ENGM272: Deep Foundation & Earth

Retaining Structures
Coursework - Critical Analysis of Foundation Failures

Michael Jury SID: 6292312 ENGM272

Table of Contents
Case Study 1: Uprooting of Pile Foundations ......................................................................................... 1
Introduction ......................................................................................................................................... 1
Failure Analysis .................................................................................................................................. 2
Conclusions ......................................................................................................................................... 4
Case Study 2: Transcona Grain Elevator ................................................................................................ 5
Introduction ......................................................................................................................................... 5
Failure Analysis .................................................................................................................................. 6
Conclusions ......................................................................................................................................... 8
Case Study 3: Kissing Tower Silo Failure .............................................................................................. 8
Introduction ......................................................................................................................................... 8
Failure Analysis .................................................................................................................................. 9
Conclusions ....................................................................................................................................... 10
Case Study 4: Shanghai High-Rise Collapse ........................................................................................ 10
Introduction ....................................................................................................................................... 10
Failure Analysis ................................................................................................................................ 11
Conclusions ....................................................................................................................................... 13
Case Study 5: Sleipner A Platform Collapse ........................................................................................ 13
Introduction ....................................................................................................................................... 13
Failure Analysis ................................................................................................................................ 14
Conclusions ....................................................................................................................................... 15
Case Study 6: Wind Tower Turbine Failures........................................................................................ 15
Introduction ....................................................................................................................................... 15
Failure Incident ................................................................................................................................. 16
Failure Analysis ................................................................................................................................ 17
Conclusions & Recommendations .................................................................................................... 20
References ............................................................................................................................................. 21
Case Study 1: Uprooting of Pile Foundations

Introduction
In March 2011 the east coast of Japan was subjected to a severe earthquake with a magnitude of 9.0 on
the Richter scale. The Great Tohoku earthquake, as it has become known caused a tsunami which
inflicted substantial damage to structures and infrastructure to towns and cities along the coast. One
such town was Onagawa, which experienced some of the most extreme structural damage experienced
as a result of the earthquake and subsequent tsunami. The structural damage varied considerably within
the town, but interestingly 6 buildings were overturned. The construction of the 6 buildings varied for
example 1 building was of steel frame construction supported on piled foundations, whilst the remaining
overturned buildings were reinforced concrete frames on either shallow foundations or piled
foundations.
This report will focus on one particular structure in order to determine the likely failure
mechanism(s) induced by either the initial quake or the subsequent tsunami or a combination of both
factors. The structure in question is illustrated in Figure 1.1 below and consisted of a 4-storey steel
frame with autoclaved lightweight concrete block walls supported on 300mm diameter hollow core
reinforced concrete piles. The form of construction of this particular building was reported by
Latcharote et al., (2014).
The building is reported to have been built between 1980-1985 (Latcharote et al., 2017) as a
residential development, but prior to the tsunami the building was used for offices. When the tsunami
hit the building was moved some 10-16m from its original location and upon inspected it was noted that
the majority of the piles had been broken at the connection between the pile and the pile cap with the
exception of one pile that was pulled from the ground. The possible method by which failure occurred
is discussed in the following section.

(a) (b)

(c) (d)
Figure 1.1: (a) Building elevations. (b) Section through single pile. (c) Plan showing pile cap and pile arrangements. (d)
Photo of structure in overturned state.
Failure Analysis
There are a number of potential factors that may have contributed to the overturning of the building in
question. Sugimura et al., (2012) reported that such factors possibly included hydrodynamic force,
buoyancy force, debris impact and liquefaction of soil (as cited in Ishida et al., (2015). The numerical
analysis for the overturning moments and restorative moments have been previously undertaken in
available literature, but for completeness the calculations will be provided below (with minor
amendments where necessary) in order to formulate relevant conclusions upon the likely failure
mechanisms.

Figure 1.2: Free body diagram indicating potential mechanisms of overturning. Diagram taken from
Latcharote et al., (2014).

Hydrodynamic Force (Fd)

ℎ = 4521.6kN

Where Cd = Drag coefficient (taken as 2.0)

ρs = Density of salt water with sediment (taken as 1200kg/m3)
b = Building width (taken as 16m)
h = Inundation depth or building height (taken as building height 12m)
u = Flow velocity (taken as 4.43m/s as simulated by Adriano et al., (2014

Buoyancy Force (Fb)

ℎ = 5695.8kN
Where Cb = Residual air space ratio 0.0 to 1.0 (0.3 as this value could be practically achieved)
ρs = Density of salt water with sediment (taken as 1200kg/m3)
g = gravitational acceleration (9.81m/s)
b = Building width (taken as 16m)
 h = Inundation depth or building height (taken as building height 12m)
d = Building depth (taken as 8.4m)

Building Self Weight

As adopted by Latcharote et al., (2014) the self weight for steel framed building was taken as 8.4kN/m2
per unit area of floor.
= 8.4 × 16 × 8.4 × 4 = 4515.8

Pile Skin Friction

Table 1.1 shows the skin friction capacity using borehole sample data from areas adjacent the
overturning building. This data also considers to possibility of soil liquefaction. Soil liquefaction occurs
when the shear strength becomes effectively zero brought about by either monotonic or cyclic loading,
such as during earthquake or wave loading. The values of CRR (cyclic resistance ratio) and CSR (cyclic
stress ratio) were used to calculate factor of safety to determine the pile skin friction capacity during
potential occurrence of soil liquefaction. When the factor of safety is less than 1.0 the pile resistance
force is considered to be zero for that particular soil type.

Table 1.1: Data taken from Latcharote et al., (2014).

