Anchor design

Design of vertical anchors

Vertical anchors may be precast concrete panels or steel plates set upright and incorporated
into the backfill behind a retaining wall. At the appropriate level of the backfill, a tie rod is connected
between the plate and the wall. Alternately tie rods can be driven or placed through augered holes.
They are usually “passive” in two senses: (i) only a slight prestressing of the tie rod if usually applied;
and (ii) the resistance to the horizontal movement of the plate comes from the passive soil resistance.
At the same time as movement develops passive resistance on the front face of the plate, active soil
pressure will develop on the back. However, if the height of the anchor is small compared to the
embedment depth the anchor will fail by ploughing through the ground producing a shear plane
extending to the ground surface.

The majority of research into anchor behaviour has been experimental and much design
practice is based on empiricism. Experimental work both under “normal gravity” conditions and
centrifuge conditions have been used to derive semi-empirical relationships. Theoretical approaches
such as limiting equilibrium, cavity expansion and limit analysis have been used. Merifield and Sloan
(2006) presented charts to determine upper and lower bound anchor pullout loads based on finite
element formulations for frictional soils. Results obtained from this study compared well with existing
theoretical and laboratory studies. Merifield et al. (2001) present similar charts for undrained clays.

Equating work done by external loads and internal stresses for a kinematically admissible failure
mechanism allows the load at which collapse occurs to be determined (i.e. an upper bound). It is also
possible to equate stresses in the soil and those generated by external loads such that the strength
of the soil is not exceeded. This gives the load up to which failure cannot occur (i.e. a lower bound).

Box 1: Merifield and Sloan (2006) method for vertical anchors

Lower bound Upper bound

1
 Merifield and Sloan (2006) found that active zones of failure can form behind the anchor when
H/B ≤ 2 and φ’ ≤ 20°. In these cases ignoring this active zone directly behind the anchor will over
predict the collapse load by up to 18%. For H/B > 2, any active soil pressure behind the anchor has
little influence on the ultimate collapse load.

Anchor roughness was found to have a significant effect on the capacity of vertical anchors.
Rough interfaces can have a pull out capacity 67% greater than that of smooth interfaces. The effect
of anchor roughness was found to decrease with an increase in the embedment ratio and is most
significant for dense soils with high friction angles (φ’ ≥ 40°).

Soil dilation was found to have significant effect on vertical anchors. In the extreme case of a
vertical anchor in a dense non-dilatant soil, the ultimate capacity was estimated to be approximately
half of that for the same anchor in an associated soil (that is φ’ = ψ). However, because of localisation
Merifield and Sloan (2006) suggest that further work should be done and this result should be treated
with caution.

Design of horizontal anchors

This type of foundation is commonly used on electricity transmission pylons to provide

resistance against uplift. Because of this frequent use, and the interesting mode of failure of the soil
above the anchor plate, it has been the subject of a great deal of research starting from approximately
1955.

In observations of a “sectioned” model test (that is the soil and half a foundation adjacent to a
glass wall) the following sequence of events occurs as the upward load is increased:

1. The soil above the bearing plate part of the foundation is consolidated and distorted, and
the plate moves up slightly, leaving an open space below it (see Figure 1). The movement
pattern of the soil is complex, with the soil moving mainly upwards, but also laterally away
from the centre. In the zone close to the edge of the plate high distortion produces significant
dilation (if the soil is dense and dilatant), otherwise the majority of the strain is consolidation
under the bearing pressure exerted by the plate.

Figure 1: Soil movements and strains above a "shallow" anchor

2
 2. A cone of soil with curved sides breaks loose from the underlying soil. The failure surface is
clearly associated with shear at the bottom end, but towards the top, there is a separation
by an opening crack between the moving and stationary soil, indicating a possible tension
mode of failure.

The shape of the failure surface varies from that shown in Figure 2 (a), for a very shallow
H/D, to that shown in Figure 2 (b) for H/D ≈ 5. With increasing H/D the amount of upward
movement that occurs, before the moving cone of soil breaks loose also increases. It is
difficult to define just at what point on the load-movement graph of the anchor that the soil
body breaks loose. However, it is after the peak load for all but very shallow anchors
(H/D < 1).

