular distortion that Skempton and
position of neutral axis (Cording et al. 2001; Son 2003). MacDonald (1956) and Bjerrum (1963) denoted as the threshold
Using this criterion, the average state of strain in a building for first cracking in panel walls and load-bearing walls for struc-
unit is evaluated. The ground movements are imposed on the base tures settling under their own weight. The upper bound of the
of the structure, producing lateral strain (extension in the outer zone in which damage is considered “moderate to severe” was
portion of the settlement trough) and angular distortion, or shear established at a critical tensile strain of 3.33⫻ 10−3, which corre-
strain, as shown in Fig. 8. A building unit can be a section be- sponds to an angular distortion of 1/150 for no horizontal strain.
tween two columns or cross walls, two different building geom- Skempton and MacDonald (1956) and Bjerrum (1963) found that
etries or stiffnesses, or two different ground displacement gradi- the angular distortion of 1/150 corresponds to the threshold angu-
ents. It is usually the portion of the building closest to the lar distortion for severe cracking and structural damage in struc-
excavation and subject to the largest distortions. Because ground tures settling under their own weight. The lateral strains were also
movements are imposed at the base of the structure, lateral strains matched to the lateral displacement criterion developed by the
resulting from the imposed lateral displacements are greatest near National Coal Board for deep mines in the United Kingdom,
the base of the structure. Bending may produce higher lateral which are subject to lateral strains but, because of the large width
strains in the upper portion of the structure, which can only occur of the settlement troughs, differential settlements result in tilt
for low building length/height ratios if vertical cracks or weak- rather than distortion (NCB 1975).
nesses extend over the height of the wall. The combination of angular distortion ␤ and lateral strain ␧L at
The criterion is based on the concept that a structure is de- a point or in a building unit results in a maximum principal tensile
formed by the combination of angular distortion and lateral strain, strain ␧ p as follows (Fig. 7);
and the maximum strain on the structure can be decided by a
principal strain formed by both the angular distortion and the ␧ p = ␧Lcos ␪2max + ␤ sin ␪maxcos ␪max
lateral strain. The maximum principal strain is compared with the
critical strains for each different damage category, which are de-
scribed below, to estimate potential damage levels. Angular dis- ␤
tan共2␪max兲 =
tortion and lateral strain for building damage estimate can be ␧L
determined by measuring vertical and horizontal displacements at
the corners, A, B, C, and D of a building unit as shown in Fig. 9. where ␪max⫽direction of crack formation and the angle of the
The critical tensile strains for different damage levels were plane on which ␧ p acts, measured from vertical plane. For ex-
determined from the results of field observations. In Fig. 8, the ample, if ␤ is zero, ␧ p = ␧L acts on the vertical plane 共␪max = 0兲 and
critical tensile strain boundaries of the zone classified as “very a crack forms along the vertical plane, and if ␧L is zero, ␧ p
slight” damage were taken as 0.5⫻ 10−3 and 0.75⫻ 10−3, which = 1 / 2␤ acts on the plane at ␪max = 45° and a crack forms at ␪max
are the strains suggested by Burland and Wroth (1974) and = −45°. Because the damage estimate criterion is based on the
Polshin and Tokar (1957) for the formation of a first visible crack. combination of the field observations and the state of strain
The upper bound of the zone in which damage is considered theory, it is a semiempirical criterion. Table 4 summarizes the

170 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Numerical analyses of two-story brick-bearing walls, 1/10th model scale

critical tensile strains corresponding to the various degrees of Angular distortion 共␤兲 is the shearing distortion of the section
damage (see Fig. 8). and defined as

