Journal of Constructional Steel Research 60 (2004) 401–410

www.elsevier.com/locate/jcsr

Analysis of reduced modulus action in

U-section steel sheet piles
M.P. Byfield , R.W. Mawer
Department of Civil and Environmental Engineering, Southampton University, Southampton,
SO17 1BJ, UK

Abstract

U-section steel sheet piles are commonly used for constructing retaining walls in marine
environments and have been in widespread use throughout the world for most of the 20th
century. Relatively recently, concern has been raised about the bending strength of this pile
section, because U-section piles are connected together by interlocking joints located along
the pile wall centreline. As the piles resist bending moments, inter-pile movement can signifi-
cantly increase bending stresses. When this occurs, the wall is said to have exhibited reduced
modulus action (RMA), reducing the bending strength and stiffness below the fully com-
posite values normally assumed during design. In view of this concern, the recently intro-
duced Eurocode 3: Part 5 has introduced strength reduction factors to account for the effect
of RMA during design. The work presented herein has been carried out in order to provide
more information as to the values that should be selected for these reduction factors. The
work is based on the experimental testing of one-eighth-scale miniature piles.
# 2003 Elsevier Ltd. All rights reserved.

Keywords: Codes of practice and standards; Piles and piling; Reduced modulus action; Retaining walls;
Steel structures; Substructures; U-section piling

1. Introduction

Steel sheet pile walls are used extensively around the world. The two main sec-
tions used in the industry are the Larssen or ‘U-section’ and Frodingham or ‘Z-sec-
tion’ piles. Both types are connected together by interlocks running the length of
the pile, which allow the sections to be slotted together to form continuous walls.
The introduction of Eurocode 3: Part 5 by CEN [1] has raised concern about the


Corresponding author. Tel.: +44-23-805-946-51; fax: +44-23-806-775-519.
E-mail address: mpb@soton.ac.uk (M.P. Byfield).

0143-974X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0143-974X(03)00119-6
402 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410

occurrence of slippage along the interlock of U-section piles. If the sections are
allowed to move relative to one another, the bending strength of the pair will fall.
This is called reduced modulus action (RMA) and can reduce the elastic bending
strength by 70% of the strength of the composite pair.
The earliest work referenced concerning the reduced bending strength caused by
RMA in U-section steel sheet piles was by Lohmeyer in 1934 [2]. The paper pro-
poses a mathematical model that attempts to quantify the effect that the incom-
plete transfer of shear caused by inter-pile movement has on bending strength. The
theory and calculations are influenced by similar work carried out around the same
time by Blum in 1931 [3]. Lohmeyer’s model applies conventional elastic long beam
theory to U-section piles and considers two boundary cases for shear transfer.
These are the full shear transfer condition (Fig. 1A) and zero shear transfer (Fig.
1B). This approach recognised that in reality interlock friction will result in the
partial transfer of shear in the interlocks, producing a stress distribution lying
within the region bounded by the full and zero shear transfer conditions (Fig. 1C).
Lohmeyer’s analysis provides a reasonable approach to quantifying how the trans-
fer of shear force to adjacent piles is proportional to their strength, but fails to
determine if RMA is a regular feature of U-section pile walls and the conditions
under which it may occur are left undefined. Whilst the model does not attempt to
predict the conditions in which RMA is likely to occur, it is observed that the most
likely cause of failure is insufficient driving. Lohmeyer’s method was reintroduced
more recently by Williams and Little [4], whose field observations of bending stres-
ses in full-scale piles confirmed the presence of RMA. Further work in this field
was carried out by Schillings and Boeraeve [5], Wolffersdorf [6], and most notably

Fig. 1. Cases of shear transfer in pile sections.