Figure 2: Zone of moving soil in "shallow" failure mode

The mode of failure of the soil above the shallow anchor (i.e. the clear separation by an opening
crack of a moving body of soil) has two implications:

1. Since the moving body has sufficient coherence to lift away from the stationary soil, the full
self-weight of the body must be contributing to the maximum pullout force.
2. Where vertical separation on a surface occurs, there can be no shear stress developed.

Very complex expressions have been proposed by researchers to define the shape of the
moving soil cone, and to calculate the shear stress acting on the surface between it and the stationary
soil. However, these are not justified in view of the fact that: (a) some breaking up of the soil cone has
been seen to occur in model tests; and (b) the extent of shear, reduced normal stress on the failure
surface, and tensile failure are not known.

3
 A reasonable simplification of the size and shape of the moving soil cone is shown in Figure 2
(c) and a reasonable compromise between the unknown extents of the contributions of the anchor
pullout load is to calculate it with the following expression:

𝑄𝑄𝑢𝑢 = 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡 𝑜𝑜𝑜𝑜 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 + 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡 𝑜𝑜𝑜𝑜 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐

+ 𝑆𝑆ℎ𝑒𝑒𝑎𝑎𝑟𝑟 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑜𝑜𝑜𝑜 𝑡𝑡ℎ𝑒𝑒 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑡𝑡ℎ𝑒𝑒 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓

The assumption that the shear stress acts only on the area of the cylinder above the bearing
plate part of the foundation (see Figure 3) is a compromise between the extremes of (a) assuming
shear under the overburden acts over the full failure surface, and (b) assuming the uplift force
removes all normal force from the failure surface and hence eliminates all shear.

Figure 3: Assumption for calculation of pullout load

Meyerhof and Adams (1968) assumed that the normal force Fn on Figure 3 (used to calculate
Fv) is that derived from the horizontal passive pressure. This gave good correlation with their test
results. In a dilatant material, such that a vigorous expansion occurs in the deforming zone above the
anchor plate, then this may well be the case. However, a better assumption would be to use K0 rather
than Kp, since the degree of conservatism compensates for the situation of low dilation, and also, to
some degree, for the fact that the moving body of soil may break up, and not impose its full weight.

More recent work based on the finite-element method by Merifield and Sloan (2006) has shown
that the failure mode of horizontal anchors consists of the upward movement of a rigid column of soil
immediately above the anchor, accompanied by lateral deformation extending out and upwards from
the anchor edge. As the anchor is pulled vertically upwards, the anchor tends to lock up as it attempts
to dilate during deformation. The effect of anchor interface roughness was found to have little or no
effect on the calculated pullout capacity for horizontal anchors at all embedment depths and friction
angles analysed. Merifield and Sloan (2006) presented a simple method to estimate the pullout
capacity based on a break out factor, as used above for vertical anchors, for frictional soils. Merifield
et al. (2001) present similar charts for horizontal anchors in undrained clays.

4
Box 2: Merifield and Sloan (2006) method for horizontal anchors

Even H/B ratios Odd H/B ratios

Dotted lines are upper bounds and solid lines are lower bounds.

5
Long cylindrical anchors

Tension and compression anchors

A tension anchor is one where the tendon adheres to the grout throughout the full fixed length
of the anchor. When a tension force is applied to the tendon, load begins to shed into the soil at the
front of the grouted body, which when attached to the tendon is also in tension. Tension anchors are
the most common type of anchor. A typical tension anchor is shown in Figure 4.