Total Q Single = 312.1kN (no soil liquefaction)
Total Q Single = 136.7kN (soil liquefaction)
Q Group = ηNQsingle
Q Group = 0.75 x 20 x 136.7 = 2050.5kN
The above pile group skin friction resistance conservatively assumes that soil liquefaction has occurred
and that the pile depth extends to a depth of 14.8m as consistent with the soil profile of Table 1.1. The
pile length is unlikely to extend to this depth. Therefore a depth of 8m considering soil liquefaction will
be used to provide the values of Q Single & Q Group.
 Total Q Single = 91.5kN (soil liquefaction)
Q Group = ηNQsingle
Q Group = 0.75 x 20 x 91.5 = 1372.5kN
Pile Self Weight = [(π x 0.32)- (π x 0.182)] x 8 x 20 x 2400 x 9.81 x 10-3 = 681.7kN

Overturning Analysis
Now that each force acting upon the building has been determined in turn a potential failure mechanism
can be evaluated as follows:
Md = 4521.6 x 6 = 27129.6kNm
Mb = 5695.8 x 4.2 = 23922.4kNm
Mw = 4515.8 x 4.2 = 18966.4kNm
Mp = (1372.5 + 681.7) x 4.2 = 8627.6kNm
Md + Mb = 51052kNm, Mw + Mp = 27594 Md + Mb > Mw + Mp

Conclusions
It can be seen from the above calculations that the overturning moment is greater that the restorative
moment and as such the building was not capable of resisting the applied hydrostatic and buoyancy
forces when combined. We obviously know this given the aftermath of the tsunami but it should be
noted that there are other factors that may have had an indirect impaction upon the failure. For example
the building in question was thought to be constructed between 1980-1985 and it was not until 1985
that Japanese design codes called for foundations to be design for seismic actions (Kaneko, et al., 2018).
This may have led to damage or weakening of the piles during the initial quake and coupled with and
soil liquefaction may partially explain why the majority the piles were sheared off at the pile cap
junction and only one pile was uprooted. This may have occurred at the point of the quake or due to the
subsequent tsunami.
It could be hypothesised that the failure is likely due to a number of factors that are hard to predict and
design for, which is why there is a lack of design guidance for such events as tsunami impact. As such
the failure is ultimately due to unforeseen circumstances and can in no way be attributed to negligence
on the engineers’ part given there are no design codes available to adhere to. However based on the
information gathered following the event there are a number of recommendations to minimise the risk
of such failures occurring in the future. Buildings constructed in potential tsunami impact zones should
consider their aspect ratios and orientation to potential wave directions, increased building dead loads,
water/air tightness, increased pile diameters and lengths, additional shear and bending reinforcement,
adequate reinforcement tying at pile cap junction and ground improvement techniques in order to reduce
soil liquefaction occurring.
Case Study 2: Transcona Grain Elevator

Introduction
On 18th October 1913 a grain elevator situated in North Transcona near Winnipeg, Canada was noted
to exhibit excessive settlement once it had been loaded. The grain elevator was constructed by the
Canadian Pacific Railway Company. The grain elevator comprised a 55m high reinforced concrete
frame work house founded on a 21.5m x 29.3m shallow raft slab and this structure was situated
approximately 3m from the bin house which was formed of 5 rows of reinforced concrete bin stores
each 28m high and 4.4m in diameter. The bin cylinders here founded on 600mm thick x 23.5 x 59.5m
reinforced raft foundation (Puzrin et al., 2010).
Following the loading of the bin stores, which were reportedly loaded in a uniform fashion and only to
87.5% of their design capacity (Puzrin et al., 2010), the bin house was observed to settle and within an
hour a settlement of 300mm was witnessed. The settlement was accompanied by tilting in the westerly
direction which progressed over a 24-hour period and approached a maximum 27o inclination from the
vertical.

Figure 2.1: Collapse of the Transcona Grain Elevator. Photo taken from Puzrin et al., 2010.

The grain elevator was built on partly farmed prairie land that was relatively flat. The ground profile
known at construction stage was considered typical for the surrounding areas where similar structures
had been built. The ground make-up was typically blue clay locally known as ‘blue gumbo’ and
although no borehole logs were taken once the excavation reached the design depth of 3.7m loaded
plate tests were undertaken on a level clay surface and indicated that a bearing pressure of 400kN/m2
was achievable. As the maximum design pressure was to be circa 300kN/m2 this, along with the
assumption that the clay properties and depths at this site were similar to depths at neighbouring sites
where similar construction had taken place, satisfied the engineers that the founding soils where capable
of resisting the expected loads without undue stress and settlement (Puzrin et al., 2010). Immediately
following the failure a number of boreholes were formed and confirmed that the grain elevator
formation was in a relatively uniform clay deposit and that properties were consistent with design
assumptions prior to construction. It was almost 40 years later until a failure mechanism was discovered.
In 1951 Peck and Bryant undertook two additional boreholes some distance from the failure site so that
any soil displacement caused by the failure would not affect the obtained samples. The undisturbed
samples were subjected to triaxial shear tests with zero confining stress (Puzrin et al., 2010) and the
results revealed some significant information as illustrated in figure 2.2 below. The soil profile may
have looked homogenous with depth as colour, grain distribution and mineral content did not differ
notably with depth but the unconfined compression strength tests revealed distinct layers of clay. The
top 7.5m layer reported as a stiff clay with compressive strength of 108 kN/m2 and undrained shear
strength of 54kN/m2 whilst the lower layer (softer clay) possessed a compressive strength of 62kN/m2
and undrained shear strength of 31kN/m2. From these results it would suggest that the failure
mechanism was brought about by insufficient bearing capacity of the underlying strata. The theory
behind bearing failures will be briefly addressed in the next section in order to illustrate the design
process at the time and subsequent checks based upon information revealed following the failure to
highlight suitable methods of analysis to determine failure pressures.

