BREAKWATER REPAIR:
MONITORING, MODELLING, AND CONSTRUCTION
S Pillay1, D Phelp2 and A Bartels3
Abstract
The trunk section of the south breakwater, at the entrance to the Port of Richards
Bay, has suffered some damage since its construction which was completed in 1976.
This paper briefly discusses the changes in the design conditions as a result of sand-trap
dredging off the breakwater, the annual photographic monitoring results showing the
excessive rates ofdamage, and the model testing ofvarious repair options, using 20t and
30t dolosse. Both 2Dflume and 3D basin model tests were carried out at various scales,
with fixed and movable bed models. Finally the construction of the optimal repair,
carried out in 1997/98, using a heavy duty mobile crane (with 48m boom reach) and
DGPSpositioning, is described.
Introduction
The Port of Richards Bay, on the east coast of South Africa, has two dolos
breakwaters, a shorter straight breakwater on the northern side of the harbour entrance
channel and a longer curved breakwater on the southern side. The south breakwater
consists of an "S" shaped rubble mound structure (Figure 1), constructed between 1973
and 1976, which stretches for approximately 1 km, almost perpendicular to the coastline.
The original armouring on this breakwater consists mainly of 20t dolosse on both sides
of the trunk (Figure 2), but includes 30t dolosse on the roundhead. The south breakwater
forms the main protection of the Richards Bay harbour entrance channel, against
dominant southerly storms and the nett littoral drift, which is from south to north.
Annual photographic monitoring of the south breakwater has shown a gradual
increase in damage to localised areas on the southern side of the trunk, despite spot
repairs using 20t dolosse, carried out by Portnet in 1985. A detailed evaluation of the
' Portnet, Port of Richards Bay, P O Box 464, Richards Bay 3600 (SA)
2
3
Environmentek, CSIR, P O Box 320, Stellenbosch 7600 South Africa
Entech Consultants (Pty) Ltd, P O Box 752, Stellenbosch 7600 (SA)
1775
1776
COASTAL ENGINEERING 1998
damage was undertaken by Zwamborn in 1988 (CSIR, 1988) which lead Portnet to
commission the CSIR in 1991 to carry out investigatory model tests in an existing 3D
1:100 scale model of the entrance to the Port of Richards Bay. These tests were to check
different repair options, taking into account the position of a dredged sand-trap along the
seaward side of the breakwater. The original idea was to use 20t and 30t dolosse,
available from a stockpile of spare dolosse at the root of the south breakwater.
Figure 1: Aerial View of South Breakwater
Due to delays in the commissioning of a suitable crane, and the use of most of the
spare dolosse on repairs to the north breakwater head (Phelp, 1996), the construction of
the repairs to the south breakwater were delayed until 1997/98. This also required the
casting of 1000 additional 30t dolosse. Between 1991 and 1996, additional 2D flume
tests were carried out at a 1:40 scale to check the stability of the rock toe of the repair
slope. These tests were carried out with a movable (sand) bed to model the effects of toe
scour. Before the commencement of the repair work it was found that, due to a gap in
the dredging programme, there was a buildup of sand along the breakwater toe. Final
model tests were therefor carried out to re-check the toe stability at this shallower depth.
Monitoring Results
The crane/helicopter photographic survey method (Phelp, 1994) has been used to
annually monitor the breakwater since 1979 on an ad-hoc basis, but regularly since 1987,
and crane and ball profiling surveys have been carried out since 1981. The photographic
survey stations are spaced at 25m, and the ball profiles at 5m intervals. Figure 2 shows
the position of the survey stations and Figure 3 the rate of damage increase since 1987.
COASTAL ENGINEERING 1998
1777
The original breakwater construction used a total of 13 400 20t dolosse on the trunk, and
2 200 30t dolosse on the roundhead, which amounts to approximately 10 dolosse per
metre of breakwater. The original depth at the head was -18m and -14m along the outer
curve of the trunk. The worst damage prior to the repair, located at station C8, was 17%
dolosse displaced or broken (Figure 2).
Ad-hoc spot repair work was carried out in this area in 1985, when 52 new 20t dolossse
were placed. Although this showed a significant improvement in the measured profiles,
the photo survey showed that half of these dolosse were broken and/or lost by 1987. This
type of spot repair, which was not model tested, with non-reinforced dolosse in a single
layer not interlocking with the surrounding dolosse, proved to be ineffective.