␤ = Slope − Tilt
Determination of State of Strain in Building Wall Lateral strain at top 关␧lat共T兲兴 is the change of lateral displace-
ment at the top over the length L of the section and defined as
The strain in the plane of a building wall is determined by mea-
suring the vertical 共v兲 and lateral 共l兲 displacements (Av, Bv, Cv, Dl − Cl
␧lat共T兲 =
Dv, Al, Bl, Cl, and Dl) at the four corners (A, B, C, D) of a section L
of the wall [Fig. 3(a)]. Typically, the angular distortion and lateral
strains were determined for the section of wall in the first third to Lateral strain at base 关␧lat共F兲兴 is the change of lateral displace-
one half of the base length (Bay 1) where the damage was con- ment at the base over the length L of the section and defined as
centrated. From these measurements, the following terms are de- Al − Bl
fined for use in description of the deformation of the section. ␧lat共F兲 =
Slope is the change of gradient at base over the length L of the L
section and is defined as A combination of tilt measurements and settlements can pro-
vide the information needed to determine angular distortion. Pre-
Av − Bv cise tape survey can be used to measure lateral strain along a
Slope =
L structure. The other option is total station measurements from
Tilt is the rigid body rotation of the section and defined as prisms on both lower and upper portions of the structure to deter-
mine settlement, lateral strain, and tilt.
共Cl − Bl兲 + 共Dl − Al兲 The full state of strain at a point would include vertical as well
Tilt = as lateral strain. However, for a building subjected to ground
2H
movements, vertical strains are usually small, and any significant
or vertical strain that takes place is likely to be induced by a hori-
zontal crack near the edge of the wall, causing the upper wall to
Cl − Bl have a lower differential vertical displacement than the base of
H the wall. Thus damage due to differential vertical displacement

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 171

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 Table 4. Critical Tensile Strains Corresponding to Various Damage critical strains for visible cracking can differ among structures
Levels because of age and structural details.
Damage level Critical tensile strain, ␧c共⫻10−3兲 In conclusion, the generalized state of strain damage criterion
estimated building damage reasonably well for the field cases,
Negligible 0 – 0.5
physical model tests, and numerical tests. The criterion also had a
Very slight 0.5– 0.75 good agreement with the “ease of repair” damage criterion, which
Slight 0.75– 1.67 uses maximum crack width, for damage levels beyond easily re-
Moderate to severe 1.67– 3.33 paired aesthetic damage levels.
Severe to very severe ⬎3.33 The generalized state of strain damage criterion considers tilt-
ing of structures with angular distortion and postcracking behav-
ior with the state of strain at a point or an average strain across a
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

