Ground anchors corrosion - the beginning of the end
MATEC Web of Conferences
Ground anchors corrosion - the beginning of the end
0 Department of Civil and Construction Engineering, National Taiwan University of Science and Technology , Taipei , Taiwan
Ground anchor corrosion is a common problem for anchored slopes in Taiwan. It is partly due to the humid climate condition and abundant groundwater in the slope and partly due to poor corrosion protection of anchor design and construction. In 2010, an anchored slope at Taiwan National Freeway No. 3 failed suddenly after 13 years of service. It buried 3 cars and killed 4 people. It caught the public's attention and initiated the island-wide program on over hauling the anchors slopes in Taiwan. Since this event, the Ministry of Transportation and Communication (MOTC) of Taiwan government had launched an extensive inspection and maintenance program for the existing anchored slopes along the freeways, highways, and railways. Totally, more than 100,000 ground anchors had been inspected. This paper will evaluate the findings from this inspection program. It includes (1) the status quo of the anchors regarding the corrosion condition and the residual load that remained on the existing anchors; (2) remedial measures taken to sustain the serviceability of existing corroding anchors; (3) measures taken to enhance the long-term durability of new anchors without changing the strand material and the practice of anchor construction commonly used by the local contractors.
Ground anchors are commonly used in Taiwan and elsewhere in the world to tie back cut
slopes or to enhance the stability of natural slopes. So far, most of the anchors have served
their purpose well to stabilize slopes. However, the long-term performance of ground anchors
has been constantly questioned by some engineers in terms of the durability of anchor
components and the loss of anchored load during the the life cycle of the anchor. In fact, the
use of ground anchors to tie back slopes is not without failures (big or small) in Taiwan .
Some local governments even tried to restrict the use of ground anchors for slope stabilization
due to its uncertainty of the long-term performance. But the restriction attempt was finally
abounded because the ground anchor was such a handy tool for geotechnical engineers and
many more ground anchors were actually installed to stabilize the cut slopes since the
restriction was issued.
In 2010, an anchored slope failed suddenly and catastrophically after 13 years of service
at Taiwan National Freeway No. 3 (Fig. 1) [
]. This failure case had fundamentally changed
the ways of design and maintenance of anchored slopes in Taiwan. After this event, the
Ministry of Transportation and Communication (MOTC) of Taiwan government had
launched an extensive island-wide inspection and maintenance program on the existing
anchored slopes along the freeways, highways, and railways [
]. This paper will evaluate
the findings from this program. The problems of anchors under humid climate and high
groundwater conditions of Taiwan will be addressed. Finally, the remedial measures taken to
sustain the serviceability of existing anchors and to secure the long-term durability of the
new anchored slopes are proposed.
2 Inspection of existing anchors
The following steps had been taken to inspect the existing anchored slopes and evaluate the
residual stability of anchored slopes along freeways, highways, and railways:
(1) Visual inspection and hammer tapping on the concrete protection cap of anchors (Fig.
2): The integrity of concrete cap can be easily detected by hammer tapping. Special
attention was paid to find the cracks on the concrete cap and the sign of groundwater
leaking out from under the concrete cap. If there is/was constant water seeping out,
calcium carbonate (white stain) will deposit under the concrete cap and can be easily
Stain of CaCO
(2) Remove the concrete cap and inspect the steel strands and the wedges on the anchor
head (Fig. 3). If the integrity of concrete cap is good, normally the appearance of steel
strands and wedges also look good. Otherwise, a clear sign of corrosion can be observed
on the strands and wedges.
Use endoscope to inspect the condition of steel strands beneath the anchor head (Fig.
4): If there was void under the anchor head, the strands were inspected using an
endoscope which provided a close-up look on the corrosion condition of steel strands.
Generally, the corrosion condition of steel strands beneath the anchor head may not
necessarily correspond to the external appearance of anchor head components observed
in Step 1.
