Engineering The Mansion’s Heritage Ken MacLeod, MacLeod Consulting, Melbourne, Australia Slav Tchepak, Vibro-pile (Aust) Pty Ltd, Melbourne, Australia A development at 83 Queens Road in Melbourne, Australia required construction of a three level basement. The normal geotechnical constraints of excavating adjacent to an existing relatively new building were challenging enough for the complicated soils profile occurring at the site. An additional significant challenge faced by the designers, was the requirement to excavate underneath the entire footprint of an existing heritage listed building, known locally as the Mansion, in the centre of the site. Maintaining the structural integrity of that building was afforded the highest priority during the 10m excavation. This paper describes geotechnical aspects of the analysis, design and construction of the secant retention system adjacent the new building and support of the Mansion enabling the 10m excavation to proceed without causing damage to the structure. The secant wall comprised hard-soft interlocking piles constructed using CFA methods. CFA piles along the length of the Mansion provided short term support, with the building load being transferred to the piles by large steel beams jacked under the building. Survey results confirmed that movements of the retaining wall and also the load bearing piles were minimal and well within design expectations. INTRODUCTION 83 Queens Road is a 20-level residential apartment structure located about 5km southeast of Melbourne’s city centre. The development has 3 levels of basement carpark, with the lowest basement being about 10m below ground surface level. The basement occupies the entire site, and extends beneath a 2-storey heritage listed Edwardian building referred to as the Mansion. Excavation for the construction of the basement required the installation of a site boundary retention system and support of the Mansion while basement excavation work proceeded. Both of these requirements are regularly encountered in building works, however this development was complicated by: •

•

The presence of an adjacent new 16storey structure on the northern boundary, the owners of which were anxious regarding the perceived risk to their building resulting from the proposed development. The owner insisted upon unusually rigorous engineering design analyses being provided, to verify that excavation for the development could proceed without damaging their structure, or disrupting their operations. The Mansion building had to be safely supported during all phases of the work, especially during excavation of the 10m

• •

deep basement. This excavation was the critical activity that could result in damage to the Mansion. Complex soil and ground water conditions. Groundwater drawdown to just below the bulk excavation level.

Given the above constraints, any retention and foundation systems adopted would have to meet demanding performance criteria, be of the highest standard of workmanship, and be closely monitored to ensure compliance with both the designers’ and statutory requirements. Wall movements for deep basements are inevitable. However it was vital that the retention system be sufficiently stiff to minimise soil movements, hence limiting the potential for damage to adjacent structures. The retention system should also be watertight to limit potential problems associated with ground water drawdown under the adjacent new building from dewatering activities. PLANNING REQUIREMENTS Although only some 200 years since European settlement in Australia, in recent decades there has been growing interest from the broad community in the preservation and conservation of heritage buildings. Relatively recently, Government regulations were introduced to protect buildings considered worthy of heritage listing. Within this context,

the commercial development of this icon site required retention of the 1896 residence in conformance with the state heritage strategy and mandated by the local government planning approval.

underlying three basements have been founded on a partially post-tensioned raft slab. This raft slab spreads across approximately 50% of the site area under the footprint of the multi level building.

HERITAGE ASPECTS

The extent of the raft slab incorporates that part of the site under the northern wall of the heritage building. The columns from the multi level building supporting the transfer slab cantilevering out above the heritage building are located immediately north of the northern most wall of the heritage building.

This prominent site at the south end of Queens Road, is the location of “Clarence,” a two storey rendered brick residence heavily decorated in the Victorian Italianate style. Further additions were made to the original residence in the 1920’s, 1950’s and 1960’s. In recent times, the historic building was used as the “Mansion” nightclub, the name by which this building is now commonly known. The conservation of heritage buildings assumes the preservation of as much of the existing fabric of the building as is possible. Relocation or demolition is actively discouraged. As the Mansion building included ornate facades, plaster ceilings and wall detailing internally, meeting the requirements of the conservation guidelines required the building to be supported insitu without distress to these finishes.

In order to transfer the 20 storeys (and potential 26 storeys) of building load through these columns, the following construction sequence was adopted: •

Construction of piles/columns.

•

Installation of steel support beams under the heritage building.

