SUMMARY
Having decided to proceed with the Steepbank Mine development in early 1996 on an accelerated schedule, Suncor needed permanent access across the Athabasca River by late 1997. A design/build contract offered the only feasible procurement method. The successful contract represented innovation in bridge engineering, uniquely blended design with construction resources, and needed to fit into a very tight construction schedule.

The 2.5 m diameter concrete filled caissons were drilled into the underlying bedrock from barge mounted equipment. The 3.3 m deep plate girder sections were erected from an ice bridge during winter, in temperatures as low as -45 degrees C. Concrete deck construction was completed in spring and summer. The design/build contract was awarded in June 1996 and the bridge was opened to traffic in August 1997. The paper describes some of the challenging and unique aspects of the project as well as the design/build process which allowed the $25 million, 391 m long bridge to be designed and built in only 15 months.

INTRODUCTION
In late February 1996 Suncor Inc. Oil Sands Group (now Suncor Energy) invited tenders, based on design/build proposals, from three pre-qualified contractors for a bridge across the Athabasca River adjacent to their extraction and upgrading plant at Tar Island, some 35 km north of Fort McMurray (Figure 1). After a three month proposal preparation period a $25 million contract was awarded to Peter Kiewit & Sons Co. Ltd.

The paper describes some of the challenging and unique aspects of the bridge design and construction that were solved to enable the bridge to be opened to traffic only 15 months after contract award.

PROJECT REQUIREMENTS
A preliminary engineering study, including geotechnical, hydraulics and ice effects, had been completed and formed part of the Terms of Reference. Proponents were encouraged to provide alternate designs that provided cost savings to the owner and complied with the following design requirements:

• The alignment and horizontal geometry was fixed in accordance with the preliminary design.
• The navigational clearance envelope of 60 m horizontal and 15.2 m vertical was prescribed - although with some flexibility for the position of the channel horizontally.
• A maximum of four in-river piers were permitted.
• The west abutment location was fixed.
• The river width was not to be less than 350 m.
• The bridge was to be open to traffic on November 30, 1998.
• Specific design criteria, including width of bridge, design loads and codes were to be used.

The evaluation criteria (in no particular order) were given as:

• Lump sum cost/unit price.
• Project schedule.
• Project team experience.
• Bridge design concept.
• Safety performance experience.
• Construction methodology.
• Resource allocation.
• Alternatives, options contract considerations offered for Owner=s selection.
• Systems and tools for contract control.

As the proposal development period progressed, environmental issues and a schedule that would enable early completion clearly became more important and would influence the selection of the successful design/build team.

Consideration of the selection criteria led us to the conclusion that the successful proposal would possess the following attributes:

• Offer a cost-effective design.
• Provide the most certainty of on-schedule completion.
• Accommodate potential changes in subsurface conditions.
• Minimize environmental impacts on the river.
• Require minimal in-service maintenance.

The calculated ice loads on the bridge piers for 20 year ice strength, 20 year ice thickness and 100 year ice elevation are given in the following table:

Table 1 - Calculated Ice Loads

Load Type	Location	Magnitude (kN)	Elevation (m)
Longitudinal (1)	front pile	7400 (2)	241.5
Transverse (1)	front pile	2100 (3)	241.5
Thermal (4)	front & rear pile	700	235.0
Ice jam	front pile	150	237.0
Vertical (4)	front & rear pile	1500	235.0

(1) Refers to load parallel to direction of flow in the river.
(2) Applied with a concurrent transverse load of 1100 kN. Concurrent loads of 3000 kN longitudinal and 450 kN transverse to be applied to other piers.
(3) Applied with a concurrent longitudinal load of 3700 kN. Concurrent loads of 860 kN transverse and 1500 kN longitudinal to be applied to other piers.
(4) Applied simultaneously to both piles

GEOTECHNICAL CONDITIONS AT THE SITE
The preliminary geotechnical investigation showed that the in river section had from 14 to 26 m of fine to medium sand, with fine to coarse gravelly layers, overlying bedrock of interbedded crystalline limestone and claystone. West of pier 1 bedrock was not encountered in two boreholes drilled to depths in the order of 50 m. Silt and clay alluvium was found near to the surface at each bank.

