You are on page 1of 7

Where is our Weir going an Unusual Upgrade!

Amanda Ament, Jon Williams, Malcolm Barker


GHD Senior Structural Engineer, Dams GHD Manager, Dams GHD Principal Engineer, Dams Aplins Weir is located on the Ross River in Townsville, downstream from the Ross River Dam. Previous work had identified Aplins Weir as exhibiting factors of safety below 1.0 under normal operating conditions, with over 1000 persons at risk today in the event of failure. Originally constructed in the early 1920s, Aplins Weir has been upgraded and repaired following various failures on a number of occasions. The end result is a complex reinforced concrete and steel sheet pile composite structure reliant for stability on a number of unreliable components. This paper presents the historical data describing the current configuration of the weir, and the analyses required to evaluate the extisting structure, leading to the design of the proposed upgrade works. The final design involves a retrofit of large diameter cast-in-place lined piles and a heavily reinforced base overlay slab designed to completely bypass all existing vulnerable substructure elements. Keywords: Upgrade, Structural, Piling, Aplins Weir, Ross River.

Introduction
Aplins Weir is located on the Ross River in Townsville, downstream from the Ross River Dam. It is the last weir before the mouth of the river, and functioned both as water supply and tidal barrier for the young Townsville. Previous preliminary work had identified Aplins Weir as exhibiting Factors Of Safety (FOS) below 1.0 under normal operating conditions, with over 1000 persons at risk today in the event of failure. This makes Aplins Weir an extreme hazard dam, which, if it were to become referable, would require an upgrade to withstand the Probable Maximum Flood (PMF) event. Some images of the existing structure can be seen in Figures 1 and 2. GHD were engaged by Townsville Water to undertake the rehabilitation design in 2010.

is a complex reinforced concrete and steel sheet pile composite structure. A good deal of historical data mining was performed in order to gain a clear picture of the present structure, which is necessary if future performance is to be predicted with any reliability. Information was sourced from drawings, historical photographs, geotechnical investigations, site surveys and inspections.

Figure 2 View of weir and adjacent footbridge, including downstream rock armouring The Original Aplins Weir The original structure constructed in the 1920s consisted of a sheet pile cutoff wall beneath the upstream edge of a hollow reinforced concrete weir and apron. It was approximately 2.3m high from top of crest to apron, and extended approximately 110m across the Ross River. The sheet pile wall appears to consist of 380mm (15in) RSJ sections installed flange to flange and held together with clutches. There is also some indication of the presence of downstream cross elements within the sheet piling, however the frequency or presence of these is not clear. Recent upstream geotechnical investigations in 2004 show that there is a zone of very loose sands in the river bed, with Standard Penetration Test (SPT) N values of

Figure 1 View of central weir looking towards left abutment

Data Mining a Timeline History of Aplins Weir


The original Aplins Weir was constructed in the early 1920s, and it has been upgraded and repaired following various failures on a number of occasions. The end result 1

less than 4. These sands form the top layer over an unknown extent of the river bed, up to a depth of approximately 6m below current bed level. Considering the construction equipment available in the 1920s it can be expected that ground improvements would have been required in this sandy zone, in the form of additional dumped rock, to enable the equipment to access the river bed and operate. This expectation is supported by drawings of the weir cross section created in 1951, which show a significant depth of rockfill downstream of the sheet piles, along with a 2.0m deep concrete capping beam on the sheet piles. 1943 Upgrade In order to increase water storage capacity, the weir was upgraded in 1943. A new reinforced concrete buttressed structure was built, on the upstream side of the original weir. A new upstream sheet pile wall was installed for the full length of the first weir to form a cutoff. This sheet pile wall appears to be a flat web type with thumb and finger joints at either end. Beyond the extents of the original weir near the right abutment, a reinforced concrete diaphragm wall was installed in lieu of steel sheet piles. The 1943 drawings also indicate cast iron pipe piles raked in the downstream direction, connected to the upstream end of the base slab. Neither the length nor the raking angle of these piles is specified, however the drawings suggest an angle in the order of 45. The same drawings also indicate that the base of the piles was belled out using explosives, but neither the success of this procedure nor the extent of the belling is known. These piles are in fact not shown on some subsequent drawings. Site investigations were performed with the aim of verifying the presence or otherwise of these piles using ground penetrating radar, but the results proved inconclusive. The new buttressed weir appears to have been simply cast up to the face of the original weir, with the original weir subsequently remaining in place for some years. The nature of the connection between the two structures during this time is unclear.