JjXjSTING PEfiCEMTApE DAMAGE
, "J*l "'S* '7.3* ' 9.,% I
-~1
SS
'
'
'
u%
-t-'i
REPAIfi
DOLOSSE
'
]3%
'
t--|
'
, 6.7% W*
iSMMM
ORIGINAL SEA
THfiqUGH
MAIN
BREAKWATER
L'uuA'/ / / /
M
v8,0m
IIPSAL JRPSS.-_SCT|0N
'
/ .
S
\
\ \//
V
^^.
/
.'
BED -13500
TRUNK
Figure 2: Plan View of South Breakwater and Cross-Section through trunk.
Factors Contributing to the High Damage
The occurrence of low pressure cyclonic storms (cyclones Demoina and Emboa
in 1984) subjected the breakwater to wave heights exceeding the 7,9m 1:50 year design
Hmo. Storm wave set-up and low atmospheric pressure associated with these storms also
had the effect of raising the water level, thereby raising the depth limited wave heights
reaching the breakwater. There have also been a number of lesser, but still powerful
storms with wave heights in excess of 6m - the latest being experienced in 1990. The
rates of damage are shown in Figure 3 for the worst stations on the trunk section.
A nett littoral drift of up to 800 000m3 northwards has necessitated the dredging
of a sand-trap against the outside of the south breakwater. Bathymetric surveys carried
out regularly by Portnet in the sand-trap area have shown that dredging has taken place
1778
COASTAL ENGINEERING 1998
much closer to the breakwater than originally anticipated (up to 60m closer). The depth
of the sand-trap has reached -24m and the side slope as steep as 1:4,7. Figure 4 shows
the dredger and Figure 5 the average position and sections through the sand-trap.
C7P1
C8P1
C9P1
C10P1
C11P1
C12P1
12 ^ 24
36 B 48 m 60 = 72 84 ^ 96~ = 108
Months from Jan 1987
Figure 3: Rates of Damage to Worst Stations since 1987
Besides the deeper trap and steep side slopes, there has also been scour at the toe
of the original breakwater (seismic surveys carried out by CSIR in 1993 confirmed the
toe erosion (CSIR,1994)). The damage profile along the breakwater (shown in Figure 2)
matched the plan of the sand-trap, with the highest damage area aligning with the deepest
parts of the trap. The 3D basin model tests, which modelled various trap layouts,
confirmed that the sand-trap allowed higher depth limited waves to reach the breakwater.
Figure 4 : Portnet Dredger Trail Dredging in the Sand-trap
COASTAL ENGINEERING 1998
1779
Figure 5: Plan and Cross-Sections showing Average Sand-trap Position.
Constraints to the Repair Design
Model tests were carried out in an existing 3D model of Richards Bay, to save
both time and costs. This original model was built at a scale of 1:100 and covered the
harbour entrance and part of the inner channel. The model test options were also
originally restricted to using available 20t and 30t dolosse from a stockpile near the
southern breakwater. A number of repair options were tested using either the 201: or 30t
dolosse, with different repair slopes between 1:1,5 and 1:2,5, both with and without a
rock toe.
The removal of rubble and pre-repair slope preparation was limited due to poor
underwater visibility and rough sea conditions normally experienced along the outer
breakwater. Contour plots of the outer slope, drawn from crane and ball profile surveys,
were used to locate damage cusps below water and guide the filling of these holes at the
toe of the armour slope. A double layer repair slope was then designed to cover the worst
damaged areas. The top and sides of the repair are then tied into the original breakwater
slope by tapering the repair. The width of the repair was limited to 40m from the splash
wall, which was the limit for the boom of the crane lifting a 30t dolos. This mobile crane
(Figure 10) was specially designed to fit onto the 6,7m wide mass-concrete capping.