•

Encapsulation of the ends of the steel beams and heads of the temporary piles in doubly reinforced, reinforced concrete ground beams parallel to the north and south walls of the heritage building.

BUILDING DESIGN ASSESSMENT

•

Excavation of the underlying soils to the required lower basement depth.

The retention of a heritage building on this commercial site imposed financial penalties on the development, as well as physical constraints. In order to address the financial penalty of preservation works, the new multi level apartment building was reconfigured internally to achieve a higher commercial yield. Allowance has been made for potentially adding a further six storeys, which if approved would result in more premium priced apartments.

•

Construction of the multi level building raft slab partly surrounding the temporary piles/columns.

•

Erection of the suspended basement floors structures including final supporting columns for the multi level tower and heritage building.

•

Specific connection to the underside and topside of the previously constructed ground beams for the major multi level building columns.

Reconfiguring the building yield to increase the number of apartments, and allowing for potential future storeys impacted on the planning requirements for carparking. To provide for sufficient carparking for the reconfigured and potential development, an additional third level of carparking was required to access the entire site, including underneath the Mansion building. Physical constraints on this site included roadways bounding two sides of the trapezoidal shaped site, and a 16-storey apartment building immediately beyond the north boundary. As the bearing capacity of the underlying soils on this site have diminishing capacity with depth, the multi level apartment building and

temporary

bearing

This structural design philosophy was based on being able to transfer the load from the temporary piles/columns to the final columns and final footings structure. The temporary piles/columns were required to have a capacity of approximately 2500kN, with some paired columns carrying up to 4250kN, whereas the permanent columns to the multi level building have a required capacity of 25,000kN. On completion of the multi level building component, the temporary support piles/columns to the heritage building were cut away. This transferred all load to the permanent footing system.

SITE LOCATION AND GEOLOGY The site is located in an area predominantly comprising office and residential developments. The northern boundary has a relatively new residential building, the nearest columns for which are about 6m from the site boundary. The adjacent building is supported on pad footings located about 1.6m below the natural surface. The pads are about 4.2m x 5.5m in plan dimension and have been designed for an allowable bearing pressure of about 375kPa.

The ground water table occurred at a depth of about 6m, i.e. about 0.40 of the maximum retained height, so dewatering would be required to enable excavation to achieve the bulk excavation level. A profile of SPT N-value with depth is shown in Fig 1. ` SPT vs Depth N (blows/300mm) 0

20

40

60

80

0 ……….. …sand. .……… ………. ………. ………. ……….. ..sand… …….. ……….. ………. ………..

clay

Ironstone 5

10 Depth (m)

The site geology comprises Tertiary aged sediments, the upper part of which is known locally as the Brighton Group, generally comprising dense sands, to a depth of about 17m. Underlying the Brighton Group is the Newport Formation, which comprises very stiff clays. Igneous intrusions in the form of Granite, usually extremely to highly weathered, have been reported at depths of 25m or greater in the area.

BH1 15

BH2 BH3

20

25

The sands of the Brighton Group vary from medium dense to very dense and contain ironstone seams of up to 300mm thickness. The ironstone seams vary from very high strength, requiring diamond drilling methods for penetration during site investigation; to very dense cemented layers that can be readily penetrated using tungsten tipped augers. The ironstone did not occur as continuous layers across the site. The sands had SPT “N-values” in the range of 14 to Refusal, but predominantly being greater than 30. The Brighton sands are underlain by the clayey silts of the Newport formation, in turn underlain by very stiff silty clays comprising the Werribee formation. Quantification of the strength properties of the silty clay was limited to two undisturbed samples, which were subjected to undrained unconsolidated triaxial tests that indicated an undrained cohesion of 63 and 87kPa with an angle of internal friction of 7 degrees. The silty clays in the area are understood to usually be of very stiff to hard consistency. At depths ranging from 21 to 26m the silty clays had traces of quartz and minerals indicating they were of granitic origin. No quantification of strength was undertaken of the residual granites over the maximum bore depth of 30m, but they were easily penetrated by wash boring methods.