Figure 2.2: (a) Soil profile (b) unconfined compression strength test values. Diagram taken from Puzrin
et al., 2010

Failure Analysis
Geometry Soil Profile Loads
Rectangular footing width Thickness of upper clay layer below footing Surface surcharge
= 23.5m = 6.0m = 0kN/m2
Rectangular footing Length = Undrained shear strength of upper clay Failure contact pressure from plate tests
59.5m = 54kN/m2 = 400kN/m2
Depth of foundation Undrained shear strength of lower clay True contact pressure at failure
= 3.7m = 31kN/m2 = 293kN/m2
Total unit weight of both clay layers
= 18.7kN/m3
Table 2.1 – Model parameter used in the following failure analysis.

Undrained bearing capacity formula:

Where: q is the surcharge load.
Nc is the bearing capacity factor
sc is the shape correction factor
dc is the depth correction factor

Based upon the original design which assumed the soil profile was homogenious and exhitbited the
strength properties of the upper clay layer the bearing capacity would be as expressed below.

If the presence and strength of the lower clay layer was known at the time and the weaker strength
values adopted the bearing capacity would be as indicated below.

Considering a 2 layer strata two forms of upper bound limit analysis could be employed. The first being
the Prandtl kinematic failure mechanism using a weighted average of the undrained shear strength and
the second being the Scoop failure mechanism analysis.

Figure 2.3: Two layer strata kinematic failure mechanism Puzrin et al., 2010.
When employing the Prandtl kinematic failure mechanism the weighted average of the undrained shear
strength is calculated as follows:

The bearing capacity then becomes:

Using the Scoop failure mechanism analysis with the

values of m & n above the value of Nc = 3.7 as obtained
from the adjacent chart the bearing capacity is
calculated as follows:
Conclusions
The Transcona grain elevator failure can be attributed to a bearing capacity failure. The reason for this
occurrence is due to a number of factors, such as an insufficient site investigation at the time of
construction. It should be noted that at the time of construction, circa 1911-1913 wash-boring were
employed to investigate the soil profile and this technique alone would not be able to determine the
shear strength parameters of the soil. Although plate tests were carried out on the clay at formation level
the bearing data yielded from these tests were not transferable to the proposed raft foundation. The
reason for this is due to scale. The plate tests employed would be relatively small in comparison to the
proposed raft foundation and the depth of the stress bulb would only extend to 2x its breadth/diameter
and as such would likely not extend into the lower clay layer as tests were conducted at a depth of 3.7m
the plates would need do have been 1.9m wide just to reach the lower clay layer. The stress bulb of the
raft foundation due to its dimensions would have interacted with the lower softer clay layer and as such
this brought about the bearing failure. Engineers also relied upon anecdotal evidence from neighbouring
sites with regards to the strength and depth properties that were not applicable to this specific site.
Given the above information it could be considered that the failure occurred due to the time period of
the project and a lack of engineering knowledge at the time given that the soil sampling and testing
techniques of this era were limited and it took circa 40 years for a failure mechanism to the revealed,
which was be partly due to advances in engineering knowledge, sampling techniques and testing
procedures. The calculations demonstrated in the previous section to predict the bearing capacity of the
soil are useful tools, especially the Scoop failure mechanism analysis technique, but these techniques
where not available in 1913 and relied upon the information of the soil profile and strength properties
that were not known at the time of construction and as such would not have been able to prevent the
failure.

Case Study 3: Kissing Tower Silo Failure

Introduction
Tower silos are agricultural structures that are used to store crops and they are typically tall bullet shaped
structures. These structures were first construction in the 1930’s and were typically constructed from
timber and were much smaller than more modern versions which are typically constructed from
concrete and reach heights of circa 30m and have capacities of 2500 tons. As reported by Bozozuk
(1976) by the mid 1970’s the number of tower silos constructed across parts of Canada alone
represented an investment of over $30 million. With time the superstructure elements have improved
from a design and construction perspective, but the substructure design and construction techniques
have not progressed at the same rate. Bozozuk (1976) reported that it is common place that foundations
for the tower silos are often provided by the farmer who may often lack the engineering knowledge and
construction technology in order to provide a suitably design and constructed foundation and as such
this highlights the need for adequate soil investigations and foundation design to be undertaken by
professionally qualified persons.
The particular tower silo failure to be reviewed within this report is that of two adjacent silos constructed
in the Red River Valley in Canada. The tower silos were founded in Lake Agassiz clay (Dogangun,
2009) and were shown to tilt due to non-uniform settlement of the clay under the foundation (Figure
3.1). The foundations to the tower silos were concrete ring foundations and the construction of the silos
were reported to be simultaneous. The following section will address the reasons for this type of failure.
Figure 3.1: Two tower silos tilting due to non-uniform settlement.