1780
COASTAL ENGINEERING 1998
Choice of Model Scale
The scale of 1:100 used for the tests gives a Reynolds number of approximately
1 x 104, which is just within the minimum range recommended by Van der Meer (Van der
Meer, 1988). Some scale effects were expected, but have been found to make the model
results more conservative. The scale effects, being similar for all the test runs, allowed
for comparisons to be made between the repair options tested. The calibration test
showed that the hind-cast of the damage which occurred in the cyclonic storms of 1984,
is in qualitative agreement with the observed prototype damage confirming the validity
of the physical model. Details of the chosen repair section were confirmed at larger
scales (1:63) in a 3m wide flume with a fixed bed and at 1:40 in a 2m wide flume with
a movable bed. Figure 5 shows the section through the sand-trap and breakwater which
was modelled in the 2D flume tests.
Wave Generation and Measurement
Up to 14 standard wire resistance wave probes were used which were coupled so
that measurements could be carried out over prescribed areas. A hinged paddle wave
generator bank for the 3D tests was 30m long situated approximately 30m seawards of
the breakwater (representing 3km by 3km in prototype). Based on a review of existing
wave data and analysis of the cyclonic storm data as recorded by a waverider buoy off the
breakwater, the following main test conditions were chosen:
Wave direction, harbour entrance area (12s) 140
Storm input, Richards Bay Spectrum, y = 2,74 with the following 1.5 m steps
Hm0 = 2.5,4,5.5,7,8.5 m with wave periods Tp = 12 s to 16s. This is above the
design wave height of 7,9 m and cyclone wave recording of 8,4 m.
Water level MHWS = +2,0 m CD which resulted in the highest damage.
The above conditions were considered applicable to reproduce the damage on the
breakwater which has been subjected to a minimum of one 1:1 year storms (Hmo>5m) per
year, three 1:10 year storms (Hmo>6m) and one storm exceeding a 1:50 year storm
(Hmo>7.9m) during the lifetime of the breakwater.
Test Procedure
In order to calibrate the physical model, a calibration test was carried out in which
the prototype damage resulting from the storm history was reproduced in the 3D model
(Figure 6). A number of repair options were then investigated, starting with the simplest
option, and extending the repair until a stable solution was found. Before each repair
was constructed, the original damage was replicated in the model. The repaired
breakwater was than exposed to the conditions described above.
After each test, the repair was removed and the original damage reconstructed,
after which the next repair option was implemented. The optimum repair option chosen
from the 3D tests was then reproduced at a larger scale in the 2D flume tests, initially
with a fixed bed and then with a movable sand bed. These larger scale flume tests were
used to give special attention to the ability of the breakwater to withstand some toe scour
resulting from sand-trap dredging.
1781
COASTAL ENGINEERING 1998
Model Damage
Prototype Damage
10 11
12
STATION
Figure 6: Prototype versus Calibrated Model Damage
Measurement of Damage
Prototype damage is assessed by counting the broken or lost dolos units and
adding the units which have been displaced more than lA h (dolos height). A number of
swing tests were carried out on full-scale 9t dolosse to determine the degree of movement
these dolosse could sustain without breakage (Zwamborn and Phelp, 1989). Based on the
results of these tests it was recommended that all movements greater than half the height
of a dolos be included as damage. This damage is then expressed as a percentage of the
total number of dolos in a particular section of breakwater (Figures 2 and 3).
In the model, the number of dolos movements was determined by the digital
analysis of video images taken before and after each run. In addition, the number of
dolosse which were rocking was recorded visually and by cine camera during each test.
However due to the difficulty in observing movements over the whole test area, it was
decided to use the video measurement of small movements (< h), as an estimation of
rocking dolosse. This was then calibrated against the recorded prototype damage to give
a calibration factor of 0.4(<h) + (>h), which gave an accurate simulation of the prototype
damage profile along the trunk of the breakwater between stations 5 and 17 (Figure 6).
Figure 6 also shows, both in model and prototype results, that the worst damage occurred
between stations 7 and 9. Hydrographic surveys of the sand-trap between 1977 and 1991
have shown that, almost since completion of the breakwater, the deepest area of the sandtrap was located opposite stations 7 to 9 (Figure 5). This also coincided with the area
where the sides of the sand-trap were steepest and closest to the toe of the breakwater.
One model test which was carried out with a larger deeper sand-trap resulted in an
increase in damage proportional to the extension of the sand-trap, which indicated that
the increased breakwater damage could be linked to the sand-trap dredging.