buttressed bays were lost. A new right abutment was constructed, including a 300mm thick concrete apron of significant extent. The crest of the original weir was demolished at this time, and the new weir was structurally connected to the substructure of the original weir. The details of this connection are unknown site measurements of the area are not consistent with any of the drawings. A detail of one version of the drawings can be seen in Figure 3, which shows the original smaller weir and the buttressed upgrade on the upstream side. Floods, 1950 During the 1950s floods, large erosion holes formed downstream of the structure. A major one formed approximately one third of the way along the structure from the left abutment. It was over 2.1m deep, 5m wide and 7.5m long along the weir. The repair of this hole can be seen in Figure 4. A second erosion hole of slightly smaller size formed near the right abutment.

Figure 4 Repaired erosion hole in apron These holes were filled with rock armour and overlaid with concrete. Also at this time, large diameter rock armour was placed approximately 15m downstream of the structure, tied together with steel cables and bars embedded into the individual rocks. A capping beam along the weir crest was also installed for the full length of the weir. And More Floods, 1956 This time the left abutment was lost, along with several bays, perhaps up to the fifth bay of the central structure. The riverbed surface was scoured significantly. The left abutment was rebuilt, consisting of a new sheet pile wall and capping beam constructed upstream of the existing structure. Here the drawings for this phase differ from the photographic evidence an upstream clay embankment protected by rock pitching was specified on the drawings, but this appears not to have been constructed. Also, the downstream scour damage appears to have been rock filled and a concrete apron installed, however no concrete apron was shown on the drawings.

Figure 3 Excerpt from drawing L4152 dated April 1943. Note the upstream sheet piles, the raked cast iron pipe piles, the original downstream sheet piles and the original 1920s cross section Floods, 1946 The first record of flood damage to Aplins Weir occurred during 1946, when the right abutment and the first two

Description of Aplins Weir Today


As the previous Timeline indicates, the reality of the structure as it exists today is complex. Each component is discussed in more detail below.

Left Abutment The left abutment consists of a sheet pile wall with a 1.2m wide reinforced concrete capping beam, all of which was constructed in 1956. The upstream face of the sheet piling is largely exposed to the reservoir, indicating that the fill installed in 1956 has been lost, or was not placed to the full extent shown on the drawing. Significant and consistent corrosion of the sheet piles can be observed. Downstream of the sheet pile wall there is a concrete apron slab, which may or may not be reinforced. Drainage holes exist at a number of locations on the apron slab. A significant void exists between the fill and the underside of the apron slab, which was observed using photographs taken down the drainage holes. Downstream of the concrete apron slab a scour hole has developed in the river bank. Typical Section The typical cross section consists of an upstream and a downstream sheet pile wall, founded into clay, and cast into a 6.9m wide, 457mm thick reinforced concrete base slab. The upstream sheet pile wall consists of 12mm thick flat web sheet piles, and the downstream sheet pile wall of 380mm RSJs installed flange to flange and coupled together with clutches. The inclined reinforced concrete weir face slab varies in thickness from 203mm at the top to 381mm at the base. The base connection between the weir wall and base slab is a pinned key joint. The face slab is supported by buttresses at 3.0m centres in a simply supported fashion, and by the continuous capping beam along the top. The buttresses are approximately 300mm thick, with a 483mm wide support beam immediately behind the face slab. Drainage holes have been provided in the base slab, and interconnecting gravel drains beneath the base slab are shown on the drawings. Water levels in the drainage holes were observed to be at, or below, the level of the top of the slab. No pressure build up is therefore present beneath the base slab. Central Section, over Sand Layer The central section of the weir is very similar to the typical section, except that the sheet piling extends significantly deeper as it passes through the loose sand layer. Raked pipe piles connected to the upstream end of the base slab are shown on the drawings, however the extents, the raking angle and the founding depths of these piles is unknown. They are called up as being 152mm diameter cast iron pipes which are filled with concrete. The base was belled out with explosives. Based on geotechnical data and drawings, it is likely that the base of the piles would have been founded in the stiff clays beneath the sand layer. Section Adjacent to Right Abutment Beyond the extents of the original 1920s weir, the substructure of the current weir is an upstream reinforced concrete diaphragm wall, 2.7m deep and 610mm thick, in lieu of the double sheet pile arrangement.