Both the elevation of the upper surface and the strength and makeup of the bedrock were highly variable. The bedrock ranged from a crystalline limestone that rang when hit by a hammer, with a strength in the range of 50 to 100 MPa, to a claystone that could be crushed by hand, with a strength of 1 MPa or less. The materials were highly interbedded and it was difficult to correlate even between closely spaced coreholes at the same general location (Figure 3).

Better definition of the rock was required to allow assessment of shaft and end bearing capacities. While design progressed a major drilling program was undertaken with 40 holes - 5 at each caisson - drilled into the rock. The final depth of each caisson was optimised using properties of the rock at each location. The rock was categorized into five categories and values assigned to the unconfined compressive strength for each category of rock.

Table 2 - Major Groups In Limestone

Class	Description	Literature (UCS)	UCS Pile Design
			Mean	99% exceeded mean	Sensible lower bound	Design
A	Strong, massive limestones or dolomites with occasional stringers of calcareous shale or argillite (<10%).	>5000 psi (34 MPa)	56	50	35	35
B	Moderately strong, generally massive or thickly medium bedded argillaceous limestones (with numerous random (10%) shale or argillite stringers).	4000 psi 28 (MPa)	25.3	17.4	16	18
C	Moderately weak to weak, thinly interbedded limestones and 10 to 20% calcareous shales. This material may be susceptible to minor slaking on exposure.	2000 psi (14 MPa)	8.4	3.4	2	3
C/D	Interbedded moderate weak to weak shale or limey shale, 30 to 50%. This material is susceptible to slaking on exposure.	1500 psi (10 MPa)	2.2	1.1	1.1	1.1
D	Weak, very thinly bedded shale or limey shale, grading to massive beds 1 to 3 feet thick. 50 to 100% calcareous shale. Material strongly susceptible to slaking on exposure.	50 to 500 psi (0.3 to 3 MPa)	0.63	0.37	0.37	0.37

SUBSTRUCTURE
The capacity of each caisson was calculated on the basis of tip only (conventional end bearing), shaft only (assuming a clean shaft only and no extra measures such as grooving of the side walls) and combined shaft and end bearing. The factored capacity of a combined socket depends on the assumption made between strain compatibility of the shaft bond and mobilization of tip resistance. Using different assumptions upper and lower bound factored capacities were calculated.

The calculated capacities assumed that the installation procedures provided a clean shaft and a relatively undisturbed base ie some care and attention was required during construction. The sides of the pile were cleaned with a fabricated wire brush to remove any loose material or build up of clay. The base was cleaned using both clean out buckets and an air lift.

Lateral loads, including ice and ship collision, are transferred by shear and bending primarily to the sand and gravel overburden. The point of fixity was critical to the structural design of the piles but was dependent on the lateral resistance provided by the sands and gravels. An iterative approach was used to solve this soil/structure interaction problem. Initial assumptions were made and moments, shears and deflections calculated at the caisson tops. These moments and shears were then input into LATPILE and revised spring stiffnesses calculated. These revised stiffnesses were then used to recalculate moments and shears. In the worst case three iterations were required for convergence. Sensitivity analyses were also conducted by varying the density of the overburden material and by assuming that scour could be 2 m deeper than calculated in the hydrotechnical studies. The various studies provided a band of maximum moment in each caisson for which reinforcing was provided.

SUPERSTRUCTURE
Steel Plate Girders
The steel plate girders were one of the most critical elements of the accelerated construction schedule for the bridge and were also one of the risk elements. There was a narrow window for supply of the steel plate and fabrication of the girders if erection was to take place using an ice bridge. The preliminary design was reviewed and a mill order placed two weeks after the award of contract.

The steel plate girders were analysed in accordance with the bridge design code CAN/CSA S6-88. A linear elastic analysis was used to compute the moments in the girders. Loads were distributed to each of the girders using a 3-D model of the structure. A similar model was used to determine the natural frequency of the structure.