Figure 4: Typical tension anchor (from FHWA-IF-99-015)

Typical forms of compression anchor are shown in Figure 5. The essential common feature is
that the tendon is not attached to the grout, but passes freely through a sleeve to the back end of the
anchor’s fixed length, where it is attached to a plate. This plate bears onto the grout column, which
goes into compression, and which then sheds load into the soil, beginning at the back end. For high
load capacity anchors, the bearing stress in the grout immediately under the plate is very high, and
the grout is sometimes reinforced with a spiral. This may still be insufficient to prevent crushing of the
grout, in which case a steel tube with outside corrugations is used to spread the compression force
more gradually into the grout body. Compression anchors can offer greater corrosion protection as
the grout remains uncracked and the tendon is sheathed along its entire length. However, they are
considerably more difficult to construct compared to tension anchors.

Composite anchors have also been used, in which the tendon is partially sleeved along the fixed
length as shown in Figure 6.

6
Figure 5: Typical compression anchor (from FHWA-IF-99-015)

Figure 6: Typical composite anchor (from FHWA-IF-99-015)

7
Anchor design

The design of long cylindrical anchors depends on four main components:

1. Load transfer into the ground

2. The type and strength of the tendon
3. The location of the fixed anchor zone, together with the uplift or pull-out capacity
4. Load transfer into the structure

Of particular importance to this course is the load transfer between the ground anchor tendon
and the soil within which it is embedded. This depends upon the design of three boundaries: (i)
anchorage grout to ground (fixed anchor length); (ii) encapsulation to anchorage grout, if appropriate
(encapsulation length); and (iii) tendon to anchorage grout or encapsulation grout (tendon bond
length). The relationship between the design anchorage force (Tdesign) and the required fixed anchor
length (L) is usually determined analogous to the design of piles:

𝑇𝑇𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 + 𝐸𝐸𝐸𝐸𝐸𝐸 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

The end bearing component is usually ignored in design, however in some cases if pressure
grouting or under-reaming is used to create an enlarged base it may be included. Ignoring any
resistance due to end bearing results in the following formula:

𝑇𝑇𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝜋𝜋𝜋𝜋𝜋𝜋 × 𝜏𝜏𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑

The first component (πdL) is the surface area of the fixed anchor length. Determining a
representative diameter is not a simple task and is often taken simply as the diameter of the drilled
hole. Values for τdesign are derived from a wide range of sources, such as pile-related bearing capacity
formulae, back analysis of pull-out tests on specifically constructed test anchors and the like.

Fixed anchor length design generally focuses on three convenient ‘groups’ of anchoring
material: granular soils, clayey soils and rocks. It will be realised that there can be some blurring of
boundaries between these idealised groups, as intact rocks weather towards ‘soils’, for instance.

As a general rule (e.g. BS8081) the fixed anchor length should be:

a) A maximum of 10 m long (because of a trend of decreasing capacity per metre run, as

discerned from back analysis of load testing). However, multi-stage anchors are able to
employ longer load transfer lengths by means of ‘stacking’ fixed anchor lengths.
b) Not less than 3.0 m; although for design working loads below 200 kN, a minimum length of
2.0 m may be acceptable in suitable circumstances.

Fixed anchor length in granular soils

A review of various proposed equations (Turner, 2012) suggested that the load-carrying
capacity in granular soils can typically be expressed by a relationship in the form of:

� 𝑦𝑦 tan 𝜙𝜙′
𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢 = 𝜋𝜋𝜋𝜋𝜋𝜋 × 𝐾𝐾𝜎𝜎′

� 𝑦𝑦 is the average effective overburden pressure along the

Where K is an empirical coefficient; 𝜎𝜎′
load transfer length; and 𝜙𝜙′ is the angle of shearing resistance of the soil. It can be seen that this is
very similar to the classic formula for calculating the ultimate skin friction of a pile in granular soils.

8
 In conventional pile design, typical values for K will vary between 0.7 and 0.9 for a bored pile,
but there might also be a reduction in 𝜙𝜙′ to account for soil loosening. This range of K for bored piles
can be compared with the much higher values obtained for ground anchors between 4 and 9 by Moller
and Wilding (1969) and 1.7 to 6.8 by Littlejohn (1980). (The higher Moller and Wilding values might
be explained because these results also include an element of end-bearing component considered
separately by Littlejohn.)