across a portion of the wall can be accounted for by considering building unit.
only the angular distortion at the base of the structure, without In addition, some cases with damage levels classified as “se-
averaging in the lesser angular distortion at the top. vere” based on maximum crack widths were observed from the
The nonhomogeneous state of strain in the structural element numerical parametric studies in the upper range of strains for the
and the loss of continuum behavior as cracks open require some “moderate to severe damage” zone in the generalized state of
judgment in the assessment of the state of strain. For example, for strain damage criterion (see Fig. 8). The strain for the “moderate
a simply supported beam with a low H / L ratio, the maximum to severe damage” range may induce severe distortion of door or
shear strains will occur in a section away from the maximum window frames or the reduction of joist bearings, which may
bending strains so that the two are not additive. However, for a cause joist bearing failure instead of cracking in some structures.
building with a high H / L ratio, the bending strains, if they do O’Rourke et al. (1976) found that a lateral extension strain of
extend to the top of the wall as a result of pre-existing cracking, approximately 1/100 is sufficient to cause bearing failure at wall–
may be additive to the angular distortions and, in fact, are likely joist connections. This strain is equal to a relative lateral displace-
to cause opening of the cracks that initially formed as a result of ment of 61 mm for two bearing walls that are originally spaced
angular distortion, as occurred in several of the model and nu- 6.1 m apart. When a relative rotation between two bearing walls
merical tests in this paper. is involved in wall separation, the critical condition for bearing
failure is a function of wall height and stiffness of bearing wall
and restraint at the wall–joist connections. A relative rotation of
Comparison of Building Damage between approximately 1/200 between two bearing walls may be sufficient
Measurements and Damage Estimation to lead to the critical condition at the roof level of a four-story
building. The relative rotation of 1/200 induces a relative lateral
Including the physical model and numerical tests in this paper, displacement of 61 mm between two bearing walls at the roof
Son (2003) summarized the results of 18 field observations (Bos- level of the four-story building, assuming the height of each story
cardin and Cording 1989), 2 physical model tests, and 122 nu- to be 3 m. These observations confirm and justify that severe
merical tests and compared damage estimates from the general- damage may occur in the upper portion of the “moderate to severe
ized state of strain damage criterion with observed damage levels damage” range from which Burland (1995) removed severe dam-
from the results (Fig. 8). The observed damage levels are based age level because none of the cases examined by Boscardin and
on the building conditions and cracks for the field data, and on Cording (1989) in that range showed severe damage.
measured and calculated maximum crack widths for the physical Deflection ratios (⌬ / L, see Fig. 9) were obtained from the
model tests and the numerical tests, respectively. The damage physical model and numerical tests in this paper and compared
categories based on crack width were chosen to compare with the with angular distortions for each case (Table 5). The deflection
damage estimates from the generalized state of strain criterion ratios were measured from the settlement profile for the part of a
because the crack width can provide direct indication for practical structure subjected to a free-field settlement trough.
and visual building damage (Burland 1995; Boone 1996; Boone Son (2003) summarized the ratios 关␤ / 共⌬ / L兲兴 between the an-
et al. 1999). However, structural damage can also be evidenced gular distortion 共␤兲 and the deflection ratio 共⌬ / L兲 of physical
by occurrences other than cracking, such as jammed doors or loss
of joist bearing. In addition, tilting or small differential ground
movement can affect sensitive machinery or utility pipe connec- Table 5. Ratios between Angular Distortion 共␤兲 and Deflection Ratio
共⌬ / L兲 of Physical Model and Numerical Tests
tions. These effects should be also considered for building re-
sponse problems. ␤a ⌬ / Lb
It is found that the damage levels observed from the field cases Case 共⫻10−3兲 共⫻10−3兲 ␤ / 共⌬ / L兲c
are in good agreement with the damage estimated by the gener- PIT5E [Fig. 3(a)] 1.88 0.55 3
alized state of strain damage criterion. A good agreement is also PIT5ENb [Fig. 3(b))] 2.29 0.59 4
found between the damage classification based on maximum N1 [Fig. 4(a)] 1.31 0.30 4
crack width and the generalized state of strain damage criterion
N2 [Fig. 4(b)] 2.82 0.79 4
for crack widths wider than 7.5 mm. But for the crack widths
N3 [Fig. 5(a)] 0.47 0.14 3
narrower than about 7.5 mm, the damage classification based on
N4 [Fig. 5(b)] 2.99 0.75 4
maximum crack width tends to give higher damage levels than
those based on the generalized state of strain criterion. As Boone N5 [Fig. 6(a)] 1.41 0.41 3
et al. (1999) pointed out, these differences are probably attribut- N6 [Fig. 6(b)] 2.25 0.56 4
a
able to the fact that, particularly when only crack widths are only Angular distortion.
b
used as a damage level indicator, an evaluation of damage sever- Deflection ratio.
c
ity for slight cracking in a structure is more subjective. Further, Ratio of angular distortion to deflection ratio.