(4) Carry out lift off test to determine the residual anchor load: A lift off test apparatus as
shown in Fig. 5 is needed in this step. Extra caution must be exercised to avoid breaking
the rusty steel strands during lift off test if anchors were suffering serious strands
corrosion. It is quite normal to have some change (say ± 20%) on anchor load over
the life cycle of anchors. But if the residual load goes beyond 120% of the design anchor
load, it can be an indication of slope movement. Further inspection and evaluation are
needed to ensure the stability of the slope. If the anchor load falls below 80% of the
design load, it may result from the creep of the fixed end, the shortening of the free end,
slip between wedges and strands and corrosion of anchor head components. If no sign
of slope instability was observed, the anchor load was likely to be in balance with slope
mass. No immediate action needs to be taken.
Table 1 shows the scores of an example anchor obtained from the above-mentioned
inspection process. This example anchor got a score of 70.75 and was graded as “Fair” (Table
2). Its residual load was between 0.8 to 1.1Tw. However, it got a low score on strands
corrosion (seriously corroded). This is the type of anchor which needs further attention. The
residual anchor load was higher than the design. It indicated that the slope might have some
downward movement. Since the strands were seriously corroded, a sudden failure might
occur on this slope due to strand breakage.
After gathering the inspection results from the slope anchors, the overall score of this
particular anchored slope can be obtained by adding up the total scores ( ) of each inspected
anchor and then dividing by the number of inspected anchors. The average overall score ( )
can be obtained and used to category grade this anchored slope (Table 3).
3 Remarks on lift off test
Among the four steps inspected, the scores obtained from steps No. 1 to 3 are qualitative.
Only step No. 4 can yield the quantitative lift off load. But the lift off test is costly and not
easy to work on if the anchors are high in the anchored slope. It might be useful to correlate
the images taken from the endoscope inspection with the residual anchor loads determined
from the lift off test.
Five anchors from the anchors remained on the failed slope of Freeway No. 3 were chosen
for lift off tests (Fig. 6). The results of the lift off tests, such as lift off load and maximum
applied load, are listed in Table 4. Among the five lifted off anchors, two yielded residual
loads of more than 90 ton (=1.5Tw, Tw = design anchor load = 60 ton); two yielded lift-off
loads of 43.65 ton and 54.8 ton (less than Tw). For the latter, steel strands broke when the
load was further increased to 50 ton and 60 ton respectively. The last one yielded a lift-off
load at 65.9 ton (lager than Tw) but strands broke shortly after the load increased to 68.7 ton.
The ultimate load of corroded anchors roughly varied from 45% to 80% of the yield load of
steel strands depending on the seriousness of corrosion.
Fig. 7 shows the endoscopic image taken before the lift off test. In general, all anchors
were subjected to serious strands corrosion and should be classified as unacceptable
following the BSI requirements for ground anchorages . In fact, some wires of the strand
were broken even before running the lift off test (Anchor III). Some strands were in moist
condition and weeds grew inside the anchor hole (Anchor V). But there is no clear correlation
between the remained anchor capacity (lift-off load) and the extent of surface corrosion of
steel strands. For example, the surface corrosion of steel strands is no better than Anchors III,
IV, and V. But the maximum pull-out load of Anchors I &II was about 50% higher than the
other three anchors. No strand breakage occurred in Anchors I & II while Anchors III, IV,
and V showed strand breakage at their maximum loads. It was also observed that the wires
in the strands suffering the most serious corrosion or subjected to the most stressing load
broke first during stressing. In other words, a wire-by-wire breaking pattern within a strand
was observed; followed by a strand-by-strand breaking pattern within an anchor. After the
breaking of an individual anchor, its load was passed to other anchors and subsequently
caused a chain-reaction type of anchor failure. As a result, a sudden slope failure occurred
As shown in Fig. 8, five out of seven strands of Anchor III were broken at a location
within 1 m from the anchor head during the lift off test. Two strands remained in the anchor
hole without breakage. When examining the broken strands, the surface of the strand was in
wet condition (Fig. 8).