The steel plate girders are non-composite for the self weight of steel girders, bracing and concrete deck, and composite for all superimposed dead and live loads. The plate web thickness was selected to avoid the need for longitudinal web stiffeners. Girder proportioning resulted in compact positive moment sections and non compact interior support sections. A constant web depth of 3300 mm was selected as optimum. The design of positive moment sections is controlled by serviceability limit state requirements while the design of interior support sections is determined by ultimate limit state considerations. The requirements for serviceability limit state Type 1 do not govern the design. Plan bracing between one pair of girders is provided for erection requirements only.

Bracing is provided at the piers to transfer part of the ice or barge impact load from the substructure to the superstructure. Steel plate girder segments are proportioned to facilitate fabrication, transportation and handling during erection. The segments are between 19 m and 21 m long and were the maximum size without resorting to butt welds. Longer sections would have required a steerable trailer with a consequent increase in transportation costs. The segments were field spliced using high-strength friction grip bolts. The design of field splices was governed by serviceability limit state requirements. All faying surfaces were blast cleaned in the fabrication shop, then wrapped to minimize contamination during the 500 km trip to the site. The level of airborne contaminants was not considered sufficient to cause serviceability issues with untreated steel hence unpainted atmospheric corrosion resistant steel was used for the girders.

Steel conforms to the following standards:

• Plates: CSA G40.21 Grade 350 AT Category 3
• Rolled sections: CSA G40.21 Grade 350 AT Category 3

All bolts are 22 mm A325M.

The steel girders were erected at temperatures ranging from -25 o C to -10 o C. Bearing plinths on the piers were positioned for this temperature range such that the centreline of each bearing stiffener was within 25 mm of the centreline of bearing.

Concrete Deck
The concrete deck varies in thickness from 450 mm over the interior girder to 350 mm over the exterior girders and tapers on the cantilever sections to a thickness of 250 mm.

The design of the concrete deck was governed by shear and bending moments resulting from the empty Haulpak 830E. For design the rear wheel load was increased by 10% to allow for residual material in the tray and/or any tray liners installed. The dynamic impact allowance was assumed to be 40%. While this is higher than the suspension characteristics obtained from the manufacturer would suggest, it was considered prudent in case of possible dynamic interaction between the truck and the deck.

The live load factor is 1.1 as specified. The deck was designed in accordance with CSA S6-88. Key elements, in particular transverse shear and longitudinal shear, were checked using BS 5400.

Moments and shear forces in the deck slab were calculated using a plate bending finite element model. The results were checked using a grillage model. Sensitivity of the modelling was assessed by varying the coarseness of the grillage in the model.

For ultimate limit state design bending moments and shear forces for the slab are the factored moments and shears averaged over a 2 m wide strip. These forces are about 20% less than the peak forces which occur at the centreline of the walls. Distribution of the moments and shears was considered reasonable for the following reasons:

• The slab has sufficient punching shear resistance and longitudinal reinforcement to enable it to distribute the load along a considerable length while preventing local failure. It is also at least 350 mm thick and the contact areas are large.
• The model is based on the assumption that the concrete slab is homogenous and linearly elastic. In practice cracks will open under the load, reducing the stiffness of the section locally and therefore redistributing the load. Sufficient flexural reinforcement is detailed to assure slab ductility.

Longitudinal (i.e., along centreline of deck) reinforcement was governed by the bending moments and also by minimum reinforcement requirements in pier areas. Transverse steel design was governed by the bending moments in the deck except at the middle girder and close to the parapet where shear forces governed. Additional bars in the bottom layer under the parapets were added to provide shear resistance required because of the concentrated shear forces at the parapets.

Serviceability limit state design is governed by the specialty vehicle loading - the 830 E Haulpak - as it is significantly heavier than highway loading and has a load factor of 1.1. In the negative moment zone, at the centre girder haunch, reinforcement design is governed by shear; serviceability tensile stresses in the top reinforcement are less than allowable stresses.