30

Fig. 1 SPT vs. Depth

ORIGINAL SOLUTION Mansion support comprised a system of capping beams resting on foundation piles that essentially act as underpinning members. It was essential that those piles had excellent load-settlement performance to ensure no damage to the Mansion. Also, the piles had to be strong enough to accommodate the design loadings without buckling, as the pile shafts would be exposed over significant lengths. The initial solution to the foundation problems suggested by the structural engineers comprised: •

•

•

A 600mm dia secant pile wall to the north boundary, with 4 rows of ground anchors and 3000mm embedment depth below bulk excavation level. A 600mm dia soldier pile wall, with piles at 2.1m centres and incorporating the use of sprayed concrete infill panels and up to four rows of soil anchors to complete the retention system to the south, east and west boundaries. Utilising large diameter bored piles to provide “temporary” support for the

capping beam arrangement designed by the structural engineer. Steel sections were proposed to be installed in 1500mm dia bored piles and concreted in place up to the bulk excavation level. Above the bulk excavation level the annulus between the steel sections and the walls of the bored pile shafts were to be backfilled with soil or stabilised sand. It was envisaged that the steel sections would be used to transfer the loads from the Mansion to the bored pile. The structural purpose of the steel tube sections was to allow simple future connection to suspended basement slab band beams as well as avoiding the permanent column footings. The original design solution had aligned these band beams to the “temporary” and permanent Mansion support columns. This proved to be a costly solution, relying heavily for its success on the end bearing capacity of the relatively short bored pile section, in variable soils, below the bulk excavation level and below the water table. FINAL SOLUTION The final solution, after consultation between the builder (Melbourne Transit), structural engineer (MacLeod Consulting) and piling contractor (Vibropile) comprised the following: •

•

The sprayed concrete infill panels of the soldier pile solution were deemed to be too risky because of the presence of the high water table. Instead, precast concrete sheet piles, restrained by temporary ground anchors, on the south, east and part of the west boundaries, were installed by the builder. Lateral movements for these walls were not critical, as they were adjacent to roadways only. The precast panels were installed by vibratory means. These 250mm thick panels were 2.5m wide and 9.75m tall, nominally embedded 2m below bulk excavation level. An anchored 600mm hard-soft secant pile retention system along the northern boundary adjacent to the neighbouring building, extending around part of the western boundary with 8500mm embedment depth below bulk excavation level. The toe of the secant pile wall was founded in very stiff clays. Because that stratum was relatively impermeable, dewatering effects would be mitigated along this “critical” section of wall. Both

•

•

the hard and the soft piles were constructed by CFA methods. Underpinning piles to support the Mansion were proposed as an alternative to the steel tube/bored pile by Vibropile, and installed using its CFA (Continuous Flight Auger) system. The CFA piles were concrete injected, not grout. The alternative solution was significantly cheaper than the conforming. Where required to meet loading requirements that were beyond the safe limits of a single CFA pile, two contiguous piles were constructed (ref Figs 5 and 6). Revision to the suspended basement slab band beam alignment to avoid the “temporary” Mansion support pile (columns). Consequently previous intrusions into the highly stressed areas of the band beams by the original solution was now avoided.

It was of paramount importance that the quality of CFA pile construction could be relied upon for the underpinning piles in particular. Vibropile’s CFA rigs are all equipped with computers that monitor every aspect of pile construction, including: • • • • • •

Penetration per revolution (mm/rev). during the drilling phase. Torque (kNm) during drilling. 3 Drilling resistance (MJm ) during drilling. Auger extraction rate (m/min) during the concreting phase. Concrete oversupply (as a percentage over the theoretical requirement). Concrete pressure (bars) during concreting.

These properties are monitored in real time and provide a full history of the construction of every pile on the project. Engineers have access to the records via modem from the office, as required. All records are scrutinised by engineers to ensure that design requirements are not compromised. This extensive monitoring system helped to ensure reliable construction of the piles and reassurance that the CFA alternative posed less risk in terms of potential construction problems. The monitored properties enable the computer to draw a profile of the expected pile shape at the conclusion of construction. This level of monitoring is considered vital to ensure the required quality of construction, and systems that only monitor concrete pressure and

volume are not considered to be satisfactory. The penetration rate during drilling can flag potential soil loosening during drilling. The auger extraction rate during drilling is vital, particularly on the “initial” pull if end-bearing is important in the pile design. Sole reliance on the Operator to maintain a consistent rate of extraction is not regarded as acceptable. A typical record for a CFA pile is shown in Fig 2. Note the presence of ironstone indicated at about 12m depth, where drilling penetration reduced and drilling resistance increased.