Fig. 9 shows the close-up look of each broken strands of Anchor III. Obviously, every
strand suffered different extents of corrosion. Those (No. 3 & 5) with less corrosion could
take more load during lift off test and showed some trace of intact steel at the broken face;
those with more corrosion showed no trace of intact steel and were expected to take much
less loading. In other words, once the strands in the anchor hole begin to corrode, their
breakage capacities decrease but not at the same paces. Hence, when the strand is stressed to
its breaking point, the wires in the strand broke in a one-by-one pattern rather than in a
grouped pattern. So, the overall pullout capacity of a rusty anchor is not the summation of all
the strands in an anchor. For example, for the three anchors broken during lift-off test,
anchors broke at a loading only marginally larger than the lift-off load, but significantly
smaller than the design anchor load.
4 Alignment of anchor head and steel strands
Anchor consists of several components such as steel strands, anchor head, and anchor hole.
In many cases, the face of ground retaining structure is not perpendicular to the alignment of
anchor hole. So an angle adjustment plate is needed to keep the anchor head in the
perpendicular position to the anchor hole. The example shown in Fig. 10 is to illustrate the
failure to use an angle adjustment plate to maintain a perpendicular position of anchor head
to the steel strands coming out from the anchor hole. Having such a situation, it will damage
the steel strands and also reduce the stressing load transferring to the fixed end of the anchor.
In addition, the wedges will not be able to grip the steel strands firmly on to the anchor head
due to unevenly distributed loads among the wires in the steel strands. Therefore, the
alignment control for all the components of the ground anchor is crucial for the anchor
construction. It requires precision, practice and patience from the ground anchor contractor.
In fact, the angle adjustment plate can do more than just adjust the angle. It can also be used
as an indicator for the soundness of free end cement grouting of the anchor. Fig. 11 shows a
specially integrated angle adjustment plate and bearing plate assembly. This bearing plate
assembly consists of (1) an extension pipe with a rubber seal to prevent groundwater from
seeping into the inside of the plastic sheath and moistening the unsheathed bare steel strands
right under the anchor head; (2) grouting opening and ventilation hole to completely fill up
the annual space between the plastic sheath of the free anchor length and drill hole; and (3)
the angle adjustment plate to keep the anchor head in alignment with the anchor hole. Cement
grout is poured into the opening of the assembly (Fig. 11), a ventilation hole is predrilled on
the bearing plate to facilitate the cement grouting process under the anchor head. The space
inside the extension pipe will also be filled with cement grout or anti-corrosion grease later.
In addition to the bearing plate assembly, a completed cement grouted ground anchor also
includes the anchor head assembly and other parts of anchors as schematically shown in Fig.
12. It should be noted that the grout seal device which is typically used in the traditional
anchors to separate the free end grouting and fixed end grouting had been removed from the
anchor assembly to facilitate the grouting process. Originally, this grout seal device is to
prevent the cement grout from flowing into the free anchor end during the fixed end grouting
and allow the strands in the free end elongate during anchor stressing. But this function can
be taken over by the strand assembly shown in Fig. 12. Each individual steel strand was
sheathed with PE tube on the free anchor length and the PE tube was sealed at the bottom
with heat shrink tube. Without the grout seal device, it can make the cement grouting work
of the anchor much easier and with a better grouting quality. The effectiveness of water
tightness of anchor can be tested by electrical resistance measurement method if required.
Lastly, the alignment of the stressing jack with the anchor hole is also an important factor
needed to address. The hydraulic jack itself is quite heavy. It needs great patience and efforts
to keep it in line with the anchor hole when the anchors are high up on the slope and with an
inclined angle. Under this difficult construction condition, it is not easy to keep a good
alignment between the jack and anchor hole. Failure to keep the alignment may result in
problems such as unevenly stressed strands and wedges. Fortunately, this problem can be
solved easily by installing locating pins on the bearing plate to keep the jack in the right
position (Fig. 13). With the locating pins, the jack can be mounted to the bearing plate at the
right position. So the load reading from the jack can be much closer to that from the load cell
as demonstrated in Fig. 13.