Right Abutment The right abutment consists of a vertical reinforced concrete wall which appears to be 610mm thick. Downstream of the wall is fill of an undescribed nature, capped with a reinforced concrete apron slab. Significant erosion and loss of the downstream rock armour has occurred adjacent to the vertical apron wall. The very end of the apron slab has been undermined, and a significant void is now present beneath the remaining slab. Downstream The condition of the downstream large diameter rock armouring is variable. Much of the armour itself is still present, except for the area adjacent to the right abutment. Movement of the individual large rocks has occurred, where the steel cables installed to tie the individual rocks together have been damaged to the point where many are no longer effective. The condition of the apron slab is also variable, with a number of damaged areas, including a zone which appears to have been caused by movement of the underlying rockfill during a flood. This suggests the rockfill beneath the apron slab, or possibly its foundation, is not entirely stable. There are also a number of large holes which appear to be unrepaired test pits. Observed Structural Distress Little structural distress was observed, in the form of structural cracking or differential deflections. The line of the weir face slab is straight with no apparent deviations.

Geotechnical Conditions
The site is generally underlain by stiff clays, with a variable thickness of loose sands above. The loose sands exhibit SPT N values of less than 4, and can reach thicknesses of 5.0m in the centre of the river channel, and towards the left abutment. Soil testing of the material beneath the base slab showed low concentrations of sulfates and chlorides, pH values around 6.5 and resistivities in excess of 5000 ohm-cm. These results suggest that the foundation materials are not aggressive. Groundwater is fresh beneath the structure, in spite of the tidal level reaching up to 2m above the downstream toe level.

Hydraulic Conditions
Earlier work performed by AECOM in 2009 demonstrated that Aplins Weir does not exhibit drownout even at a flood of at least the 1 in 2000 Annual Exceedence Probability (AEP) event (AECOM, 2009). Indeed, a significant afflux of approximately 2.0m persists from the 1 in 100 AEP event up to the 1 in 2000 AEP event (AECOM, 2009). Discharge for the 1 in 2000 AEP event is 1920 cumecs (AECOM, 2009). GHDs structural stability investigations demonstrated that the critical hydraulic loading condition occurs during floods below the 1 in 50 AEP event. This corresponds to a discharge of 300 cumecs over the weir.

Analysis Calibration
The first task for the desktop analysis was to create structural models which demonstrated behaviour consistent with that observed in the real structure. This was an iterative process performed in conjunction with the data mining exercise, as the desktop work highlighted critical structural areas and suggested geometric and structural requirements which in some cases were hazily defined by the structural drawings if they were defined at all. The aim of this activity was to gain an understanding of the particular features of Aplins Weir which are vital to its overall stability. Two series of models were created for this task cross sectional models, and an overall longitudinal model. Cross Sectional Models Three different cross sectional models of the main weir were created, corresponding with the three different structural geometries present along the weir from one bank to the other. These models were three dimensional plate models of a single bay of the structure. The three locations modelled were The cross section adjacent to the right abutment, with the diaphragm wall in leiu of the two rows of sheet piling The typical cross section, founded directly into the stiff clays The central cross section, across the loose sand profile Sensitivity analyses were then performed, with the key output parameter being deflection in the downstream direction under FSL (Full Supply Level) reservoir loading, in order to calculate an effective downstream stiffness value for later input into the longitudinal model. The parameters altered during these analyses were The presence or absence of the cast iron raked pipe piles The moduli of subgrade reaction, for both the loose sands and the clays The section properties of the downstream sheet piling between RSJ sections and flat web type sections Forcing volumetric conservation of the material beneath the base slab The presence or absence of the downstream rockfill placed during the original 1920s construction The structural nature of the connection between the original and upgraded weir, between a pinned connection and a rigid connection (as there are no consistent records of this area) For the first cross section with the diaphragm wall, the modulus of subgrade reaction was the only variable, as all other elements were not applicable. Deflection ranged from approximately 20mm to 35mm. Deflections of the second cross section, the typical section, ranged similarly from 15mm to approximately 30mm. Negligible effects were noted from the cast iron raked pipe piles or volumetric constraint of the foundation the dominant effect was in fact the alteration from a