Failure Analysis
Once the silos are loaded (assuming the loads within the silo are evenly distributed) this uniform
pressure is transferred to the founding soil and results in a stress bulb as illustrated in Figure 3.2(a)
below. The shape and extent of the pressure bulb is directly proportional to the diameter of the loaded
area (Bozozuk, 1976). It can be seen from Figure 3.2(a) that the foundations are loaded with the same
uniform load, but the larger foundation exhibits a larger and deeper stress bulb. The maximum pressure
will occur directly below the foundation and dissipates to circa 10% at a depth equal to 2x the footing
diameter (Bozozuk, 1976) and if the pressure within the bulb does not exceed the shear strength of the
soil the structure will remain stable.
If the silo is not loaded equally this will distort the pressure bulb as illustrated in Figure 3.2(b), this can
also occur to due wind loading especially to tall silos and can result in localised overloading to the
founding strata leading to non-uniform settlement causing the silo to tilt and potentially even overturn
if the pressure exceeds the ultimate bearing capacity. Similarly where silos are constructed in close
proximity to each other, as with the two tower silos in the Red River Valley under consideration here,
the pressure bulbs overlap and result in a combined pressure bulb that is larger and extends to greater
depths. The soil within the area of the bulb overlap will be experience greater stresses and the soil in
this area will settle to a greater extent resulting in the silos tilting in towards each other.

(a) (b)
Figure 3.2: (a) Pressure bulbs under differing sized foundations (b) non-uniform & overlapping pressure
bulbs.
Bozozuk (1976) reports that many foundations failures in clay soils are the result of loading the silos at
a rapid rate for the first time. The initial loading applies pressure on to the soil particles and the pore
water pressure with the voids between the clay particles. The pressures generated within the pore water
reduces the friction between the clay particles, this in turn reduces the shear strength of the soil. Given
the rate of loading the pore water does not have time to dissipate causing soil to consolidate where shear
strength is restored and failure may occur if the shear stresses exceeds the reduced shear strength of the
soil. A further complication to the above failure mechanism is related to the silage stored in the silos.
The silage contains large amounts of water and typically the silos are constructed without floor slabs
and as such the liquid can seep into the underlying soil. This can increase pore pressures further
exacerbating the shear strength reduction or prolonging the dissipation of pore pressures (Bozozuk,
1976). The foundations can also be undermined due to soil washout caused by silage water flowing
through the soil under hydraulic pressure.

Conclusions
Based on the above information the failure of the ‘Kissing Silos’ is likely caused by overlapping of the
pressure bulbs under the concrete ring foundations. This will have caused localised stress increases
where the pressure bulbs overlap resulting in non-uniform settlement in this location which causes the
tilting illustrated in Figure 3.1. The tilting could have been further exacerbated if the silos were unevenly
or rapidly loaded.
The mechanics of the failure have been addressed but the reason to why this was allowed to occur is
likely due lack of engineering knowledge, lack of design for the silo foundation and no soil
investigation, as it has been noted that often silo foundations are provided by unqualified individuals as
could be hypothesised here. It would be expected that if a qualified geotechnical or structural engineer
were responsible for the foundation design that the silos would not be located so close together, it is
recommended that silos are separated by a distance equal to at least the diameter of the required
foundation to avoid interaction of pressure bulbs Bozozuk (1976) or a combined raft or piled foundation
be adopted to resist the required forces.

Case Study 4: Shanghai High-Rise Collapse

Introduction
On 27th June 2009 a recently constructed building situated on Lianhuanan Road in Shanghai City
unexpectedly collapsed. One week prior to the collapse a 4.6m deep excavation had been carried out to
one side of the structure and the cut from this excavation was stored on the opposite side of the building
forming a 10m high fill slope (Wang et al., 2018). The building was a 13-storey reinforced concrete
frame which was supported off a capping beam and 118 lightly reinforced piles. The ground beam was
600mm wide x 700mm deep. The piles were pre-cast displacement piles that extended to a depth of
33m. Piles were pretensioned and formed in high strength concrete (grade C80). The piles were open
ended with an overall diameter of 400mm and a wall thickness of 80mm. Piles were designed as friction
piles with an estimated axial capacity of 2300kN per pile (Wang et al., 2018). Following the collapse
the piles were inspected and it was revealed that all the piles supporting the structure had been broken.
The piles on the excavated side (Figure 4.1) were reportedly broken into 3 to 4 pieces along their length,
whereas the central piles were broken into 3 pieces and the piles on the surcharged side of the building
were broken at the junction of the steel caisson and the pile which was circa 2m below ground level.
Figure 4.1: Schematic site arrangement prior to collapse (Wang et al., 2018).

The soil profile at the collapse site consisted of 9 layers of clay on top of a layer of fine sand. The clay
layers can be broadly grouped into 2 main layers. The top layer is reported as being a soft clay and
extends to a depth of 13.5m whilst the second layer is a stiff clay from 13.5m to 29.7m. The sandy layer
to the base of the lower stiff clay achieved a SPT value of 27 to 32. The soil profile is illustrated in
Figure 4.2.

Failure Analysis
Following the collapse of the structure an investigation was undertaken in order to determine to cause
and mechanism of the failure. The first forensic investigation employed three-dimensional finite-
element analysis, but there was a lack of comprehensive information relating to the undrained shear
strength of the soil at the construction site, therefore a typical value of cu for Shanghai clay was adopted
for use in the initial numerical analysis. The findings of that investigation were surprising as it was
suggested that the piles were unlikely to break which is obviously in stark contrast to the results of the
actual failure observed on site. These findings understandably led to further investigations into the
failure given that the initial investigation was at odds with the actual failure. One such study conducted
by Wang et al. (2018) undertook a forensic study using three-dimensional centrifuge and numerical
modelling (Figure 4.3).

Figure 4.3: (a) Scaled centrifuge model arrangement (b) 3D view of finite element mesh (c) Scale model
used in centrifuge, units = mm and (d) pile layout plan for centrifuge model. (Wang et al., 2018).