It is understood that the anti-corrosion capacity of the ground anchor can be significantly
improved by using reinforced fibre glass or carbon fibre strands or using the epoxy coated
steel strands to replace the traditional steel strands. However, the government system is
generally slow to pick up new materials or new methods for the civil engineering work,
especially the cost of the new materials or new methods tends to be higher. If not using new
material, then using the factory-made ground anchor assembly can be an alternative. Factory
made anchors can provide a better control on the details of anti-corrosion measures compared
to those assembled on the job site. However, due to the fact that factory made pre-fabricated
anchor cannot always meet the changing ground or construction conditions on site, the
prefabricated anchors are not popular in the ground anchor industry. The proposed ground
anchor system shown in Fig. 12 does not involve any new material or require any new
construction methods. But it can provide the same anti-corrosion function needed for the
permanent use of anchors. So, it is relatively easier to be accepted by the government agencies
and also by the engineering design companies.
5 Monitor the anchor load change
Anchor load monitoring is an important practice for monitoring the stability of an anchored
slope. But since ground anchors are mostly pre-stressed during the construction stage, so it
is the change of stressed load which should be the concern of an anchored slope rather than
the residual anchor load. For example, a clear load increase of anchor load on an anchored
slope can be an indication of downward sliding of the slope. But long-term measuring of the
anchor load change is not a straightforward task. Typically, anchor load change is measured
with the electrical load cells or by lift off test. Although the electrical load cell is good at
measuring the locked-in anchor load with high accuracy and can be linked to the automatic
slope monitoring network, it can only survive for a limited period of time when used in an
outdoor environment . On the other hand, the lift off test is rather simple in principle but
often has site accessibility problems when carried out on the existing anchored slope. Due to
the above-mentioned restrictions, both load cell installation and lift off test cannot be carried
out in large numbers. But limited numbers of anchor testing can further complicate the
problem because it will be difficult to evaluate the status quo of the slope stability based on
a limited amount of data, especially if these data are themselves scattering. To solve this
problem, a Smart Anchor which can reliably measure the anchor load change over a long
period has been developed and implemented in Taiwan .
The anchor load change monitoring device of Smart Anchor is similar to the tell-tale
device  in principle. The tell-tale, which uses an unstressed rod mounted alongside a
stressed structure member, can be used to indicate the change in length of the stressed
member. The change in length is then converted to strain or change in load provided that the
length of the stressed structure member is known. Nevertheless, the tell-tale is actually a
foreign object mounted to a strand of ground anchors; thus, extra care is required to facilitate
the survival of the tall-tale during anchor construction. Practically, successfully installing a
tell-tale is difficult during routine anchor construction. The method proposed in this paper is
to convert the tell-tale device to become part of the anchor itself. Basically, this method alters
nothing in the anchor assembly but introduces one extra strand as the reference strand. As
depicted in Fig. 14, the reference strand is not connected to the anchorage head by omitting
the lock-in wedges. In other words, the reference strand is not engaged in the movement of
the anchorage head. So when the anchorage head moves (i.e., anchor load changes) because
of slope movement, deterioration of anchor components or any other causes, the reference
strand does not elongate or shorten as other engaged strands do. Then a relative deformation
of the reference strand to the engaged strands is generated. If the anchor load decreases, the
reference strand extends outward with respect to other engaged strands (negative , Fig. 14c).
On the other end, if the anchor load increases, the reference strand is subsided (positive ,
If the measured relative deformation ( ) of the reference strand is known, the change of
the anchor load ( P) can be estimated from the following equation:
where is the relative deformation of the reference strand in response to anchor load
change; E is Young’s modulus of steel strand and equals to 2000 t/cm2; A is the total
crosssectional area of all engaged steel strands (A = 0.9871 cm2 for a 7-wire strand with a nominal
diameter of 12.7 mm; A = 1.3870 cm2 for a 7-wire strand with a nominal diameter of 15.2
mm); and Leff is the effective free strand length.