fixed to a pinned connection between the original weir and the upgrade works. By far the greatest range of deflections was obtained by the central cross sectional model, with values anywhere between 15mm to 260mm depending on parameters. By a very significant margin, the parameter having the most effect in reducing downstream deflections was the rockfill placed downstream of the original sheet pile, underneath the apron slab. Longitudinal Model The final stage of the calibration was to create a longitudinal model of the weir with the three different section types included where appropriate along its length. This model enabled an understanding to be gained of the structural implications of having varying downstream stiffnesses along the length of the weir. The model consisted of the base slab, at full extent, with transverse spring stiffnesses applied at 3.05m centres. The stiffness values corresponded to the appropriate cross sections, and were calculated using the downstream deflection values determined from the cross sectional models created earlier. A consistent suite of values was applied for each run of the longitudinal model. In this way, the parameters most important in contributing to the overall stability of the weir were determined. This was done by correlating the outputs from the longitudinal model most particularly the in plane shear demands in the base slab with observed structural distress (very little) in the actual structure. Calculations demonstrated that the capacity of the existing base slab in shear is very low, due to the low volumes and strength of the reinforcement provided (even this has a degree of uncertainty different drawings show different amounts of reinforcement). This, therefore, places an upper limit on the amount of differential deflection the structure can resist along its length before the base slab exhibits shear distress. The ability for the structure to span across softer zones in the cross valley direction is therefore quite limited. Results The results of this first calibration phase were very interesting. In essence, no combination of parameters except for the inclusion of the passive restraint of the compacted rockfill downstream of the section over the sand layer accounts for the ongoing structural stability of the weir. If that rockfill were to be lost, the analyses indicate that the base slab would fail in diaphragm action and the weir would be lost. This event has come close to occurring, with the development of the large 2.0m deep 7.5m wide erosion hole which opened up in the apron in the 1950s. If that erosion hole had extended further upstream and compromised the degree of compaction of the rockfill at the toe, a large section of the weir may have been lost at that time in addition to the right abutment that failed during that event. Additionally, it was shown that the connection between the original weir and the upgrade works acts with a fair

degree of rigidity. Modelling this connection as pinned also results in high values of diaphragm shear in the base slab.

Paths to Failure
The next phase of analysis was an investigation of the paths to failure of the weir into the future. This considered issues such as ongoing corrosion of structural steel elements, as parts of the structure are 85 years old. A number of analysis runs were conducted using the combination of parameters which led to the desktop output being consistent with the real structure. In these runs, the following components were disabled in turn, modelling loss due to deterioration: Pipe piles Upstream sheet piling Downstream sheet piling Of the above, only the loss of the downstream sheet piling was of any significance from a structural perspective. Loss of vertical support beneath the toe of the structure results in bearing failure and rotation of the structure. Seismic Considerations The loose sands in the area of the weir have a high potential for liquefaction under a minor to moderate earthquake. Liquefaction could lead to loss of lateral support to the downstream sheet piling, resulting in buckling and settlement / rotation of the structure. Therefore, the downstream sheet pile is considered to be vulnerable under both the deterioration scenario and the earthquake scenario.

heavily reinforced concrete overlay slab dowelled to the existing structure. The intent of this concept is to completely bypass all existing vulnerable substructure elements, and result in a structure with a quantifiable capacity.

Design Philosophy
Design scenarios were grouped into base load cases and extreme load cases. The load cases and their individual design scenarios are described in the following paragraphs. Base Load Cases Base load cases (BLCs) were assigned a required FOS (Factor of Safety) of 1.1. The relevant design scenarios were: BLC1: Reservoir at FSL following flood which caused loss of downstream rockfill BLC2: 1 in 10,000 Average Recurrence Interval (ARI) earthquake, with downstream rockfill present and effective BLC3: All flood cases, with downstream rockfill present and effective Extreme Load Cases Extreme load cases (ELCs) were assigned a required FOS of 1.0. The relevant design scenarios were: ELC1: 1 in 10,000 ARI earthquake, with downstream rockfill either not present or ineffective ELC2: All flood cases, coupled with loss of downstream rockfill

Upgrade Concept
To satisfy current safety criteria, the upgrade concept for Aplins Weir needed to answer the following requirements: The weir should no longer be reliant upon the downstream rockfill for sliding stability, which is vulnerable to erosion under flood events and difficult to quantify from a capacity calculation perspective The weir should no longer be reliant upon the competence of the moment connection between the original and the new weir, which was constructed to an unknown detail The weir must have robust vertical support to the downstream edge of the base slab. Currently the vertical support provided by the RSJ sheet pile wall is contingent not only on the condition of the steel, but also on the ability of the substrates to provide sufficient lateral restraint to prevent column buckling failure of the steel members Any solution should be structurally quantifiable The chosen upgrade concept answering all the above requirements consists of retrofitting a 1200mm diameter lined cast in place pile in each bay, at the downstream one-third point of the existing base slab, centrally placed between buttresses. The pile will be structurally connected to the existing weir through a 650mm thick 5