172 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Estimation and validation of equivalent building stiffnesses

model tests, numerical tests, and field observations. For all cases
investigated, the ratio was in the range of 2 and 4. It was indicated
that for elastic (including frame structures) and for minor crack-
冉 EsoilL2
GbuildHb
冊
ing conditions (Fig. 9), the ratio was in the range of 2, but for where Esoil⫽soil stiffness in the region of footing influence;
severe cracking in a structure, the ratios increased up to 4. The L⫽length of building portion subjected to ground movement;
low ratio of 2 indicates that the distortion in a structure is less Gbuild⫽elastic shear modulus of a building; H⫽height of a build-
than change in ground slope 共⌬GS兲 that would occur in the ab- ing; and b⫽building wall thickness. Increasing values of soil/
sence of the structure, while the high ratio of 4 indicates that the structure shear stiffness (decreasing structure stiffness) cause the
distortion in a structure is close to the change in ground slope. building distortion to conform more closely to the ground settle-
ment profile, and the ratio of ␤ / ⌬GS to approach one.
The building will also become more compliant with the
Interaction between Soil and Structure ground settlement profile as cracks in the masonry mortar form
and open. Cracking will be a function of the magnitude of the
This study proposes an approach for considering soil–structure strain imposed on the structure with respect to the strains required
interaction of masonry structures in estimating building damage to initiate cracks and is quantified in terms of the change in
due to excavation-induced ground movement. ground slope with respect to structure cracking strain 共⌬GS/ ␧t兲.
As indicated previously, to estimate building damage based on The change in ground slope ⌬GS is determined from the free-
the generalized state of strain criterion, angular distortion, and field ground movement slopes at the location of two adjacent
lateral strain need to be determined. If the angular distortion and units, one of which is the unit for damage estimation, and the
lateral strain are determined from a free-field ground movement other being the unit adjacent to it. ⌬GS is obtained by subtracting
profile, it may result in too conservative a damage estimate be- the slope of the latter from the slope of the former.
cause a pre-existing building may modify the free-field ground For a building that is composed of staggered bricks or blocks
movement profile. In other words, if a building is stiff and strong and has openings, the shear modulus, Gbuild can be determined
enough to modify a free-field ground movement profile, the esti- from relationships developed from numerical tests (Fig. 10).
mate based on the free-field ground movement profile can cause Son (2003) summarized the results of 4 field case studies, 3
unnecessary and expensive protection measures. This condition physical model tests, and 209 numerical tests in terms of the
leads to a conclusion that building response should be evaluated normalized angular distortion ␤ / ⌬GS the relative soil–structure
based on soil–structure interaction to provide more reasonable shear stiffness EsoilL2 / GbuildHb and the ratio between change in
damage estimates. ground slope and structure cracking strain ⌬GS/ ␧t.
Fig. 11 shows the relationship among the normalized angular
distortion, the relative stiffness, and the ratio between ground
Relative Soil/Structure Stiffness movement slope change and structure cracking strain. The results
from field observations, physical model tests, and numerical
A relationship was developed to consider the effect of soil– model parametric studies are plotted on the figure, together with
structure interaction in reducing the ratio of angular distortion 共␤兲 the results of elastic analysis. The figure clearly shows four
in a building to change in ground slope 共⌬GS兲 in a free-field trends. One trend is that the value ␤ / ⌬GS gradually increases as
ground settlement profile (Son 2003). The relationship is a func- the relative soil/structure shear stiffness increases (and the struc-
tion of the relative soil/structure shear stiffness ture shear stiffness decreases) if no cracks, or minor cracks, form

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 173

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Relationship between or among relative soil–structure stiffness, normalized angular distortions, ratios between ground slope change, and
structure cracking strain