Discussion on the Repair Strategy Followed
Static tests on dolosse have shown that a dolos can carry 4 to 6 times its own
weight without breaking; this implies that a number of layers can be constructed without
breakage under static load. Thus it was considered feasible to place a 1 to 2 layer thick
20t to 30t dolos strengthening layer, safely on top of the existing damaged 20t dolosse.
1782
COASTAL ENGINEERING 1998
Although the quality of the underlying 20t dolosse is questionable, the dynamic loading
over the past 20 years has caused the weaker dolosse to break, and careful placing of new
dolosse should not result in significant further breakages, besides the initial "shake down
damage". However, since most parts of the repair sections will consist of a number of
already broken units, the repair itself was designed as well interlocked armour, finished
to a uniform slope, which should be able to stand on its own. Although stresses cannot
be modelled, extensive prototype observations and structural tests on full size dolosse
support the above conclusions.
Repair Options Tested in 3D Basin
Comparative tests were first carried out using the same wave conditions and sandtrap configuration. Later tests included the option of extending and deepening the sandtrap. The first repair option tested involved covering only the worst stations (C7 and
C8) with a double layer of 30t dolosse, with 20t dolosse on either side to tie into the
existing slope. A total of 150 30t and 250 20t dolosse were used for repair option 1,
placed at 1:1,5 slope with no rock toe. After this proved unsuccessful, repair option 2
was tried, with 30t dolosse and a rock toe stretching from stations C5 to C12. A total of
504 dolosse were placed covering a distance of 165m. Although repair option 2 showed
less damage, it was still unacceptably high. Repair option 3 was similar to option 2 but
with a flatter 1:2,5 slope from +3,5m. A total of 670 30t dolosse were used. This repair
1:2,5 slope option was repeated unsuccessfully with 950 20t dolosse and then with 785
30t dolosse but without a rock berm.
Repair Options Tested in 2D Flume
Repair option 3 was then repeated in a larger 1:63 scale 2D flume (3m wide). The
effect of extending and deepening the sand-trap was also re-tested in the 2D flume. The
latter test confirmed the relationship between high damage and the deepest part of the
sand-trap (Figure 5). Because of the vulnerability of the breakwater toe to scour resulting
from sand-trap dredging, it was decided to optimise the size and position of the rock toe
by running some tests at an even larger 1:40 scale in a 2m movable bed flume. This was
done to check stability of the toe at low tide, and its ability to accommodate settlement
and erosion, but still maintain support for the bottom row of repair dolosse. The
movement of the rock and change in profile of the toe were carefully monitored. These
tests showed the need for the rock toe and first two rows of dolosse to be placed first, and
allowed to settle, before the rest of the repair dolosse were placed.
Change in design conditions
After acceptance of the above repair design, a total of 1000 new 30t dolosse were
cast near the root of the south breakwater (Figure 7). The start of construction of the
repairs was however delayed for more than a year because the Portnet crane that was to
be used for the repair at Richards Bay was unavailable. In this time there was substantial
accretion, (of up to 3m) especially along the trunk of the breakwater (Figure 8). The
breakwater repair section was again tested in the 2 D flume with the reduced depth. The
depth at the toe at some sections was as shallow as -5m CD. During the re-testing of the
COASTAL ENGINEERING 1998
1783
model it was found that the proposed rock toe was unstable, with the rocks being
displaced into the dolos slope. The dolosse would have sustained more damage and
become clogged (lower porosity) in such a scenario. Various options were then
investigated to solve this problem, such as dredging a trench in front of the breakwater
to lower the toe, or to use heavier rock (> 5t), or to do away with the rock altogether.
Figure 7: Casting Yard for New Dolosse
Figure 8: Final Repair Design Profile
Implementation of Model Test Results
A comparison of the damage at the end of each test run in the 3D basin showed
that option 3, with the 30t repair dolosse at a flatter 1:2,5 slope was clearly the best
option, although some "shake down" damage was expected from the repair settling into
the existing dolosse, from the pre-settlement of the first rows of repair dolosse and from
possible future toe scour. Research by the CSIR (Zwamborn and Phelp, 1989 and Luger,
1994) has shown that armour unit strength can be enhanced by rail reinforcing and by
increasing the size of the dolos fillet between the fluke and shank. For this reason, the
new dolos shape with large curved fillets was used and one third of the repair dolosse
were rail reinforced for use in potentially high damage areas.