 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410 403

by the Delft University of Technology in the Netherlands, where a large-scale field

test was carried out by Hartmann-Linden et al. [7].
The British Standard BS8002 [8] defines a range of soil conditions that are likely
to lead to RMA. In situations defined as prone to RMA, engineers are advised to
drive the piles in ‘‘crimped pairs’’. Crimping involves the pressing together of the
common pair interlocks to preventing inter-pile movement. Structural testing of
crimped U-section piles by Hartmann-Linden has identified a phenomenon known
as oblique bending. This occurs due to the asymmetric shape of a crimped pair of
piles and can lead to a 24% loss of bending strength. These results have led to an
inclusion in the Eurocode of an allowance for oblique bending which downgrades
bending strength. Recent research by Crawford and Byfield [9] provides a method
for predicting the bending stresses in crimped pairs of U-piles, and shows that the
effects of oblique bending can be largely ignored where piles are restrained against
sideways movement.
It is often easier to install piles in singles, rather than crimped pairs. For this and
other reasons, it has been common practice in many countries including in the UK
to ignore RMA during design. The anecdotal evidence from this widespread prac-
tice is that problems from RMA are rare and isolated to a well-defined set of cir-
cumstances, such as those that are clearly defined in BS8002 [8].
However, this anecdotal evidence from practice runs contrary to evidence from
the laboratory. Experimental tests carried out at the University of Liège in Belgium
[5] clearly showed that interlock friction has only a minor effect on bending
strength. These tests were carried out in three-point bending and the specimens
were fabricated from steel plate or section onto which the clutches from piles were
welded, thus forming miniature pile sections, see Fig. 2. These piles were tested
elastically to investigate the effect that interlocks filled with sand had on bending
strength. The tests demonstrated that the frictional effect of sand raised the bend-
ing stiffness of the piles by between 2% and 14%, depending on the density of the
sand pressed into the interlocks. However, it can be argued that these tests provide
only a partial model of the behaviour of restrained U-pile walls because:

(a) The specimens were smaller than conventional U-section piles, being only 1 m
long, whereas in practice piles are typically 15 m in length. The depth of the
test piles was also significantly less than found in practice. The combination of
short length and shallow depth produced relative displacements between the

Fig. 2. Sketch of tests carried out to investigate RMA at University of Liège [5].
404 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410

Fig. 3. Realistic loading arrangement for permanent steel sheet pile wall.

piles of less than 1 mm. This is in contrast to the several centimetres of dis-
placement that a full-scale pile wall should be expected to develop during the
development of RMA. The small relative displacements may have been insuf-
ficient to generate the full frictional force in the sand introduced into the full-
scale interlocks tested.
(b) The loading arrangement used three-point bending (Fig. 2). This is significantly
different to the loading conditions typical in practical retaining walls, in which
a complicated arrangement of active and passive pressures (Fig. 3) creates two
or three points of contra-flexure. Furthermore, the position of maximum shear
force is typically located away from the ends of the piles, whereas the
maximum shear forces from three-point bending are adjacent to the free ends
of the specimens. Therefore, three-point bending may present a situation more
conducive to RMA than that found in practical pile walls.
(c) Practical restrained steel sheet piles usually have a capping beam at the head of
the piles. Capping beams are commonly formed from reinforced concrete and
will in effect prevent inter-pile movement at the pile head. The Liège tests did
not assess if the resulting restraint against inter-pile movement has a significant
influence on strength.
(d) The effect of friction between the surface of the piles and the soil is not asses-
sed, but is likely to influence the development of RMA.
(e) The effect of corrosion at the interlocks on the transmission of shear between
the piles is not assessed.

Fig. 4. Dimensions of test specimens.

 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410 405

Table 1
Section properties of test specimens shown in Fig. 4
Single pile Non-composite pair Composite pair of
of piles piles
Area (mm2) 341.57 683.15 683.15
Second moment of area (mm4) 41,390 82,780 339,441
Elastic section modulus (mm3) 1807 3614.3 12,011.7

Because of these factors, it is likely that the Liège tests provide a lower-bound
estimate of the potential that interlock friction has for moderating the effect of
RMA. Despite this potential conservatism, the results have been used as the basis
of substantial strength and stiffness reduction factors applied to U-pile design for
the recently introduced Eurocode 3: Part 5. The resulting factors lead to a
reduction in bending strength of up to 45% and a reduction in stiffness of up to
65%. Clearly, this downgrading in performance has implications on the economic
viability of U-section steel sheet piles.
This paper presents results from the testing of one-eighth-scale piles extruded
from aluminium. The scale pile tests simulate the complex bending effects found in
practical pile walls and thus attempt to discover if some of the factors listed (a)–(e)
significantly effect the bending strength of U-section piles.