It can be envisaged therefore that the increase in K value exhibited by ground anchors in
granular soil, compared with ‘conventional’ piles, can be attributed to:

1. Scale differences between a typical pile diameter of perhaps 450 to 750 mm, compared with
a ground anchor of 100 to 200 mm.
2. Penetration of the fluid grout into the surrounding granular soil, so increasing the effective
diameter of the fixed anchor length.
3. The influence of grouting pressure on anchor capacity (by both permeation and by
increasing the confining pressure around the fixed anchor length). The techniques of
pressure grouting includes, in order of increasing load capacity: grouting under hydrostatic
head only, low-pressure grouting (less than hydro-fracture pressures), or high-pressure
post-grouted systems (using tube-a-manchette, or similar, techniques).
4. The particle-size distribution of the granular strata: where coarser soils will, in general,
permit a greater penetration of the anchor grout.

With respect to high-pressure post-grouted systems, empirical design curves for such
techniques in sands and gravels are illustrated in Figure 7.

Figure 7: Empirical relationships for the determination of the ultimate skin friction for sand and gravel using IGU
and IRP post-grouting techniques (from Turner (2012))

9
Fixed anchor length design in clayey soils

A typical design relationship for clayey soils is expressed in the form:

𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢 = 𝜋𝜋𝜋𝜋𝜋𝜋 × 𝛼𝛼𝑠𝑠𝑢𝑢

Where su is the undrained shear strength; and α is an adhesion factor. Again it can be seen that
this design approach is similar to that employed in bored-pile design practice.

Design is almost always based on undrained shear strength and total stress analysis rather
than using the more fundamental concept of drained strength parameters and effective stress
analysis. Drilling and grouting activities, usually using auger or water flush systems and fluid cement
grouts, are difficult to model reliably using effective stress design. In addition, load-testing is of short
duration compared with the time for steady-state effective conditions to stabilise. Hence, the testing
mode is more suited to a total stress approach.

It should be noted that increasingly, anchor load testing involves not only investigating the
ultimate ‘short term’ load-holding capacity but also confirming acceptable creep values.

For straight-shafted tremie grouted anchors with low or no pressure, quoted α values for stiff
clay (su > 90 kPa) are typically around 0.3-0.35. For very stiff to hard clay (su = 287 kPa) a design
value of 0.45 is suggested.

For multi-stage anchors, each strand is deemed to have its own short individual load transfer
length within the overall fixed anchor length. Pullout test data suggests this allows the use of markedly
higher α values for design; values as high as 0.95 or 1.0.

For post-grouted anchors, additional grout is pumped under high, controlled pressure into the
fixed anchor length using valved injection tubes, once the primary grout has achieved its initial set.
Design chart for such anchors are given below.

10
Figure 8: Ultimate skin friction in cohesive soils for various anchor lengths, and with and without post-grouting
(from Turner (2012))

11
 Figure 9: Influence of post-grouting pressure in cohesive soils (from Turner (2012))

Figure 10: Empirical relationship for the determination of the ultimate skin friction for 'silty clay soils' using the
IGU and IRP post-grouting techniques (from Turner (2012))

Fixed anchor length design in rock

A typical design relationship for anchors in rock is of the form:

𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢 = 𝜋𝜋𝜋𝜋𝜋𝜋 × 𝜏𝜏𝑢𝑢𝑢𝑢𝑢𝑢

Where 𝜏𝜏𝑢𝑢𝑢𝑢𝑢𝑢 is a value for the ultimate (or ‘failure’) bond stress to be used for design, based upon
previous pullout test data or established by empirical relationship. Established data may be site-
specific or based upon tests in similar or related strata or conditions. Almost all design bond-value
data is empirical, based upon published case studies of field experience.

For strong intact rocks, an ultimate bond value equal to 10% of the unconfined compressive
strength of the rock is often recommended. A maximum value of 𝜏𝜏𝑢𝑢𝑢𝑢𝑢𝑢 of 4.0 N/mm2 is usually

12
recommended, based upon a typical design of the unconfined compressive strength of cement grout
of 40 N/mm2.