in a structure. The building conforms more closely with the free- Building Damage Estimation Procedure
field ground movement slope when the building is very flexible,
and the value of GbuildHb is small. In that case, the building will Protection of adjacent or overlying structures occupies a major
conform to a free-field ground movement and ␤ / ⌬GS will ap- part of the cost, schedule, and third-party impacts of urban under-
proach one. The second trend is that when a building is cracked ground construction. Methods used to limit damage or mitigate
extensively, the effective building stiffness is no longer described their effects include siting to avoid the structure; limiting move-
by elastic stiffness alone, because the extent of cracking strongly ments at the excavation source with controlled construction mea-
controls the effective building stiffness. The value of ␤ / ⌬GS in sures; modifying the ground or using compensation methods to
this case becomes close to one. For moderate cracking, the value replace ground loss; and underpinning, reinforcing, repairing, or
of ␤ / ⌬GS is between values for no or minor cracking and values replacing structures. Predicting and then monitoring building per-
for extensive cracking. The third trend is related to structure formance is a key aspect in planning, design, and implementation
strength and shows that the value of ␤ / ⌬GS increases as the ratio of such measures.
between ⌬GS and ␧t (structure cracking strain), ⌬GS/ ␧t, in- There are several levels in evaluating the effect of ground
creases for the same relative stiffness. In other words, the greater movements on building performance (Cording et al. 2001; Son
the change in ground slope and the weaker the structure, the 2003). An initial screening determines the potential affected struc-
larger the value of ␤ / ⌬GS. The fourth trend is related to the tures along the project alignment by considering the width, maxi-
effect of downdrag force from an adjacent façade wall. The trend
shows that a structure subjected to a downdrag force has higher
␤ / ⌬GS for the same relative stiffness.
Lower values of ␤ / ⌬GS were obtained for the various combi-
nations of higher-story buildings, higher tensile strength, fewer
window openings, smaller change in ground slope, less downdrag
force, and stiffer structures with respect to the soil.
Comparing the field observations and physical model tests
with the results based on numerical parametric studies, good
agreement was found for the relationship with some degree of
conservatism. Therefore, the normalized relationship was vali-
dated with the limited data and can be extended to estimate the
responses of buildings to ground movements other than the cases
examined.
Boscardin and Cording (1989) investigated the effect of grade
beams to reduce horizontal ground movement-induced lateral
strain in a structure (Fig. 12). The greatest benefit was found
when going from no grade beams to a light grade beam. The
relationship can be used to consider the effect of grade beams on
strain in a structure induced by lateral ground movement due to
open cutting. Fig. 12. Effect of grade beams for two-story and three-bay structures

174 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 Table 6. Typical Values of Maximum Building Slope and Settlement for Damage Risk Assessment [after Rankin (1988)]
Maximum slope of Maximum settlement of
Risk category building building (mm) Description of risk
1 Less than 1/500 Less than 10 Negligible: superficial
damage unlikely
2 1 / 500– 1 / 200 10– 50 Slight: possible superficial
damage which is unlikely to
have structural significance
3 1 / 200– 1 / 50 50– 75 Moderate: expected
superficial damage and
possible structural damage
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

to buildings, possible
damage to relatively rigid
pipelines
4 Greater than 1/50 Greater than 75 High: expected structural
damage to buildings.
Expected damage to rigid
pipelines, possible damage
to other pipelines

mum settlement, and ground slopes of the free-field trough. For 2. Determine if the structure is within the zone of influence of
example, Rankin (1988) notes negligible damage would be ex- the ground movement profile.
pected for slopes less than 1/500 and settlements less than 10 mm 3. If so, determine the maximum settlement and settlement
(Table 6). slope expected across the structure (Table 6).
A second level of investigation is to consider the location of 4. For most structures, if the maximum settlement or slope is
the building and its width with respect to the settlement trough below the limits, the investigation may finish at this step,
and estimate lateral strains and distortions by imposing the antici- although it is prudent to gain additional perspective on the
pated ground displacements on the building. Making the conser- magnitude of the actual distortions that could develop by
vative assumption that the structure is flexible enough to be com- proceeding to the next step. If the maximum settlement or
pliant with the ground is most appropriate for masonry structures slope is above the limits, proceed to the second estimation
that are not tied or reinforced. Distortion can be estimated by phase.
integrating curvature across a length of a structural unit, but in
practice it is appropriate to consider differential slopes between
Second Estimation Phase
structural units, recognizing that the structure is stiff enough to
distribute loads and smooth out locally high or abrupt curvatures. 1. Use the free-field ground movement and assume that a struc-
A further level of investigation is to consider the ability of the ture conforms with the free-field ground movement.
building to modify ground movement patterns; in particular, to 2. Determine the change in ground slope between adjacent sec-
consider the axial stiffness of grade beams or the structural frame
which limits lateral strains, and the shear and bending stiffness of
the building with respect to the soil which causes the angular
distortion of the structure to be lower than the distortions deter-
mined from the free-field settlement profile. Additionally, the ge-
ometry and variations in stiffness and strength of the structure,
and its tolerance for different levels of distortion and damage,
should be considered.
From the outlined points, the building damage estimation pro-
cedure shown in Fig. 13 was developed. The procedure is de-
scribed below in detail.
The procedure starts from a conservative screening that elimi-
nates from further study buildings that do not fall within a sig-
nificant damage category using even the conservative estimate.
Buildings that exceed the acceptable damage using the conserva-
tive estimate category are investigated further. This procedure al-
lows a number of buildings in urban areas to be investigated in a
limited time.