2. Experimental testing

U-section piles are amongst the largest hot rolled sections available. A single pile
is typically 600 mm wide and lengths range up to 30 m. Given the resulting diffi-
culties associated with full-scale testing, the tests reported herein were carried out
using scale model piles. The cross-section of the test specimens is sketched in Fig. 4
and the geometric properties are defined in Table 1. The test piles were designed to
interlock to form a box section so that no interlocks are left unconnected. Free
interlocks were avoided because local buckling of the free interlocks would have
initiated premature failure in the piles.

Fig. 5. The loading arrangement used during the experimental testing.

406 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410

Fig. 6. Dimensions of loading arrangement.

The general arrangement of the test set-up is sketched in Fig. 5, with the exact
positions of the loading points shown in Fig. 6. Load was applied to a primary
loading beam that was supported by two secondary beams, both of which had
PVC reaction points that loaded a total of four final loading beams. PTFE bear-
ings between the four final loading beams and the test specimens were used to
achieve low friction bearings. This loading system produced a triangular distri-
bution of loads similar to that found in an actual steel sheet pile retaining wall, see
Fig. 7. The corresponding distribution of bending moments for a total load of 1
kN is sketched in Fig. 8.
A photograph of the test set-up is shown in Fig. 9, in which a total of four verti-
cal lateral restraints can be seen. These stabilise the loading arrangement without
significantly introducing friction sites, because PTFE strips cover the interface

Fig. 7. Load distribution per 1 kN of total load applied centrally.

 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410 407

Fig. 8. Distribution of bending moments (for 1 kN of applied load).

between the supports and the loading beams and test specimens. In total, five sets
of tests are reported. These include:

Test 1—Interlocks greased.

Test 2—Plain sections.
Test 3—Interlocks filled with ungraded sand.
Test 4—Interlocks filled with ungraded sand and riveted at the head of the piles
(a total of six 4 mm diameter pop rivets passing through 4.1 mm holes drilled
through the interlocks to prevent slippage and simulate the effect of a capping
beam).

Fig. 9. View of test set-up.

408 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410

Fig. 10. Load vs. deflection.

Test 5—Interlocks riveted (3.2 mm diameter rivets passing through 3.2 mm holts
drilled through interlocks on both sides at 100 mm centres).

3. Results and discussion

The load vs. deflection responses recorded during the tests are shown in Fig. 10.
Furthermore, the ultimate bending strengths and stiffnesses (recorded during the
initial linear elastic region) are listed in Table 2. A brief inspection of these results
shows that a significant improvement in stiffness has been achieved by the addition
of sand to the interlocks. In addition, RMA did not significantly affect ultimate
bending strength.
Test 1 (greased interlocks) provides the benchmark test. Comparison between
the expected and observed deflections showed that the grease was successful in
removing interlock friction, thus allowing the full effects of RMA to develop. Test
2 was carried out on plain (ungreased) aluminium piles. Whilst the piles were free
to slide along the interlocks, friction was responsible for a 16% increase in stiffness

Table 2
Test results
Load/displacement (N/mm) Ultimate load capacity (kN)
Test 1: Greased interlocks 137.68 15.33
Test 2: Plain interlocks 159.92 17.40
Test 3: Sand filled interlocks 164.61 19.87
Test 4: Sand and rivets at pile head 220.04 15.54
Test 5: Riveted interlocks 386.12 16.35
 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410 409