Table 1: Presumptive average ultimate bond stress for ground/grout interface along anchor bond zone (FHWA-IF-
99-015)

13
Mechanics of cylindrical grouted anchors

Distribution of surface shear stress

Consider the application of load to the grouted length of a tension anchor. The leading end of
the grout body will move, inducing shear stresses between the grout and the soil, and shedding load
from the anchor into the surrounding ground. A rough estimate of the load distribution along such a
grout body is shown in Figure 11 (a), together with a formula for the movement along the leading end
caused by elastic extension of the steel tendon alone. If the stress in the steel were 300 N/mm2, and
the grouted length 5 m (typical values) then the extension of the steel at the leading end would be
2.5 mm.

Figure 11: Anchor surface shear stress, fundamental considerations

In Figure 11 (b) is shown a typical shear stress – movement relationship for shearing on a plane
in soil. For a dense soil, the movement to peak shear stress is in the region of 1.5 to 2.5 mm. Therefore
it can be seen that the stretch of the grouted body is sufficient to produce a wide variety of shear
stress levels along its length as the load is applied.

From measurement of strain on the steel tendon, using electrical resistance strain gauges, the
load in the tendon can be deduced. From knowledge of the stress-strain behaviour of the grout, the
load in the whole grouted body can be deduced. The difference in load between two points must be
the load shed into the soil. Hence the mean surface shear stress (τs) at the grout-soil interface
between two such points is the difference in load divided by the area of that interface.

Various researchers have used this technique to produce τs distributions. An early example is
that of Ostermayer and Scheele (1978) shown in Figure 12. The migration of the peak τs backwards
along the anchor is clearly apparent. The maximum load capacity would be reached if, for an
attempted increase in load, the anchor move so that the peak τs movement caused the area under
the τs distribution graph to decrease.

14
 Figure 12: Development of τs distribution as load on anchor increases (from Ostermayer and Scheele (1978))

Practical experience with anchors on site has shown that if the grouted length is increased in
an attempt to increase the maximum load capacity, then a “law of diminishing returns” applies,
particularity in stiff ground materials. The reason for this is illustrated in Figure 13 from Ostermeyer
and Scheele, which shows the τs distribution at ultimate load for a 2 m and a 4.5 m grouted length.
On the 4.5 m long anchor, the maximum value τs is located towards the rear of the anchor, and at the
front, the level of τs has dropped to its residual value. On the short, 2 m length anchor, the peak region
of the τs distribution occupies the whole anchor. Therefore the average τs is higher for the short anchor
than the long one, with the result that the short one works more efficiently in terms of load capacity
per unit length.

15
 Figure 13: τs distribution at the ultimate load on a 2 m and a 4.5 m anchor (from Ostermayer and Scheele (1978))

Although long anchors are less efficient in the manner just mentioned, from a practical point of
view, the relatively small amount of money spent on making an anchor slightly longer than the
minimum for the required load brings three returns:

1. The maximum load capacity will be increased, if only by the residual τs value
2. The probability of an excessively large portion of the grout body being in a weak zone
of ground is reduced.
3. The movement of the anchor head required to bring the grout body to maximum load is
increased, hence increasing the energy absorption capacity of the anchor, and also
improving the amount of warning given before ultimate failure.

Some attempts have been made to include efficiency factors to account for the reduction in
ability to carry load as the fixed length increases. Examples of such relationships from Barley (1997)
are given below. However, these remain somewhat empirical and case specific.

16
 Figure 14: Anchors in Stiff to Very Stiff Clays, Diameter 105 to 190 mm (from Barley (1997))

Figure 15: Injection Anchors Using 152 mm Casing in Alluvial Soils, Taipei (from Barley (1997))

A complication in the derivation of the shear stresses on the surface of a tension anchor’s
grouted length from measurements of strain in the steel tendon, is whether or not the grout is cracked.
Figure 16 (a) shows a hypothetical anchor with a uniform τs distribution. Figure 16 (b) shows the
resulting linear load distribution along the anchor, and, making the assumption that the grout is

17
uncracked, the division of the load between the steel (Ps) and the grout (Pg) is shown. If the strain
gauges were fixed to the steel in this anchor, the strain distribution would be as shown by the line ab
in Figure 16 (c). However, if at all strains greater than, say, 150με the grout were cracked at very
close centres, then at point c, the steel would suddenly take all the load, and the graph of the
measured steel strain would follow the path acde.