First Estimation Phase

1. Estimate a free-field ground movement, both vertical and
horizontal movements from empirical or semiempirical rela-
tionships, physical model tests, or numerical tests for a given Fig. 13. Procedure for building damage estimate to excavation-
excavation. induced ground movement

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 175

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 tions based on column spacing, footing spacing, cross-wall ment increases distortion near the end of the bearing wall and
spacing, or ground movement gradient. Assume that the causes cracks to concentrate near the façade wall or the first
change in ground slope is equal to angular distortion. column of windows. For a structure without any lateral re-
3. Determine the horizontal (lateral) free-field ground move- straint, opening of cracks in bending can be induced in upper
ment strain between adjacent sections based on column spac- floors if a weak joint exists or if cracking continues to de-
ing, footing spacing, cross-wall spacing, or ground move- velop upward. For wide settlement troughs, as the ground
ment gradient. movement extends further out beneath the building, the
4. If calculated shear distortions are high throughout the height building may begin to experience bending deformation and
of the structure, or if major pre-existing vertical joints or crack development at the top of the building due to large
weaknesses are present in the structure, determine the bend- L / H (where L⫽building length subjected to ground move-
ing (lateral) strain at the top of a building unit, using the ment and H⫽building height).
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

height of the building unit and the radius of a curvature of 2. Cracks initially form around opening areas and propagate
ground movement. further with increasing ground movement. Crack formation
5. Using the angular distortion and lateral strains, determine the decreases building stiffness and increases the tendency for a
damage level from the generalized state of strain damage building to conform more closely to the free-field ground
criterion (Fig. 8). movement profile. As distortions increase, cracking strongly
6. If the damage level from the state of strain criterion is ac- controls building response to ground movement, and elastic
ceptable, finish at this step. If it is unacceptable, proceed to analysis alone can produce misleading results about building
the final estimation phase. response. At small distortions, with little cracking, the stiff-
ness of the structure remains high and the angular distortion
Final Estimation Phase with respect to the change in ground slope, ␤ / ⌬GS is low.
As distortions increase, cracks extend and grow so that
1. Use the change in ground slope of the free-field settlement ␤ / ⌬GS becomes high. Thus there is an amplification effect
determined at Step 1 of the “Second Estimation Phase.” as cracks develop with increasing change in ground slope:
2. Determine the relative stiffness between soil and structure, doubling the change in ground slope will more than double ␤
considering the opening of a building (Fig. 10). and may even lead to development of bending cracks in the
3. Find the normalized angular distortion from the relationship upper floors.
between or among the normalized angular distortion, the 3. An updated damage criterion summarized here and in Cord-
relative stiffness, and the ratio between change in ground ing et al. (2001) and Son (2003) is a generalized state of
slope and structure cracking strain. Then determine an angu- strain damage criterion that is not dependent on L / H, E / G,
lar distortion (Fig. 11). and the position of a neutral axis. The generalized damage
4. From the lateral strains obtained at Step 2 of the “Second criterion is based on the state of strain at a point, or the
Estimation Phase,” determine a lateral strain considering fac- average strain across a building unit between two columns,
tors of influence such as the effects of a grade beam (Fig. two cross walls, two different building geometries or build-
12); an interface between soil and building, shear cracks ing stiffnesess, or two different ground displacement gradi-
which extend to upper floor levels; pre-existing cracks at the ents. The generalized state of strain damage criterion was