Detailed Analysis of Upgrade


The aim of the analysis of the upgraded structure was to quantify the design actions within the new weir components in order to proceed with the detailed design. In this case, however, this was not a single step process, as is often the case. The main complications were: The complex nature of the existing foundations and ground conditions The limited ability of the existing structural elements to plastically redistribute under ultimate conditions The existing structure will still be under partial load during the installation of the strengthening elements, as a drawdown of the reservoir of only 1.5m is expected to be possible Therefore, in order to obtain meaningful results, time step modelling which included construction and load sequencing was performed. Geometry A three dimensional finite element model of the entire structure was created, with emphasis placed on capturing an accurate structural mass and geometry. This was achieved using plate elements for the existing weir components and the new overlay slab, and beam elements for the retrofitted piles. The new base overlay slab was modelled separately from the existing, with the two connected using master-slave links. These links were the

only modelled connection between the existing structure and the retrofitted components, reflecting the reality of the final structure. Materials 20MPa concrete was used for the existing weir components. The new works have been specified as 40MPa concrete. The final soil parameters used are listed below: Modulus of subgrade reaction for rockfill at downstream toe 80,000kN/m3 Modulus of subgrade reaction for stiff clays 80,000kN/m3 Modulus of subgrade reaction for loose saturated sands 20,000kN/m3 Spring stiffness applied to retrofitted pile varying with depth, and obtained from the SPT values from the geotechnical investigations. Values ranged from 14,400N/mm at the top of the stiff clay surface, to 44,640N/mm at 7m depth Analysis Procedure The nonlinear solver was used to capture the time step effects, as follows: Applying self weight and reservoir load at construction drawdown levels, disengage the retrofitted components from the existing structure in the model. The results of this run represent the preexisting stress condition in the existing structure at the time of construction Re-engage the retrofitted components to the existing structure, and continue the analysis to apply reservoir load at FSL. This provides results at normal operating conditions Continue the analysis applying the 300 cumec flood load case. This provides the results for BLC3, defined earlier Continue the analysis, removing the restraining effect of the downstream rockfill. This provides the results for the second of the extreme load cases, described earlier Continue the analysis, reducing the water level back to FSL. This provides the results for BLC1. An example of this output can be seen in Figure 5 Conducting the analysis in this fashion tracks the load transfer in the structure, as significant events occur.

Figure 5 Preliminary modelling output of nonlinear time step process, capturing the partially loaded existing structure and subsequent strain compatibility with the retrofit components. Smaller preliminary model shown instead of final complete structure for clarity The seismic load effects were obtained using the spectral response solver. The acceleration response spectra from the seismicity study for the Ross River Dam were used for Aplins Weir, due to the relatively close proximity of the two sites. An initial structural damping factor of 5% was used. This value is appropriate if the structure is operating entirely within the elastic range, and therefore not accessing the energy dissipation associated with the creation of plastic hinges. This was later verified following the detailed design, taking the contribution of the steel liner into account during the calculation of the pile capacity. Concrete cast in place piles with permanent steel liners exhibit capacities greatly in excess of the normal attributed capacity calculated when ignoring the contribution of the steel liner.

Piping and Seepage Considerations


Seepage is currently controlled by a double row of sheet piles, which will remain in place following the upgrade. As stated earlier, the soils beneath the structure are not considered aggressive. As such, a corrosion rate of approximately 0.01mm/y can be expected, at which rate the design life of 100 years will be easily achieved from a seepage perspective. As the soils under the structure are sandy, piping and dispersion are considered unlikely.

Conclusion
This paper presents the investigation, analysis and design activities undertaken by GHD during the design of the upgrade of Aplins Weir, on the Ross River in Townsville. Extensive historical data mining was undertaken due to the long and interesting history of the structure, in order to gain an understanding of the existing situation. Nonlinear time step finite element analysis was required to obtain meaningful results for the detailed design phase, as more simple analyses would not have captured the stress history effects. The final upgrade solution is an

unusual one, requiring an intensively structural retrofit of large diameter cast in place piles with a heavily reinforced concrete overlay slab.

Acknowledgements
The authors gratefully acknowledge the support of the Client, Townsville Water, in the publication of this paper. The supply of the photographs and drawings from their archives for this project was very much appreciated.

References
AECOM 2009. Aplins, Gleesons and Black Weir Risk Assessment.