Based on the finding of the Wang et al. study it was determined that the likely failure mechanism was
initiated by the surcharge loading applied from the spoil heaps. The surcharge loading resulted in a
bearing capacity failure of the upper layer of soft clay which was exhibited by the upper layer settling
downwards and being displaced laterally forming a slip surface in the direction of the excavation and
as such the piles became laterally loaded meaning substantial bending moments occurred in the piles.
The highest bending moments were exhibited in the front piles on the excavation side and it was these
piles that ruptured first at the intersection of the two clay layers which were shortly followed by a second
breakage of these same piles at the slip surface. This sequence of soil and pile failures resulted in the
building initially tilting towards the surcharge side until the first pile ruptures occurred when the
building tilt was momentarily restored until tilting to the excavation side occurred due to further pile
breakages. This caused significant tension forces in the piles on the surcharge side of the building which
could not be resisted by the lightly reinforced piles resulting in separation of the pile and the capping
beam leading to complete collapse of the structure (Wang et al., 2018).

The failure mechanism described above (and illustrated in Figure 4.4) is based on the observations from
the centrifuge model, but these findings were fairly representative of the actual collapse mechanism
observed in the field. One area where the modelled and field failure mechanisms varied was the location
of the intermediate pile breaks to the front (excavation side) piles. In the field these breaks were at a
shallower depth compared to the centrifuge model. This could be attributed to a zone of 3-4m at the
base of the upper soft clay layer that was notably weaker than the clay above this depth. This may have
resulted in the slip plane forming at a shallower depth in the field as this weaker zone was not accounted
for in the centrifuge model (Wang et al., 2018).
Figure 4.4: Proposed collapse mechanism (a) development of slip surface arises from surcharge loading
(b) excavation side piles break at clay layer boundary (c) further pile breaks causing building tilt (d)
fully collapse.

Conclusions
Based upon the above discussions it is apparent that the cause of the collapse stems from the surcharge
loading induced by the spoil heaps. This is obviously an unsafe construction practice given that the
bearing capacity of the upper clay layer has been exceeded, but how this was allowed to occur is unclear
from the literature review. It would be expected that the contractor consulted the design or resident
engineer regarding the excavation works and subsequent stock piling of spoil material on the opposite
of the building. If this is the case then the blame falls on the Engineer as consideration has not been
given to the impact upon soil capacity or pile design. On the other hand the contractor may have either
undertaken these works without the knowledge of the design engineer or the extent of the spoil pile
exceed the original proposals resulting in the catastrophic failure witnessed. Whatever the reason for
this occurrence it highlights the need for closer supervision on site by suitability qualified Engineers
and competent construction professionals.

Case Study 5: Sleipner A Platform Collapse

Introduction
On 23rd August 1991 a catastrophic failure occurred when the Sleipner A platform collapsed in the
Gandsfjord near Stavanger, Norway (Barry 2013). The Sleipner A platform was a Gravity Base
Structure (GBS) and was intended to be used to extract oil and gas from the North Sea. The GBS took
the form of a concrete deep water structure or ‘Condeep’, which consists of a series of buoyancy cells
in order to float the structure. Figure 5.1 illustrates the typical arrangement of Sleipner platform. The
buoyancy cells are formed in reinforced concrete and are filled will water ballast in order to control the
depth of the lower structure within the water. A number of these buoyancy cells extend vertically
forming shafts which support the deck that is above sea level (Jakobsen, 1994). Typically the cells and
shafts are partially formed in a dry dock and then floated to transport into the required location once the
dock is flooded. Once in position the buoyancy cells are capped off and the shaft structures are vertically
extended as required via slip forming to receive the deck structure. Prior to deck mating occurring it is
required that the GBS is subjected to a control ballast test in order to trial the equipment under service
loads and check for any leaks. It was during a controlled ballast test on 23rd August 1991 when the GBS
started to take on water uncontrollably at a depth of 97.5m. It was reported that immediately prior to
taking on water a deep bang originating from one of the shafts was heard followed by vibrations and
the sound of rushing water (Jakobson, 1994). Shortly after a smaller second bang was heard and upon
investigation it was discovered that drill shaft D3 was the site of the failure. At this point emergency
ballasting was initiated and all 14 individuals were evacuated from the platform. Just 18 minutes from
the initial onset of failure the entire platform was submerged and eventually hit the seabed at a depth of
220m causing a seismic event registering 3.0 on the Richter scale (Barry, 2013)

Figure 5.1: (a) Sleipner A platform (b) Section through buoyancy cells.

Failure Analysis
Two investigations were undertaken simultaneously, one by the Contractor and one by the oil company
(Statoil). It was concluded by both parties that the two weak points in the structure were the tri-cell
walls and the supports at the cell wall joints and at the time of failure these walls were calculated to be
80-120% utilised. The weaknesses in these regions were attributed to a number of factors such as the
unfavourable geometric shape of finite elements in the analysis programme. This was further
exacerbated by errors in the processing of analysis results leading to circa 45% underestimation of shear
forces in the wall supports. Errors in reinforcement detailing were also noted which meant some
reinforcing bars were too short or were not provided (Jakobsen 1994). Figure 5.2 illustrates where
failure was likely to occur and this failure was manifested as crushing of concrete and shear failure in
the tri-cell wall above the transverse reinforcement.
Subsequent scaled model tests were undertaken by the investigating teams in order to further verify
their findings upon the anticipated failure mechanism(s) and although the models could be considered
an oversimplification of the events experienced in the field the modes of failure exhibited during the
model testing agreed with the hypothesised failure modes. It should be noted that in at least one of the
model test the T-headed bar noted in Figure 5.2 was increased in length by a mere 500mm and this
corresponded to a shear capacity increase of approximately 70% and it is suggested that this minor
change in reinforcement detailing alone would have likely avoided the disaster (Jakobsen, 1994).
Figure 5.2: probably failure locations in tri-cell joint and shaft wall.