DESIGN AND CONSTRUCTION NORTHERN RETAINING WALL

OF

A hard-soft secant wall was adopted as the most cost-effective solution for this boundary. This wall comprised 600mm dia “hard” CFA piles reinforced to accommodate the soils, water and surcharge loadings, cut into and overlapping the soft unreinforced piles by 100mm. All piles were founded at a depth of 17m in accordance with design requirements. The soft pile mix was accomplished by using a

Fig. 2. Typical PL2000 computer output CFA pile (note presence of ironstone layer for this pile) bentonite-cement mixture. By having the piles cut into each other, watertightness is The changes to the retention system were enhanced. The finished wall would also be driven by cost, with precast reinforced suitable for the permanent basement wall. concrete sheet piling proving to be an economic alternative. However, that system was not used on the northern boundary The design for the secant retention system primarily because the panels could not be was done on a “design and construct” basis by installed to the required depths, but also Vibropile using the program WALLAP (written because the wall movements of the relatively and distributed by Geosolve UK). WALLAP is flexible panels would be too high even with limited to analysing a retaining wall as a beam four rows of ground anchors. in which the supports are represented as springs. Whilst providing estimates of

movements of wall itself, it cannot provide information on excavation-induced movements. The desired estimates of lateral and vertical movements of the adjacent footings as a result of the adjacent excavation were undertaken by geotechnical consultants using the program PLAXIS (Plaxis BV, The Netherlands).

carried out by local consulting geotechnical engineers Coffey Geosciences, this time yielding more believable results. The design parameters adopted by Coffeys are tabulated in Table 1 below. The difference in results between the two analyses was the consequence of the initial analysis use of an over-simplistic soil model.

The initial PLAXIS analyses done by one consultant yielded questionable results. Those results indicated that the adjacent footings would heave upwards as the secant wall moved laterally during excavation of the basement. An independent analysis was

The opportunity was taken to compare the results of the WALLAP and final PLAXIS analyses. This parametric study was done essentially using the same soil and wall design parameters, the results of which are summarised in Table 2.

Depth (m) 0 2

Description

c’ (kPa)

φ’ (deg)

ψ’ (deg)

E’ (MPa)

µ’

γ 3 (kN/m )

Ko

Fill

0

33

10

20

0.3

20

1.0

0

37

25

70

0.3

20

1.1

Medium dense – dense SAND (N = 14 to >50)

0

35

15

50

0.3

20

1.0

Stiff v.stiff CLAY (N typ 17)

5

25

5

30

0.3

19

1.1

Dense SAND

0

37

25

55

0.3

20

1.1

5

25

5

30

0.3

19

1.1

Dense SAND (N typ >50) 6

17

23 27

stiff – v. stiff

CLAY (strength quantified)

not

50 Granite rock (assumed) (strength not quantified) 60 Table 1. Design parameters for secant pile wall design.

Table 2. Results of retention analyses Program PLAXIS WALLAP

S1 (mm) 2 -

S2 (mm) 5 -

∗1 (mm) -5 -24

where S1 =

settlement of near end of first line of adjacent footings S2 = settlement of far end of first line of adjacent footings ∗1 = lateral displacement at top of retaining wall ∗2 = lateral displacement of retaining wall at final excavation level Fmax = Maximum anchor force developed Mmax = Maximum bending moments developed in the retaining wall at any stage during excavation.

The following observations were made from the above-mentioned analyses: •

•

•

The importance of selection of appropriate design parameters was highlighted in the initial PLAXIS analyses, as that was the cause for prediction of adjacent footings being in the “wrong” direction. The estimated maximum anchor forces and bending moments from the WALLAP and final PLAXIS analyses were essentially equal. The estimated wall movements from the WALLAP and PLAXIS analyses differed significantly.

Whilst the maximum anchor forces and maximum bending moments were remarkably similar, the variation of the profiles of bending moment versus depth between the two analyses (Fig 3) is of interest. The results of the PLAXIS analysis were adopted for the structural design of the hard piles. Consequently, the “hard” piles were reinforced over the top 14m only as the bending moments below that depth could be safely resisted by plain concrete. The upper 10m of the cage comprised 8N32 bars with 4N32 bars extending an additional 4m (“N” grade structural steel in Australia is the standard and refers to ductile steel with a nominal yield strength of 500MPa). 40MPa concrete was used for the hard piles. The soft mix comprised a cement-bentonite mix having typical unconfined compression strength of about 5MPa.