top of a building; weak planes; and restraint provided by combined with field observations, so that it is a semiempir-
joists, roof, or connections to cross walls. ical criterion. Damage estimation using the generalized state
5. Find a damage level from the generalized state of strain dam- of strain damage criterion was in good agreement with field
age criterion using the angular distortion and the lateral strain observations. Based on the data from physical model tests
(Fig. 8). and numerical parametric studies, the criterion was also over-
6. If the damage level at Step 3 of the “Final Estimation Phase” all in good agreement with the damage classification using
is acceptable, finish at this step. If this is unacceptable, de- maximum crack width.
termine appropriate protection measures or means of mitigat- 4. Some cases with severe damage levels based on maximum
ing estimated effects as described in the “Production Phase.” crack widths were observed from the numerical parametric
studies in the upper range of strains for the “moderate to
Protection Phase severe damage” zone in the generalized state of strain dam-
age criterion. The strain for the “moderate to severe damage”
Find a protection scheme to prevent damage or limit damage to range may induce severe distortion of door or window
acceptable levels: frames or reduction of joist bearings, which may cause bear-
• At source of ground movements (tunnel or excavation); ing failure instead of cracking in some structures. These ob-
• Ground modification/replacement; and servations confirm and justify that severe damage, although
• Building reinforcements. not expected over the full range of the “moderate to severe
damage” category, may occur in the upper range of the cat-
egory.
Conclusions 5. The ratio 关␤ / 共⌬ / L兲兴 between the angular distortion 共␤兲 and
the deflection ratio 共⌬ / L兲 was investigated. The ratios ranged
1. Bearing walls oriented in a direction perpendicular to an ex- between 2 and 4. For elastic (including frame structures) and
cavation wall tend to become distorted with shear strain and for minor cracking conditions, the ratios were in the range of
lateral strain at the foundation level. When ground movement 2, but for severe cracking in a structure, the ratios increased
initially impinges on the front of a building, the building is up to 4. The low ratio of 2 indicates that the distortion in a
primarily subjected to shear distortion and lateral strain at the structure is less than change in ground slope 共⌬GS兲 that
base. In addition, façade downdrag induced by ground settle- would occur in the absence of the structure, while the high

176 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005

J. Geotech. Geoenviron. Eng. 2005.131:162-177.

 ratio of 4 indicates that the distortion in a structure is close to Geotech. Eng., 122(11), 886–896.
the change in ground slope. Boone, S. J., Westland, J., and Nusink, R. (1999). “Comparative evalua-
6. A relationship was developed to consider the effect of soil– tion of building response to an adjacent braced excavation.” Can.
structure interaction in reducing the ratio of angular distor- Geotech. J., 36, 210–223.
tion 共␤兲 in a building to change in ground slope 共⌬GS兲 in a Boscardin, M. D., and Cording, E. J. (1989). “Building response to
excavation-induced settlement.” J. Geotech. Eng., 115(1), 1–21.
free-field ground settlement profile. The relationship is a
Burland, J. B. (1995). “Assessment of risk of damage to buildings due to
function of the relative soil/structure shear stiffness
tunneling and excavation.” Proc., 1st Int. Conf. on Earthquake Geo-
共EsoilL2 / GbuildHb兲, change in ground slope with respect to
technical Engineering, IS, Tokyo.
structure cracking strain 共⌬GS/ ␧t兲, and building downdrag
Burland, J. B., Broms, B. B., and de Mello, V. F. B. (1977). “Behavior of
force. The increase in angular distortion with increase in the
foundations and structures.” Proc., 9th Int. Conf. on Soil Mechanics
ratio of ground/structure shear stiffness (decrease in building
and Foundation Engineering, II, State of the Art Report, Tokyo, 495–
Downloaded from ascelibrary.org by NATIONAL UNIV OF SINGAPORE on 07/04/13. Copyright ASCE. For personal use only; all rights reserved.