Three test anchors were used to examine the effective free length of working anchors.
The assembly of all test anchors was exactly the same, as that illustrated in Fig. 12. Each
anchor used seven 12.7 mm steel strands (Grade 270) with a design free length of 15 m and
design fixed length of 10 m. Among the strands, six were engaged to the anchorage head and
one was used as the reference strand. In this field test, several pre-determined loading cycles
were applied to the anchors during the anchor suitability test . The initial length of the
reference strand extruding from the head of the jack was measured using a caliper. Repeating
this procedure for each loading cycle and then subtracting the initial reading is done to obtain
the relative deformations of the anchor head at different loadings. Since the deformation of
the reference strand was measured from the head of the jack, the free length of this test should
be the summation of the sheathed strand length and the strand length inside the jack and load
cell (=1 m). Through a substitution of the measured relative deformations ( ) and the anchor
load changes at each corresponding loading cycle into Eq. 1, Leff of the test anchors was
calculated and compared with the design free anchor length in Fig. 15. In general, there is
only 1%–2% (0.16 m/16 m or 0.34 m/16 m) difference in length, demonstrating that the
calculated effective free length (Leff) was very close to the design (i.e., sheathed) free length
in the anchor assembly under a working anchor load. Thus, if the anchors were assembled as
shown in Fig. 12, the design free length could be used directly in Eq. 1 for the calculation of
anchor load change.
The residual load (Pr) of the anchor at the time that is measured is equal to the
summation of the anchor load change P and the initial locked-in load (Pi) of the anchor:
Pr =Pi +ΔP
Three field anchors were used to check the locked-in loads with lift-off tests to verify the
accuracy of the proposed method. Each anchor used 7 strands (12.7 mm- ) with the design
free length of 15 m and design fixed the length at 15 m. Among them, 6 were engaged strands
and one was the reference strand. Prior to the test, a set of split rings (approximately 1 cm in
thickness) was placed under the anchor head of the test anchors. Lift-off test was performed
to determine the locked-in anchor loads before and after the removal of the split ring. As
shown in Fig. 16, the reference strand clearly extruded out from the engaged strands after the
split ring was removed and the load was reduced. The threads that appeared on the anchor
head in the photo were for the stressing of the lift-off test. But the load change measurement
method proposed here can be used easily with any regular anchor heads. The load change
determined from the lift-off test was compared with that calculated from Eq. 1 by using the
relative deformations of the reference strand measured before and after the removal of the
split ring (Fig. 17). Table 5 lists the test anchor data, results from the lift-off test, and
calculated loads. In general, the load change calculated from Eq. 1 was in good agreement
with that determined from the lift-off test. The average difference ranges from 1.4 % to 4.7
% relative to the initial locked-in load (Pi). This indicates that this simple method can be
satisfactorily used to monitor the long-term anchor load change with reasonable accuracy.
a P1: residual load before split ring removed
b P2: residual load after split ring removed
c ΔPmeasured: measured anchor load change
d ΔPcalculated: calculated anchor load change from Eq. 1
6 strands (12.7mm-) per anchor engaged with anchorage head
Design free strand length = 15m, Design fixed length = 15m
e ΔPdiff: difference of anchor load change = abs [ΔPmeasured-ΔPcalculated]
In 2010, the sudden failure of a tied back cut slope of National Freeway No. 3 in Taiwan had
revealed the problems of ground anchors of anchored slopes in Taiwan and changed the
practice of design, construction, and maintenance of the anchored slopes. Since voids under
the anchor head were found in the majority of ground anchors and the steel components of
anchor were corroded at different extents, anti-corrosion measures had been used to prevent
the corrosion from happening on the existing anchors as well as the new anchors. The
following conclusions are drawn from the anti-corrosion exercise on anchors in the anchored
slopes in Taiwan:
(1) Not properly sealed voids underneath the anchor head was found to be the main area of
steel strands corrosion on the existing anchors. It was treated by sealing off the voids
with cement grout to stop further corrosion. For the new anchors, a slightly modified
strands assembly and anchor head assembly was used to upgrade corrosion protection
without introducing the non-traditional ground anchor materials. The seal device which
was commonly used to separate the fixed end grouting from the free end grouting of the
anchor is removed from the new ground anchors to facilitate the grouting process and
to minimize the risk of not filling up the whole anchor with cement grout.
