Retaining wall design considerations to AS 4678-2002 and AS 2159-2009

Common uses and failure modes for sleeper retaining walls

Sleeper retaining walls are used whenever there is a need to retain sloped soil, whether to create a terrace garden, flatten a backyard for usable space, or prevent soil erosion around an existing foundation. Brooks Smith noted they are a common form of residential construction in Australia, in contrast to the United States where this type of wall, known there as a soldier pile retaining wall, is almost exclusively used for temporary construction works.

Brooks drew on photos taken along the Dobbin Creek Trail in Bundoora to illustrate three typical failure modes. The first was sleepers deflecting outward and showing signs of rot, with plants growing through the wall, pointing to inadequate drainage behind it. The second and most serious was a post leaning heavily forward, indicating that it had rotated within the soil due to insufficient embedment depth below the retained surface. The third was a wall that had been propped with diagonal supports, which Brooks described as a sign the original posts were too weak and the wall had already failed structurally.

The four-step design process

Brooks outlined four sequential steps for designing a sleeper retaining wall, each referencing a different standard.

The first step is determining soil properties and calculating lateral loading using AS 4678-2002. This standard is relatively vague by geotechnical standards and does not prescribe equations step by step, but it does define the limit states that must be satisfied. The most common approach for calculating active earth pressure is the Rankine-Bell method, which treats the dead load and groundwater as a triangular distributed load and any surcharge as a uniform rectangular load over the full retained height. Brooks emphasised that Amendment 1 to AS 4678 raised the minimum live load surcharge to 5 kPa for walls over 1.5 m high, and that some manufacturer span tables still use the superseded 2.5 kPa value.

The second step is assessing the geotechnical capacity of the laterally loaded pile under AS 2159-2009. This covers the translation and rotation limit states. Calcs.com uses the Broms method, which simplifies the actual soil reaction curve into two rectangles, analogous to the Whitney stress block used in concrete design. Brooks explained this choice reflects a balance: Hosking gives lower embedment depths that have been associated with real failures, while P-Y curve methods require detailed soil data that is rarely available on residential projects.

The third step is structural design of the steel post to AS 4100. Brooks noted that the main difference from a standard beam-column design is that the load combination includes soil lateral pressure, which engineers may not commonly encounter. The concrete pile below the post is conservatively assumed to contribute no structural strength, as the critical stress point is at the base of the retaining wall above the pile.

The fourth step is designing the sleepers. Sleepers can be timber, concrete, or other durable materials, and in Calcs.com they are designed by linking to a separate beam calculator for the chosen material. Because the triangular soil pressure is largest at the bottom of the wall, the bottom sleeper carries the highest loads. Engineers can use a single sleeper size for the full height, or optimise by specifying different sizes at different heights.

Comparing design methods for laterally loaded piles

Brooks presented a comparison of several methods for calculating minimum pile embedment, focusing on soft clay as a representative case. The methods compared included Hosking, Broms, Zerny-Ack, the now-deprecated AS 4676 prescriptive method, and two US methods from IBC and the Outdoor Advertising Association of America.

The US methods produced considerably deeper embedment requirements than any of the Australian methods, which Brooks suggested may explain why sleeper retaining walls are uncommon as permanent structures in the United States. Among the Australian methods, Hosking is the least conservative and gives the shallowest embedment depths. The deprecated AS 4676 method produced a different curve shape, with low embedment for short walls that increased steeply for taller walls.

Broms sits in the middle of the Australian methods, and Brooks said it had good backing in the scholarly literature and a track record of accuracy in published research on sleeper retaining walls. He also cautioned that comparing results between software packages is not straightforward, because different packages may only include post deflection, or both post deflection and pile rotation, and soil properties may be entered with very different levels of detail and with different underlying assumptions.

Deflection and serviceability considerations

Brooks explained two approaches to the serviceability deflection check. The first considers post deflection only, treating the pile as a fixed support at the soil surface. The second also includes pile rotation within the soil, giving a larger total deflection figure.

The choice between them depends on context. For shorter walls where the post is deliberately embedded at an angle, commonly around five degrees, towards the backfill, the pre-rotation counteracts expected forward deflection and including pile rotation in the check may not be necessary. For taller walls, or where consequences of movement are more significant, including pile rotation provides a more complete serviceability assessment. Brooks noted that AS 2159 does not prescribe which approach to use, so the decision rests with the designer's judgement of the specific situation.

Using the Calcs.com calculator: worked example

Brooks demonstrated the calculator using a worked example: a 1.4 m high wall retaining cohesionless gravel backfill (class 1 controlled fill, friction angle 25 degrees, unit weight 18 kN/m3), with a chain link fence of 1.8 m height attached to the top, post spacing of 1.5 m, and a firm clay foundation soil with undrained shear strength of 50 kPa.

He showed that the minimum embedment depth is calculated automatically and defaults to the Broms minimum for the current inputs, changing live when soil parameters are updated. In the example, entering a firmer clay reduced the required embedment from 2.75 m to 1.9 m. The calculator also defaults the coefficient of horizontal subgrade reaction, a value not commonly reported in residential geotechnical reports, to a conservative figure derived from the undrained shear strength, with a table provided so engineers can verify or refine the value.

Post sizing is handled through a member selector that shows which sections pass or fail instantly. Brooks selected a 100 UC after the default section showed very low utilisation ratios. The sleeper design is completed by creating a separate timber or concrete beam calculation and linking it to the retaining wall, which passes through the relevant loads automatically. Changing the retained height in the main retaining wall calculator is immediately reflected in the linked sleeper calculation.

Brooks confirmed that at the time of the webinar the calculator supports cohesive foundation soils only. Users needing cohesionless foundation soil support were encouraged to submit a request via the Calcs.com help email.