∗2 (mm) 26 12

Fmax (kN/m) 378 389

Mmax (kNm/m) 221 239

The actual soil movements were significantly less than the design estimates and no damage resulted to any structure. Unfortunately no lateral movement measurements of the secant pile retaining wall were done. However, a series of levels were taken at a number of locations at the adjacent building site. The maximum vertical movement recorded was 1mm for the closest footings, the southern edge of which are located about 6m from the secant pile wall. MANSION SUPPORT The underpinning piles were redesigned using CFA piles which were selected primarily because of low cost, speed of construction and their installation being unaffected by the presence of water. The CFA piles were a far more economical solution than the conforming scheme, but more importantly, by designing piles to found at depths of up to 33m, pile settlements were estimated to be minimal, thus reducing the risk of damage to the Mansion. A further desirable feature of Vibropile’s CFA system was the comprehensive computerised monitoring system mentioned above, that ensured that a structurally sound product could be reliably produced – a most important aspect given that the upper 10m of pile shafts would be exposed and required to resist buckling. Because no quantification of the strength of the soils below 25m was undertaken the design of the underpinning piles had to be done on a conservative basis. The initial design parameters below the bulk excavation level were taken as: • •

ultimate shaft friction resistance “fsu” = 60kPa ultimate end-bearing pressure “fbu” = 2.5MPa in the stiff clays.

During the installation of the first piles it became obvious that the soils below 23m depth comprised residual granite of hard consistency, with the rock fabric visible on disturbed samples on the auger flights. As a consequence the design parameters were cautiously increased to 100kPa and 4MPa for ultimate shaft friction and end-bearing respectively below 25m depth..

83 Queens Rd Retention analyses Bending moment (kNm/m) -200

-100

0

100

200

300

6 4

Reduced Level (m)

2 0 Wallapmin -2

Wallapmax Plaxismin

-4

Plaxismax -6 -8 -10 -12

Fig 3. Comparison of results of PLAXIS and WALLAP analyses. The CFA piles were 600mm and 750mm nominal diameter founded at depths ranging from 20 to 33m to accommodate the design loadings ranging from 1000kN to 4250kN, that were transferred from the Mansion to the piles via the capping beam system. Where the design indicated that loads might be too high for single piles, possibly resulting in settlements greater than desired, two contiguous (touching) piles were installed to ensure desired performance (shown in Figs 5 and 6). CONCLUDING COMMENTS The development at 83 Queens Road had a number of constraints that affected the commercial viability and the engineering requirements. These constraints included the preservation of a heritage listed building on the site whilst a three level basement was excavated beneath and the presence of an adjacent 16-storey building close to one boundary.

delivering the stated aim of preserving the integrity of that building and its internal finishes. Photographs of various aspects of the work, including the partially completed development are depicted in Figs 4 to 7. The foundation and retention system adopted was the result of comprehensive consultation between the builder, structural engineer and piling contractor and successfully met the high standards of excellence in preserving the heritage listed structure. ACKNOWLEDGEMENT The input of Prof Harry Poulos of Coffey Geosciences, who carried out the PLAXIS analyses of the secant pile wall and provided comments on a number of aspects of the content of this paper, is gratefully acknowledged. REFERENCES

The design and performance of the hard-soft secant piles wall resulted in negligible movements of the footings of the adjacent building. The underpinning piles were constructed by fully instrumented CFA methods. The performance of those piles also resulted in negligible movement of the heritage building,

Heritage Victoria, 2004. Victoria's heritage 2010 Heritage Victoria, 1999. Draft guidelines for the assessment of heritage planning applications Heritage Victoria, 2001. Heritage listings and

property valuations in Victoria. Australia ICOMOS, 1999. The burra charter: The Australia ICOMOS charter for places of cultural significance

Fig 4. The Mansion. CFA rig to left and adjacent building right background.

Fig 6. CFA Mansion support piles.

Fig 5. CFA Mansion support piles with secant wall in background.

Fig 7. Partially completed tower with Mansion