Conclusions
It can be concluded that the cause of the failure was a result of negligence on part of the design
Engineer(s), given that errors occurred in the finite element analysis model, shear forces were
underestimated and reinforcement detailing was inadequate for the stresses applied to the structure. This
highlights the need for extreme care in design and detailing along with the need for competent and
experienced engineers that are able to effectively interpret analysis models and verify the output by
alternative means such as hand calculations.

Case Study 6: Wind Tower Turbine Failures

Introduction
Wind power is becoming increasingly more prevalent with the World Wind Energy Associated
reporting that the world wide capacity of wind power reached 600GW with 53.9GW being added in
2018. This equates to almost 6% of global electricity demand (wwindea.org 2019). 200GW of the
600GW stated originates from China. The development of wind harnessing infrastructure within China
has rapidly grown in the past two decades which is evident from the abundance of Wind Tower Turbines
(WTT) in use today. This rapid growth is linked to the Country’s increased energy demands associate
with its economic development. Chinas most densely populated areas are the coastal regions that
frequently experience power shortages (Goa et al., 2015) which is why much of Chinas WTT resource
is located in the North Western and Eastern coastal regions. These areas are subjected to extreme
weather conditions such as Typhoons which was resulted in extensive losses to coastal wind farms
during the last decade and it is reported that global warming has potential to impact the intensity and
frequency of future typhoons (Goa et al., 2015) and as such careful consideration should be given to
the structural design of WTT and the associated infrastructure to ensure structural robustness for the
design life of the elements.
Wind tower turbines typically take the form of a large circular concrete gravity base in order to
adequately anchor the turbine. The current trend is to provide larger more efficient turbines that are
capable of generating over 3MW of energy and typically turbines of this size exceed 100m from ground
level to the hub centre (Loubser & Jacobs, 2016). The turbines are subjected to extreme loading
conditions and for this reason the foundation base is circular in shape in order to best resist the variable
directional design loadings. The diameter of these bases can be between 15-20m and this is to account
for both soil bearing capacity and stability of the tower (Zhou & Kong, 2015). The concrete gravity
base is connected to the circular steel tower shaft by various methods such as cast-in high tensile bolts
that are post-stressed following their installation (Loubser & Jacobs, 2016) or via an embedded circular
steel tube which connects to the steel tower shaft via flange plates. Typical WTT bases can require
between 350m3 and 450m3 of concrete with reinforcement densities of between 80-100kg/m3 (Loubser
& Jacobs, 2016).

(a) (b)

(c)
Figure 6.1: (a) Typical wind tower turbine arrangement (b) Typical wind tower turbine gravity base (c)
embedded flange plate and foundation arrangement.

The wind tower turbines under consideration in this case study are located on the east coast of china
and form a wind power station of 66 WTTs that was built 2010. Each WTT has a hub height of 70m
and the circular steel tower shaft is 4.2m in diameter supporting a 1.5MW three bladed head. The tower
shaft is connected to the base via an embedded circular steel tube with top flange plates.

Failure Incident
On 8th April 2013 alarms were activated on three of the WTTs which were eventually shut down. The
turbines are fixed with an alarm system that constantly monitors the vibration frequency of the turbine
under normal working conditions and if the vibration frequency drops below a specified value (in this
case 0.400Hz) the alarms sound and shutdown ensues. Following an extensive inspection of the WTTs
there were three unusual occurrences noted:

• Large amounts of ground concrete powder were discovered next to the embedded
circular steel tube and on top of the reinforced concrete base. The concrete powder
originated from the minute gap between the embedded steel tube and the concrete of
the reinforced base and it is suggested that a grinding action was experienced due to
slippage between the concrete surface and the surface of the embedded tube. This
slippage reduced the stiffness of the tower turbine resulting in a reduction in
 vibration frequency below 0.400Hz. Further intrusive inspection works were carried
out to determine the reason for the slippage and the extent of any associated damage.
It was discovered that concrete over the lower flange plate had been crushed and
there were instances where the horizontal reinforcing bars passing through the
embedded tube had sheared (Zhou & Kong, 2015).

• In addition to the seriously damaged turbines mentioned above a further 19 turbines

recorded vibration frequencies of 0.375Hz and cracking occurred to the concrete
adjacent the embedded steel tube along with concrete powder in the same vicinity.

• A further 14 turbines reported a vibration frequency of circa 0.400Hz, visible shake

and the presence of concrete powder, although to a lesser extent to the turbines
mentioned above.

Failure Analysis
Following the completion of the investigation works discussed previously analysis of the failure was
undertaken to understand how the overall failure occurred and explain the reasons for the damage
witnessed during the inspections. A number of forms of analysis were carried out and were focused
upon analysis of loading during typical working conditions against that of a recent typhoon, bond
behaviour between the concrete and the embedded circular steel tube, numerical analysis of the
reinforced concrete foundation and finite element modelling analysis of the foundation and embedded
circular steel tube.