Figure 16: Simplified anchor behaviour

A further aspect of anchor behaviour as revealed by strain gauge measurements on the tendon
of a tension anchor is shown in Figure 17, from Shields et al. (1978). In this test the load was first
raised to a high value of 588 kN, and then lowered to a nominal value of 6 kN. Figure 17 shows the
load and the shear stress distributions for these two states of loading. A significantly large tensile
force (13% of the originally applied load) is left locked in the anchor. It is kept there by shear stresses,
applied by the soil to the anchor, of the magnitude of 21% of those induced by the original load. The
mechanism is that when the anchor head is unloaded, the tendon and grout body rebound backwards,
but because a large amount of plastic strain has occurred in the soil round the front end of the anchor
during initial loading, the soil develops shear stresses to oppose the rebound.

18
 a) Load distribution

b) Surface shear stress distribution

Figure 17: Load and shear stress distributions on a loaded and unloaded anchor

All instrumented anchors reviewed so far have been the tension type. Mastrantuono and
Tomiolo (1978) tested both a tension and a compression anchor in the same soil. The distributions of
strains along the grouted bulbs are given in Figure 18. No indication of the diameter of the grouted
zone of these anchors was given by Mastrantuono and Tomiolo (1978), hence τs values cannot be

19
calculated. However, Figure 18 (b) shows the relative shapes of the τs distribution, being the change
in the strain between adjacent points, divided by the distance between them. (It is to be noted that
different conversion factors would be needed for the tension and compression anchors, to convert the
graphs of Figure 18 (b) to τs because of different stiffnesses in tension and compression). Only two
loads for each anchor are given in Figure 18 (b), but they are sufficient to show that the development
and backward movement of the peak τs for the tension anchor are similar to other tension anchors.
As might be expected, the development of the τs distribution for the compression anchor is the mirror
image of the tension anchor, in that the peak τs increases and moves forward.

Also significant is the fact that for virtually the same magnitudes of applied loads, the strain in
the compression anchor are much lower than those in the tension anchor. This indicates that the
stiffness of the grout body in compression is greater than the tendon steel alone in tension.

a) Load distribution in tension and b) Comparative development of surface shear stress

compression anchors
Figure 18: Comparison between the behaviour of tension and compression anchors (from Mastrantuono and
Tomiolo (1978))

Kim (2003) carried out comparative tests on tension and compression anchors in weathered
soils. Results from these tests reproduced in Figure 19 and Figure 20 show general agreement with
those presented in Figure 18.

20
Figure 19: Tension anchor (from Kim (2003))

21
 Figure 20: Compression anchor (from Kim (2003))

Shear-induced dilation

Since narrow cylindrical anchors were first constructed in soil in the late 1950’s, it has been
recognised that the average shear stress developed on them was greater than could be derived from
overburden pressure alone. This is because of the repression of dilation, causing high normal stress.
The mechanism is that when the anchor is moved longitudinally, the greater shear strain occurs
adjacent to the anchor, hence dilation begins there first. The dilation is prevented, to some extent, by
the surrounding soil, which develops a radial pressure, normal to the surface of the anchor. This
enables high shear stress to develop.