1784
COASTAL ENGINEERING 1998
The solution that was eventually found to provide a stable repair at the shallower
toe depth was to replace the rock toe with an additional three rows of "sacrificial"dolosse.
There would ultimately thus be 5 rows of dolos lying on the accreted seabed. These dolos
were allowed to pre-settle into the sand over a length of time, before placing the rest of
the repair slope. In the model, the maximum settlement was recorded at 2-3m at the toe
(Figure 9). It was also found that the dolos had to be placed at a packing density of 0,75
for the maximum pre-settlement to occur. The rest of the breakwater repair was then
placed at a packing density of 0,85.
In reality the dolos would settle 2-3m or until they reached the previous rock toe,
or the remnants thereof. The pre-settlement dolosse placed directly onto the sand were
all to be rail reinforced for additional strength, and their settlement was monitored by ball
surveys. Figure 9 gives a typical profile before and after placing the first 5 rows of
dolosse, which shows the dolos 2m above the sand, indicating a settlement of about 2m.
The results of the crane and ball survey were analysed and contours plotted of the data.
These contours were then analysed and large holes were identified where additional
dolosse were placed to ensure a smoother profile before placement of the new double
layer repair. Each dolos was given a fixed co-ordinate (Figure 10), calculated to achieve
the desired packing density and final repair profile.
Figure 9: Pre-settlement of Toe Dolosse - Prototype and Model
Construction Methods
Based on the results of the model tests, only 30t dolosse were to be used for the
repair. These dolosse were brought from the casting yard (1001 new 30t units) and old
stockpile (37 old 20t and 88 old 30t units left over from the original breakwater
construction) on the south side of the entrance channel directly onto the breakwater.
Three double direction trailers were then used to transport the dolosse, but as these
trailers could only pass when unloaded, it meant that only one 3 Ot dolos could be brought
onto the breakwater at any one time (Figure 11). A portal crane was used to handle the
dolosse from the casting yard onto the stockplie and from there onto the trailers.
1785
COASTAL ENGINEERING 1998
Initial crane and ball surveys were
done with 5 m profile intervals over the
damaged areas. Repair dolos placing
grids were then calculated and the presettlement dolosse were placed. Another
ball survey was then carried out to check
the pre-settlement, from which the final
repair dolos placing grid could be recalculated if necessary. The smoother the
under-layer profile, the easier it was to set
placing grids for uniform packing density.
The crane hook was fitted with a
15m sling (to ensure the hook and pulley
remained out of the seawater), a quick
release hook and a double cable sling.
The double slings which support the
dolosse were hand spliced (instead of
swage joined) to allow easy removal of
the slings once the dolos was in position.
The quick release hook was hung from a
swivel and fitted with two torque bars,
which allowed easy rotation of the
dolosse to ensure good interlocking. The
torque bars were attached to 1 Omm nylon
(light and water resistant) ropes, which
were pulled perpendicularly from the
mass capping to orientate the dolosse.
The front row of toe dolosse were placed
with the vertical fluke facing seawards.
It was found that to ensure correct
packing density, the dolos placing must be
kept as close as possible to the grid
coordinates. The final orientation and
positioning of the dolos is then done by
eye to ensure good interlocking. Dolosse
are placed with a minimum of three
contact points to reduce the chance of
rocking under wave action. After all the
grid positions were full, it was found that
up to 5% additional dolosse (using old 20t
and 30t units from the stockpile) had to be
placed "in holes" to ensure a well
interlocked uniform profile. To identify
these "holes" an aerial view of the slope
was studied from a helicopter.
;s\*.
?'
O
00
a
'3
aj
CU
CO
O
O
a
9
-d
+J
!-<
3
o
C
UJ
CO
CO
O UJ
ri ">
O
co
Q o
u_ -i
oo
Q
M
"cd
uj u.
a,
a)
i-<
i>
[t,
fc
<, en
s>
ti
oo
1786
COASTAL ENGINEERING 1998
Figure 11: Portnet Crane Offloading Dolosse from Trailor
DGPS for Crane Positioning
For both the crane and ball surveys of the slope profiles, and the correct placing
of the dolosse, there was a need to accurately position the hook of the crane. A
differential GPS system was introduced using satellite positioning linked to a portable
computer onboard the crane. The satellite receiver is positioned on top of the crane
boom, directly above the position of the hook. The pre-determined positions are entered
into AutoCAD software on the computer, and standard survey software enters the real
time navigation parameters which indicate the position of the boom. By entering the
standing position of the crane along the breakwater, the boom reach and safety circle can
also be indicated on the screen. The crane operator can then immediately see which
dolosse can be placed from the present position of the crane. The AutoCAD dolos
placing grid is shown in Figure 10.