Wind Loading Conditions

Seven months prior to the failure incident occurring a powerful typhoon, called Bolaven swept across
the east coast of China where the power station is located. Table 6.1 below details the design wind
speeds adopted along with the resultant design loads and the recorded wind speeds during storm
Bolaven and resultant loading for comparison.
Design Wind Speeds (m/s) Design Loads
Vert Horiz M Tq
(kN) (kN) (kNm) (kNm)
Cut-in Speed 3.0 SLS -1962.1 256.3 16939 277.0
Rated Speed 14.0 ULS -1948.3 487.9 35618 594.6
Cut-out Speed 22.0
Survival Speed 37.5 (Max duration 10min)
Maximum Survival Speed 52.5 (Max duration 3s)
Storm Bolaven Wind Speeds (m/s) Storm Bolaven Loads
Vert Horiz M Tq
(kN) (kN) (kNm) (kNm)
Max Wind Speed @ epicentre 48.0 Cut-in Load -1948.3 396.4 27751 594.6
Wind speed at power station 38.9 (duration 10h) Cut-out Load -1948.3 568.2 39777 594.6
Idling Load -1948.3 483.4 33835 594.6
Blocking Load -1948.3 633.8 44369 594.6
Table 6.1: Comparison of design wind speeds and loading to wind speeds and loads experienced during
storm Bolaven (data from Zhou & Kong, 2015).
From the data shown in Table 6.1. it can be seen that during storm Bolaven the wind turbines were
exposed to loading conditions and wind speeds that exceed the design values. The wind speed
experienced during the storm was above the design survival speed which was considered over a
maximum 10 minute duration. The 38.9m/s wind speed recorded during the storm was sustained for
circa 10 hours (Zhou & Kong, 2015). The turbines were subjected to loading beyond their design
capacities for a substantially prolonged period.
Bond Behaviour
Between the face of the embedded circular steel tube and the concrete bond shear stresses exist and
the behaviour is illustrated in Figure 6.2 and shows that near the tension zone the bond stress gradually
increases along the steel as tensile forces increase until stress distribution occurs continually and bond
failure ensues (Zhou & Kong, 2015). The bond shear stresses between the tube wall and the concrete
are not typically considered in the design given the relatively small force magnitudes experienced
during serviceability loading and if bond stress is lost during severe wind loading the friction force
generated between the wall and the concrete can be ignored. When bond shear stress is lost the lower
flange plate is subjected to the total load transferred from the tower and the concrete either under or
over the lower flange plate is a repetitive state of compression given the force direction applied.

Figure 6.2: Stress diagrams upon tube wall and flange plates dependent upon force direction (Zhou &
Kong, 2015).

Foundation Analysis
Under normal conditions it is considered that the moment, axial and shear forces generated by the wind
and gravity loads will be produced on the top flange plate of the embedded circular steel tube and
provided there is no slip between this tube and the surrounding concrete these forces will be resisted
by the foundation alone. If slip does occur, as would be suggested by the findings of the subsequent
inspections, then forces will be transferred to the lower flange plate of the circular steel tube. This will
result in the concrete under and over the lower flange plate being in an upward or downwardly
compressive state alternately through a 360 degree range. The stresses in the circular tube wall and
flange (Table 6.2) along with the load carrying capacity of the concrete in local compression were
calculated as follows:

σα = ±

Where NZ = Axial force from wind load

Mr = Moment from wind load
A = Cross-sectional area
W = Sectional flexural modulus

F1 = βcβ1fcA1 = 46.4MPa

Where βc = 1.0 if fcuk < 50Mpa if fcuk = 80MPa βc = 0.8

β1 = strength enhanced coefficient due to local compression of concrete = (Ab/A1)0.5 = 1.732
Ab = calculated area under the compression state = 3A1
A1 = local compression area
fc = prism axial compressive strength of concrete
 Normal State Ultimate Work State
Up Down Up Down
Circular Tube Normal Stress (MPa) 82.31 92.83 178.73 189.30
Flange Plate Normal Stress (MPa) 12.05 13.25 25.40 26.64
Circular Tube Shear Stress (MPa) 1.43 2.64
Circular Tube Torque Stress (MPa) 0.36 0.77

Table 6.2.

Finite Element Model Analysis

Finite element modelling was employed to investigate the various stress states experienced by the
foundation when subjected to the extreme loading conditions during storm Bolaven. The model
allowed for the varying behaviour of concrete when in a state of tension or compression and contact
elements were utilised to replicate the behaviour at the junction between the tube wall and the concrete
and the flange plate and the concrete. It was assumed that the interaction between the foundation and
the supporting strata was rigid i.e. soil deformation did not occur.
The results of the FEM analysis generally coincided with the analytical solution and output from the
FEM (Figure 6.3) supports the field observations noted. It can be see that from figure 6.3(a) that a slip
plane occurs between the concrete and the steel tube and there is an inclined stress field evident, This
would suggest that when this stress profile occurs the concrete adjacent the flange plate will obliquely
fracture and the concrete over the flange plate with be crushed. The slip plane is also evident in figure
6.3(b) but this image indicates are larger downward compression region to one side and a smaller
upward compression region.

Figure 6.3:(a) von Mises stress ribbon under blocking load (b) vertical stress ribbon under blocking
load.(Zhou & Kong, 2015)

Failure Mechanism.
It was reported that when the typhoon hit the power station the turbines were in the blocking load state
(worst case loading) and under repeated and prolonged extreme wind loads the bond between the
concrete and the circular steel tower tubes was lost resulting in the transfer of wind loads directly to
the lower flange plate. Due to variation in wind load direction the concrete surrounding the lower
flange plate was in a alternating state of upward or downward compression. When a downward
compressive state the concrete under the flange plate had sufficient capacity to resist the applied forces.
The concrete over the top of the flange plate was put in to a state of punching and shearing and an
oblique fracture in the concrete occurred. Overtime with repeated wind loading the oblique crack grew
in conjunction with the increase bond slippage and the cracked concrete was crushed which was
eventually pushed out of the crack zone to pile upon on the foundation surface. It is suggested that the
bond failure and the oblique crack formation occurred at the time of the typhoon, but the concrete over
the flange plate was not fully crushed at the time. The crushing of the concrete progressed gradually
under normal loading conditions following the typhoon as the damage sustained during the typhoon
was not sufficient to stop the turbines working under normal condition (Zhou & Kong, 2015). As the
breakdown of concrete continued through the seven month period between the typhoon and the point
the extent of the failure was realised this resulted in ever increasing levels of vibration occurring, as
crushing and slippage increased until the threshold level was reached and the warning alarms sounded.
This is why upon inspection many if the turbines were in a state of failure but the alarms had not been
triggered.