Figure 21 from Wernick (1978) shows the relationship between normal and shear stresses
measured on the surface of a cylindrical anchor by means of in-built load cells. The anchor is oriented
vertically, in a round test bin 2.5 m in diameter by 3 m high. Stage 1 to 2 on Figure 21 is the filling of
sand round the anchor. The normal stress on the anchor increases, and because the anchor is drawn
down as further sand is placed above the level of the load cell, negative shear stress is developed.
Stage 2 to 23 is the stress path that occurs when load is applied to the end of the anchor, and it is
moved in a strain controlled manner. Peak surface shear stress occurs at stage 17, falling to the
residual at stage 23. The anchor is unloaded from stage 23 to 27, and the shear stress falls to zero

22
and the normal stress returns to the at-rest value. Because only the at-rest pressure existed before
the anchor was moved, the eight-fold increase in radial stress on the anchor during the rise of the
shear stress to its peak could only have been caused by restrain of dilation.

Figure 21: Radial (σr) and shear stresses (τrz) measured on the surface of an anchor during pullout (from Wernick
(1978))

Figure 22 shows the results of tests carried out at Wits (Luker). An anchor 600 mm long was
moved in a dense sand, and the surface shear stress measured on three separate parts of its length.
The radial stress was also measured at four points, all on the middle length of the anchor. Again it
can clearly be seen that the radial stress increases dramatically with movement of the anchor,
inferring that restraint of dilation was present.

23
 Figure 22: Mean radial stress at surface of mid length (Luker)

Creep movement

Because anchors are usually expected to work by maintaining their working load after
prestressing, if movement of the tendon, grout, or surrounding soil occurs after the anchor has been
prestressed and locked-off, then the force exerted by the head of the anchor will decrease. Relaxation
of the steel used for the tendons can cause up to 8% loss of prestress after 100 hours (42 days), but
5.6% loss for normal relaxation strand and 1.1% for low relaxation strand are more usual values in
practice.

Time dependent behaviour of soil is most commonly associated with clay soils, but creep also
occurs on anchors in cohesionless soils. Figure 23 from Ostermayer and Scheele (1978) shows how
shear stress is redistributed when a tension anchor in a gravelly sand is held at constant load. The
magnitude of peak τs generally decreases and moves backwards, whilst the area under the curve
remains essentially constant. This is different from the behaviour under load increase, when the peak
moves backwards and generally rises at the same time.

24
 Figure 23: Change in τs distribution with time for constant load on an anchor in sand (from Ostermayer and
Scheele (1978))

If the load on an anchor is quickly increased, and then held constant, the anchor head will
continue to move, but providing the load is below the failure load, the rate of movement will decrease
with time. If the load increase brings the anchor to its failure load, then although it may appear to carry
load, the rate of movement does not decrease with time.

Below the failure load, a graph of movement against log time is a straight line, as shown in
Figure 24. The slope of this line is called the creep coefficient, α. If α is plotted against load, as shown
in Figure 25, then the graph line is initially straight, but suddenly rises as the failure load is
approached. The load at which the graph suddenly turns is called the critical load. An assessment of
the proximity of the working load to the critical load is used as a criterion to judge the acceptability of
an anchor. The method of assessment varies between different codes.

Creep of the anchor head caused by the tendon and the grout amounts to an α value of
approximately 0.4. Values above this are caused by the soil.

25
 ∆2 − ∆1
𝛼𝛼 = 𝑡𝑡
log( 2 )
𝑡𝑡1

Figure 24: Method of determining creep coefficient (from Hanna (1982))

α

Load

Figure 25: Variation of creep coefficient with anchor load

26
Examples

Example 1

Determine the ultimate pullout load of a 1 m high vertical anchor plate placed at a depth of 5 m
in a dry cohesionless soil (φ’ = 25° and γ = 16 kN/m3).

Example 2

Determine the ultimate pullout load of a 0.25 m wide horizontal anchor plate placed at a depth
of 0.75 m in a dry frictional soil (φ’ = 33° and γ = 18 kN/m3).

Example 3

a) What are the mechanics of behaviour of a horizontal anchor and its surrounding soil when it
is subjected to a vertical upward pull?
b) What major problem exists in the determination of a mathematical model of this behaviour for
design purposes?

Example 4

A circular horizontal anchor (Diameter = 0.75 m, Weight = 1.1 kN) is placed at a depth of 1.25 m
in dense sand (φ’ = 37° and γ = 19 kN/m3). By considering the failure cone determine the maximum
pullout capacity of the anchor.