The positioning software includes the following useful features:
Zoom in and out, and centring the cranes position on the screen.
The entry of up to 20 predetermined crane standing positions on the breakwater,
including facility to orientate and offset.
The entry of up to 1000 top and bottom layer dolosse, including an indication of
size and numbering (colour options)
COASTAL ENGINEERING 1998
1787
The facility to import and do editing of an AutoCAD or other CAD drawing of
the breakwater eg: the "as-built" layout.
Indication and editing of the safe radius of the cranes reach.
The input and storage of the placed positions of the dolosse.
A backup system where the polar coordinates can be entered to position the crane,
should the DGPS signal fail.
The dolosse were numbered as per their sequence of placement. A top layer dolos
was always placed centred between 4 under-layer dolos. The sequence thus entailed the
placing of alternate bottom and top rows of dolos, moving up the slope. The placing of
the pre-settlement under-layer dolosse started at the root ofthe breakwater and progressed
towards the head. The crane then returned to the root to place the rest of the repair
dolosse after the lower dolosse had had a chance to settle. The end of the repair was
always left tapered at 45 up the slope, thus ensuring no unstable units which might be
displaced before the repair could be completed (during storm conditions or breaks in
construction). The limiting operating conditions forthe crane were wind speeds of 50kph
or swell heights above 2m.
- *> 1
^
Efc*:^*.v>g
Is
!
Figure 11: Successfully Completed South Breakwater Repair
1788
COASTAL ENGINEERING 1998
Conclusions
Close cooperation was maintained between Portnet, the Client/Port Authority who
also partook in the model tests and constructed the repairs, the CSIR Research Laboratory
which undertakes the annual breakwater surveys and who carried out the model studies,
and Entech Coastal Consulting Engineers who assisted with parts of the design and
construction. This ensured problems encountered could be quickly investigated and
amendments incorporated into the final design. The early warning provided by the
annual breakwater monitoring also meant that there was sufficient time to carry out the
model tests and come up with an optimum repair design.
The completed repair can be seen in Figure 12, which shows the uniform profile
and good integration with the original structure. A final ball survey of the repair showed
that for most profiles, there was a perfect match between the design and surveyed profile.
This was achieved by accurate dolos placing with the aid of DGPS on the crane. This
repair which was completed in June 1998, has already withstood two storms in excess of
the 1: lyr design wave height above 4m. A photographic aerial survey done after these
storms showed less than 1% damage resulting from the initial "shake down". Annual
surveys will be continued, to monitor the performance of the new shaped 30t dolosse.
Acknowledgements
The authors would like to thank those concerned for the team spirit which
prevailed between Portnet, the CSIR and Entech for the successful completion of this
project and for the contributions made to this paper.
References
CSIR (1988). Richards Bay South Breakwater, Evaluation of Damage and Proposals for
Repair. CSIR report EMAS-C 88114, Stellenbosch, South Africa.
Van der Meer J (1988). Rock slopes and gravel beaches under wave attack. Doctoral
thesis. Delft University of Technology.
Zwamborn J A and Phelp D (1989). Structural tests on dolosse. Seminar on stresses in
concrete armour units. Vicksburg, USA.
CSIR (1992). Port of Richards Bay: Model Study to Optimise Breakwater Repairs. CSIR
report EMAS-C 93049, Stellenbosch, South Africa.
Luger et al (1994). Increased Dolos Strength by Shape Modification. 24 ICCE 1994,
Kobe, Japan.
Phelp et al (1994). Results of Field Monitoring of Dolos Breakwaters. ICCE 1994, Kobe.
Phelp et al (1996). Richards Bay North Breakwater - Repair of a Roundhead. 25 ICCE
1996, Orlando, Florida.