(2) A new anchor strands assembly method is proposed to measure the load change of
prestressed anchors. It makes the load change monitoring become part of the anchor itself.
So it is simple, inexpensive, reliable, and above all very durable. It can be done easily
during anchors construction. Although its accuracy may not be as good as the electrical
load cells, it is good enough for practical purposes and can be used in large quantities.
Having such an abundant load change information of tieback anchors, the engineers will
be able to evaluate the stability of anchored slopes in a more confident way.
The Authors wish to thank the Directorate General of Highways, the National Freeway Bureau of
Taiwan government and the Department of Rapid Transit System of Taipei city government for
providing financial support and test sites to carry out the ground anchors experiment works for this
research. The Authors also wish to thank the local ground anchor contractors and engineers for
providing technical support and suggestions throughout this study.
British Standard Institute (BSI DD81 , BS 8081) (1989) British Standard Code of Practice for Ground Anchorage .
Dunnicliff , J. ( 1988 ) Geotechnical Instrumentation for Monitoring Field Performance , John Wiley, New York.
ISO/DIS 22477-5: Geotechnical Investigation and Testing - Testing of Geotechnical Structures, Part 5 , Testing of Anchorages, International Organization for Standardization, Geneva, Switzerland, 2010 , www.iso.org.
Lee , Wei F. , Liao , H. J. , Chang , M. S. , Wang , C. W., S. Y. Chi , and Lin , C. C. ( 2013 ) “ Failure Analysis of a Highway Dip Slope” , Journal of Performance of Constructed Facilities, ASCE , 27 , No. 1 , pp116 - 131 Liao , H. J. and Cheng, S. H. ( 2011 ) “ Failure cases of Anchors and Anchored Slopes in Taiwan” , Proc. of the 5th Cross-strait Conference on Structural and Geotechnical Engineering , Hong Kong.
Liao , H. J. , Lee , Wei F., and Wang , C. W. ( 2013 ) “A Tale of Twin Cut Slopes in Taiwan” , Forensic Engineering , Proceedings of the Institution of Civil Engineers , 166 , Issue 2, pp. 72 - 80 , doi: 10.1680/feng.12.00024 Liao, H. J. and Cheng, S. H. ( 2014 ) “Overhaul the Anchored Slopes in Taiwan” , Proc.
of 6th Japan-Taiwan Joint Workshop on Geotechnical Hazards from Large Earthquakes and Heavy Rainfalls , Kita-Kyushu, Paper No. TW024 .
Liao , Hung-Jiun, Cheng, Shih-Hao, Chen, Huang-Ren and Chen, Chun-Chun (2017a) “A simple method to measure long term load change of ground anchors,” Geotechnical Testing Journal , March 2017 Volume 40 , Issue 2GTJ20160110 Liao, Hung-Jiun, Cheng, Shih-Hao, Chen, Huang-Ren and Chen, Chun-Chun (2017b) “Cement grouting to seal off voids below anchor head , ” Proc. of Grouting , Deep Mixing, and Diaphragm Walls ( Grouting 2017 ), Honolulu.
10. Taiwan Geotechnical Society (TGS) ( 2011 ) Forensic Study on the Dip Slope Failure at Chainage 3 . 1k of National Freeway No.3 , Taiwan ( in Chinese)