Conclusions & Recommendations

It can be concluded from the above case study that the failure was a result of extreme loading
experienced during storm Bolaven. This failure was exhibited in two distinct regions at the foundation
and circular steel tube junction, namely loss of bond between these two elements and punching shear
failure of the concrete over the lower flange plates. It should be noted that the foundations have not
behaved poorly given that they have performed beyond their design expectations as the loading
experienced was above what was expected in the design stage.
It is therefore recommended that the magnitude of wind loading and the duration of application are
reviewed in future designs to best predict the nature of extreme weather conditions such as typhoons.
This however is not an easy task given how variable the intensity and frequency of these weather
phenomenon are not to mention how global warming has been suggested to alter these, but from all the
documented storm data in the region studies should be conducted to determine peak wind loads and
durations and to identify any trends in these values over the past years. This would allow designers to
better estimate the load magnitudes that the WTTs are likely to be experienced.
Consideration should be given to how the bond strength between the foundation concrete can be
improved given its importance in ensuring the integrity of the surrounding elements. Further study is
likely needed to fully understand the bond behaviour under various loading conditions but efforts should
be made in future designs to increase the surface area and roughness of the embedded circular tube or
provide welded studs to the embedded tube wall in efforts to increase the shear resistance between the
tube wall and concrete. Adequate waterproofing should also be provided to ensure water ingress does
not occur at micro gaps between the tube wall and concrete foundation which would increase the
likelihood of the circular steel tube slipping.
Finally provisions should be made to increase the punching shear capacity of the concrete over the
lower flange plate. This could be done in a number of ways such as increased concrete thickness,
increasing the concrete grade, increased reinforcement or providing specific shear reinforcing bars such
as loose bar links, studded bars or ring shaped steel strips.
References

• BARRY, J. (2013). The loss of the Sleipner A platform, Journal of Undergraduate Engineering
Research and Scholarship, March (2013).

• BOZOZUK, M. (1976). Tower silo foundations, Canadian Building Digest, CDB 177, May
(1976).

• DOGANGUN. A., KARACA. Z., DURMUS. A. and SEZEN. H. (2009). Cause of damage and
failures in silo structures, Journal of Performance of Constructed Facilities, 23(2009) pp, 65-
71.

• GOA. M., NING. J. and WU. X. (2015). Normal and extreme wind conditions for power at
coastal locations in China. PlosONE, [online] 10(8). Available at:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4551742/ [accessed 03.05.19].

• ISHIDA. M., TOKIMATSU. K. and INOUE. S. (2015). Overturning of buildings induced by

tsunami run-up at Onagawa in the 2011 Great Tohuku earthquake, 6th International Conference
on Earthquake Geotechnical Engineering, November (2015)

• JAKOBSEN. B. (1994). The Sleipner accident and its causes, Engineering Failure Analysis,
1(03) pp 193-199.

• KANEKO. O., KAWAMATA. S., NAKAI. S., SEKIGUCHI. T and MUKAI. T. (2018).
Analytical study of the main causes of damage to pile foundations during the 2011 off the
pacific coast of Tohoku earthquake, Japan Architectural Review, 1(2) pp 235-244.

• LATCHAROTE. P., SUPPASRI. A., YAMASHITA. A., ADRIANO, B., KOSHIMURA S.,
KAI. Y. and IMAMURA. F. (2014). Mechanism and stability analysis of overturned buildings
by the 2011 Great east Japan earthquake and tsunami in Onagawa town, The 14th Japan
Earthquake Engineering Symposium (2014).

• LATCHAROTE. P., SUPPASRI. A., YAMASHITA. A., ADRIANO, B., KOSHIMURA S.,
KAI. Y. and IMAMURA. F. (2017). Possible failure mechanism of buildings overturned during
the 2011 Great East Japan tsunami in the town of Onagawa, Frontiers in Built Environment,
3(March) pp 1-18.

• LOUBSER. P.B. and JACOBS. A.R. (2016). Optimised design of wind turbine gravity
foundations. In: Insights and Innovations in Structural Engineering, Mechanics and
Computation, London: Taylor & Francis Group.

• PUZRIN. A.M., ALONSO. E.E. and PINYOL. N,M. (2010). Geomechanics of Failures.
London: Springer

• SUGIMURA. Y. and MITSUJI. K. (2012). The survey of buildings overturned and carried out
by tsunami. Foundation Engineering and Equipment, 40(12) pp 71-75.

• WANG. W.D., NG. C.W.W., HONG. Y.I., HU. Y. and LI. Q. (2018). Forensic study on the
collapse of a high rise building in Shanghai: 3D centrifuge and numerical modelling.
Geotechnique 16 pp 1-16,

• WWINDEA.ORG (2019). Wind power capacity worldwide reaches 600GW, 53.9GW added in
2018 [online] available at: https://wwindea.org/blog/2019/02/25/wind-power-capacity-
worldwide-reaches-600-gw-539-gw-added-in-2018/ [accessed 03.05.19]

• ZHOU. X. and KONG. H. (2015). Failure case study of reinforced concrete foundations od
wind turbines, Forensic Engineering, 168(FE4) pp 158-166.