Example 5

A vertical anchor plate is required to carry an anchor load of 315 kN/m as indicated in the
diagram below. Determine the required anchor plate height if a clean sand is hydraulically placed as
backfill. State all assumptions made.

Figure 26: Diagram for Example 5

Example 6

By comparing the pullout capacity of horizontal anchor calculated by considering the moving
cone of soil to that obtained with the Merifield and Sloan (2006) method determine whether it is more
appropriate to use K0 or Kp. State all assumptions made.

27
Example 7

Figure 27 below shows an anchor in a dense sand.

a) The anchor is loaded to its maximum capacity, when the force distribution along the
anchor is as given in Figure 28. Calculate and draw the surface shear stress distribution
along the anchor.
b) Describe with the aid of sketches the likely changes in the shear stress distribution as
the load on the anchor increased from zero to the maximum, giving reasons for them.
c) If the anchor load is held constant at its maximum capacity and some creep movement
occurs at the head, discuss the likely shear stress distribution immediately after the
anchor is unloaded to 5% of the maximum capacity.

Figure 27: Site cross section

Figure 28: Force distribution along anchor

Example 8

a) Describe, with the aid of diagrams, the mechanism of the behaviour of a long cylindrical
compression anchor in stiff soil under the following conditions:
i. Load applied in four equal steps up to the ultimate load (Pult).
ii. Load of 0.5Pult is applied, and held constant, for a long time.
iii. A load of 0.5Pult applied, kept constant for a long time, then reduced to 0.01Pult.
b) The amount by which the estimated value of an anchor’s ‘critical load’ exceeds the
required working load is used in some codes of practice to judge whether an anchor (as
installed) is acceptable.

Explain with the aid of diagrams what is meant by the ‘critical load’.

28
References and further reading

These notes have been developed from the various texts listed below, the author admits some
sections are copied verbatim and therefore does not pass this work of as his own.

BARLEY, A. 1997. ANCHORAGE BEHAVIOR OF SHAFT ANCHORS IN ALLUVIAL SOILS.

DISCUSSION AND CLOSURE. Journal of Geotechnical and Geoenvironmental Engineering,
123.

HANNA, T. H. 1982. Foundation in tension, Germany, Trans Tech Publications.

KIM, N.-K. 2003. Performance of tension and compression anchors in weathered soil. Journal of
Geotechnical and Geoenvironmental Engineering, 129, 1138-1150.

LITTLEJOHN, G. S. 1980. Design estimation of the ultimate load-holding capacity of ground anchors.
Ground Engineering, 13, 25-39.

MASTRANTUONO, C. & TOMIOLO, A. 1978. First application of a totally protected anchorage. Revue
Française de Géotechnique, 107-112.

MERIFIELD, R. & SLOAN, S. 2006. The ultimate pullout capacity of anchors in frictional soils.
Canadian Geotechnical Journal, 43, 852-868.

MERIFIELD, R., SLOAN, S. & YU, H. 2001. Stability of plate anchors in undrained clay.
Geotechnique, 51, 141-153.

MEYERHOF, G. & ADAMS, J. 1968. The ultimate uplift capacity of foundations. Canadian
geotechnical journal, 5, 225-244.

OSTERMAYER, H. & SCHEELE, F. 1978. Research on ground anchors in non-cohesive soils. Revue
Française de Géotechnique, 92-97.

SHIELDS, D., SCHNABEL JR, H. & WEATHERBY, D. 1978. Load transfer in pressure injected
anchors. Journal of Geotechnical and Geoenvironmental Engineering, 104.

TURNER, M. 2012. Geotechnical design of ground anchors. In: BURLAND, J. B., CHAPMAN, T.,
SKINNER, H. & BROWN, M. (eds.) ICE manual of geotechnical engineering. London: ICE
Publishing.

WERNICK, E. 1978. Stresses and strains on the surface of anchors. Revue Française de
Géotechnique, 113-119.

29