Serviceability tensile stresses govern in the positive moment zone between pairs of girders. The following criteria were used in the calculation of service stresses:

• The peak moment value was used (ie the moment was not distributed longitudinally).
• Both wheel loads of one axle are used to calculate the moment as the wheels are on opposite sides of the middle girder, so each wheel reduces the bending moment in the slab under the other.
• Dynamic impact factor of 1.3. Data supplied by the truck manufacturer indicates a factor of 1.16. A value of 1.4 was used in the analysis for ULS because of possible dynamic interaction between the truck and the bridge. The lower value is considered acceptable for serviceability because of the low number of cycles of the mining truck expected during the life of the bridge.

Punching shear in the deck was checked according to both CSA S6 and BS 5400 and there is a reasonable margin between the demand and the resistance. It is also noted that punching shear is affected by the number of load cycles on the bridge and that after the first few years Haulpaks will cross the bridge infrequently.

Wearing Surface
The waterproofing and asphalt wearing surface specified in the Technical Specifications of the Bid Documents were replaced with a 50 mm thick silica fume overlay over the roadway. Although coarse aggregate in the Fort McMurray area complied with the requirements of CSA A23.1, it also contained a small percentage of ironstone. Ironstone is known to cause local spalling (pop-outs) on exposed flat surfaces, especially if the concrete is exposed to an aggressive environment such as wetting and drying, and freeze/thaw. The ironstone aggregate particles absorb moisture and swell, especially after freezing occurs. This swelling action creates high internal pressures which cause the surface concrete or asphalt to rupture creating the popout holes. While primarily cosmetic, the pop-outs also could reduce the durability of the deck in the long term.

The 50 mm thick silica fume overlay contains imported, better quality aggregate to provide a dense concrete traffic surface giving the desired protection against deicing salts.

THE DESIGN/BUILD PROCESS
Design/build is simply single point responsibility for design and construction. Traditionally a project owner will contract with a Consultant for the design and then separately with a Contractor based on the previously prepared design. There is no contractual relationship between the designer and Contractor - although courts have indicated some standards of care under which there could be a tort liability. With a design/build contract the owner has just one contract for both design and construction with the Contractor.

For the Consultant design/build can offer a fee based on value rather than hours - but also incurs higher risk. Probably one of the biggest risks to be aware of is that of schedule. One of the big advantages of design is the time saving to the owner. On the Suncor bridge Kiewit designed and built the $25M structure in about 15 months after a three month proposal development period (18 months total) - compared with a minimum of six months to design, two months to tender and award, and 15 months to build (23 months total) under traditional methods of procurement. This five month saving to Suncor represents real dollars as the Steepbank mine is constructed. But for the design consultant it is "just-in-time" engineering design. The Contractor needs it now and you better have it ready. The Contractor will not be thinking ahead so you had better - will all the details fit? To respond to the design pressures the lead designer was set up in the site office with all the tools necessary to design and detail a bridge.

The Suncor bridge is a successful design/build project. Why? I believe that there are three reasons for this:

• Appropriate allocation of risk. Many design/build contracts try to pass all risk to the Contractor. In this case Suncor recognized that the geotechnical conditions were variable and requested an add/delete price for piling, thereby sharing the risk and reducing the Contractor’s risk cost.
• An outside project manager. Suncor recognized that their expertise was in oil sand processing and did not extend to bridge design and construction. By bringing in an experienced outside project manager Kiewit had the advantage of dealing with a knowledgeable Owner.
• A positive attitude from all involved. Although no formal "partnering" was instituted, all involved worked positively toward timely completion. Timely reviews of partial submittals were critical.

CONCLUSIONS
The Suncor Bridge project demonstrated that the design/build concept can be successfully applied to larger civil type projects if care and thought is applied in the planning. For Suncor, the advantages of the process were:

• Early knowledge of firm cos
• Early knowledge of firm costs.
• Time savings.
• Cost savings.

For the consultant, there was a realization that traditional organizational structures need to be modified. Material is ordered based on preliminary designs and construction proceeds while the detailed design is completed. A strong concept review and checking process is required to make sure that all appropriate inputs are considered and that the separate components all fit together.