(in comparison with Test 1). Tests 3 and 4 were carried out with sand introduced
into the interlocks. Test 3 showed that the sand increased the bending stiffness by
20%. Test 4 was carried out with the interlocks riveted at the head of the piles to
simulate the effect of a capping beam and stiffness was observed to be 60% higher
than that for Test 1.
The original tests by Schillings and Boeraeve [5], carried out to assess the effects
of interlock friction on the development of RMA, showed that sand introduced
into the interlocks is capable of increasing the bending stiffness by between 2% and
12%. Tests 2–4 may have demonstrated higher stiffnesses because of the triangular
distribution of load used during the tests, which more accurately models the load-
ing commonly found in practice, in which the active and passive pressure distribu-
tions may in effect help to ‘‘clamp’’ the base of the piles (Fig. 3). This clamping
effect will raise the reactions between the piles and therefore generate relatively
high friction forces. This conclusion is justified from observations made during the
tests. No interlock slippage was observed at the toe of the piles during Test 2,
although significant inter-pile movement was observed at the pile head. Further-
more, zero movement in the interlocks was observed at either the head or toe of
the specimen during Test 4. Thus, the tests presented herein provide an explanation
as to why the tests carried out in support of the development of Eurocode 3: Part 5
showed only a marginal impact on RMA from interlock friction.
As would be expected, Test 1 (greased sections) showed the lowest ultimate
bending strength of all the tests. Test 2 (plain sections) showed a 13.5% increase in
ultimate bending strength. Test 3 (sand filled interlocks) produced the highest ulti-
mate bending strength, showing an increase of 30% over the greased sections.
However, both Tests 4 and 5 produced bending strengths similar to that for Test 1.
The sections tested were relatively slender in cross-sections and failed due to local
buckling of either the flange or the web. U-section piles are generally class 1 or 2 in
classification. It is therefore not possible to make direct conclusions concerning the
effect of RMA on the plastic moment of resistance. However, RMA should have
reduced the bending strength of the test piles because the ratio between the elastic
bending strength of the pair of piles acting compositely vs. the non-composite elas-
tic strength of the pair is 3.3 to 1 (see Table 1). Therefore, RMA should theoreti-
cally have led to significant reduction in strength. The absence of any significant
strength reduction may help to explain why RMA has a clear impact in theory, but
has rarely been observed in practice.

4. Conclusions

This paper presents data from scaled tests on U-section piles. These are used to
assess the effect that different interlock conditions have on the strength and stiffness
of U-section piles. The testing builds on similar tests [5] carried out in support of
the development of Eurocode 3: Part 5, in which the presence of sand in the inter-
locks was shown to increase stiffness by between 2% and 12%. The present work
provides evidence that the previous research may have provided a conservative esti-
410 M.P. Byfield, R.W. Mawer / Journal of Constructional Steel Research 60 (2004) 401–410

mate of the stiffening effect of sand in the interlocks. The test results presented
herein reveal that there is a great advantage in preventing slippage at the head of a
pile wall, for example by a reinforced concrete capping beam. Tests demonstrate
that considerable interlock friction is developed at the toe of the pile by the clamp-
ing effect of active and passive pressures simulated during the experiments. This
effectively prevents relative movement at both ends of the piles and leads to a 60%
increase in stiffness in comparison with greased sections.

References
[1] CEN Eurocode 3: Design of steel structures, Part 5: Piling. BSI, London, UK, 1996.
[2] Lohmeyer E. Discussion to ‘Analysis of sheet pile bulkheads’ by P. Baumann. Proc Am Soc Civ Eng
1934;61(3):347–55.
[3] Blum. Einspannungsverhaeltnisse bei Bohlwerken. Berlin: Wilhelm Ernst & Sohn; 1931.
[4] Williams SGO, Little JA. Structural behaviour of steel piles interlocked at the centre of gravity of
the combined section. Proc Inst Civ Eng Struct Bldg 1992;94(2):229–38.
[5] Schillings R, Boeraeve P. Design rules for steel sheet piles—ECSC Project 7210-SA 127/523/840.
CRIF Department of Steel Construction, Liège, Belgium, 1996.
[6] von Wolffersdorf PA. Verformungsprongnosen fur Stutzkonstruktionen. Germany: Universität Fri-
dericiana in Kalrsruhe; 1997.
[7] Hartmann-Linden R, et al. ECSC-Project 7210-SA 127/523/840: final report. Development of rules
for steel sheet piles and introduction into Eurocode 3: Part 5. RWTH Aachen, Germany, 1997.
[8] BS8002: Earth retaining structures, BSI, London, 1994.
[9] Crawford RJ, Byfield MP. A numerical model for predicting the bending strength of Larssen steel
sheet piles. Proceedings of the Conference on Structural Engineering, Mechanics and Computation
(SEMC), Capetown. 2001, p. 393–400.