shear stiffness) was examined for both elastic and cracked 546.
building walls. Cracking significantly reduced effective wall Burland, J. B., and Wroth, C. P. (1974). “Settlement behavior of buildings
stiffness making the wall more conformable to the ground and associated damage.” Proc., Conf. on Settlement of Structures,
settlement profile. This increased angular distortion, causing Pentech Press, London, 611–654.
it to approach the distortion (change in ground slope) that Cording, E. J., Long, J. H., Son, M., and Laefer, D. F. (2001). “Modelling
would occur in the absence of the structure. Lower values of and analysis of excavation-induced building distortion and damage
␤ / ⌬GS were obtained for the various combinations of using a strain-based damage criterion.” Proc., London Conf. for Re-
higher-story buildings, higher tensile strength, fewer window sponses of Buildings to Excavation-Induced Ground Movements in
openings, smaller change in ground slope, less downdrag London, London.
force, and stiffer structures with respect to the soil. Compar- Itasca Consulting Group, Inc. (2000). UDEC 3.1 theory and manual,
ing the field observations and physical model tests with the Minn.
results based on numerical parametric studies, good agree- Laefer, D. F. (2001). “Prediction and assessment of ground movement
ment was found for the relationship with some degree of and building damage induced by adjacent excavation.” PhD thesis,
conservatism. Univ. of Illinois at Urbana-Champaign, Urbana, Ill.
7. From the results of the investigation, a phased procedure was National Coal Board (NCB). (1975). Subsidence engineers handbook,
developed for estimating building damage due to excavation- National Coal Board Production Dept., London.
induced ground movements for practical use in project plan- O’Rourke, T. D., Cording, E. J., and Boscardin, M. (1976). “The ground
ning and design stages. movements related to braced excavation and their influence on adja-
cent buildings.” Univ. of Illinois Rep. No. DOT-TST-76T-22, Prepared
for U.S. Dept. of Transportation, Washington, D. C.
Polshin, D. E., and Tokar, R. A. (1957). “Maximum allowable non-
Acknowledgments
uniform settlement of structures.” Proc., 4th Int. Conf. on Soil Me-
chanics and Foundation Engineering, Vol. 1, Butterworth, England,
This study was funded by the National Science Foundation (NSF)
402–405.
and Schnabel Foundation Co. Their support is gratefully acknowl- Rankin, W. J. (1988). “Ground movements resulting from urban tunnel-
edged. ling: Predictions and effects.” Engineering geology of underground
movements, F. G. Bell, M. G. Colshaw, J. C. Cripps, and M. A. Lov-
ell, eds., Geological Society, London, 79–92.
References Skempton, A. W., and MacDonald, D. H. (1956). “The allowable settle-
Bjerrum, L. (1963). “Discussion on: Proc., European Conf. on Soil Me- ment of buildings.” Proc. Inst. Civ. Eng., Struct. Build., 5, 727 – 784.
chanics and Foundation Engineering, Vol. III.” Publication No 98, Son, M. (2003). “The response of buildings to excavation-induced ground
Norwegian Geotechnical Institute, Oslo, Norway, 1–3. movements.” PhD thesis, Univ. of Illinois at Urbana-Champaign, Ur-
Boone, S. J. (1996). “Ground-movement-related building damage.” J. bana, Ill.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / FEBRUARY 2005 / 177

J. Geotech. Geoenviron. Eng. 2005.131:162-177.